Phasing out Broker Integration(Copy)

I'm sorry Dan, but this needs to be on the front page. I would not have known otherwise. I have some other thoughts about your post, but I will respond to that later. As for now here is a copy.
Hello everyone,

We're shutting down the broker integration feature at the end of September. Everything else on Quantopian will continue unchanged - Quantopian Research, our backtester, our community, and all of our data. We continue to make allocations to selected algorithms, and we're still running the Quantopian Open every month. We've already contacted the few hundred community members who use the feature to let them know. For most of the community, this decision means we can spend more of our time and energy on improving Quantopian. We will be making it easier to learn about quantitative finance, to research strategies, and perhaps get an allocation.

We thought long and hard before we made this decision. A key element of our company philosophy is to keep our goals aligned with our community's goals. This ensures that we're delivering the best experience and product that we can; we're working for you, and for us, all the time. At the end of last year that was still true - we were making allocations and trading our company's capital using the Interactive Brokers integration. Our success depended on the broker integration, and we felt the pain of every limitation and bug in that integration ourselves.

That pain made sure we were investing and improving the integration every day. In April we started trading on behalf of our investment clients using a prime broker, an integration that uses a different codebase. That was the moment when the company goals significantly diverged from the goals of personal traders. Our focus has been on helping the community create high-quality algorithms that we can fund with allocations of millions of dollars, and not on the broker integrations. We have concluded that we can't support personal trading at the level of quality that you deserve and expect.

Moving forward, we want to help people find other ways to trade their own money algorithmically. One possible option is the zipline-live open source project that is derived from our open source backtester, zipline. This project has the potential to be an alternative for personal trading. We've talked to the project leaders, and we've agreed to support the future development of zipline-live. Interactive Brokers has agreed to provide some assistance as well. We support more than a dozen different open source projects already and zipline-live fits our mission. We are also soliciting additional corporate sponsors to help support the project. If you're interested in using this project for your own personal trading, or in helping to build this project please visit http://www.zipline-live.io/.

We launched the Quantopian community in 2012 with just a simple backtester. The feature requests came in fast, and we kept building - first fetcher, and later we added more rigorous alternative data sources. We built broker integrations, history, pipeline, futures, and more. We thank everyone for their help from the bottom of our hearts. The community has pushed us to keep building a high-quality platform. We can't thank you all enough for believing in us and working with us.

I said earlier in this post that our focus has evolved. Still, I want to repeat that our core principles have not and will not change. We're here to democratize finance, to educate people from all over the world, to teach about quantitative finance. We are just sorry that we don't have the resources to do everything we want to. We thank you for all of your support.

Sincerely,

fawce

DISCLAIMER
The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by Quantopian. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. No information contained herein should be regarded as a suggestion to engage in or refrain from any investment-related course of action as none of Quantopian nor any of its affiliates is undertaking to provide investment advice, act as an adviser to any plan or entity subject to the Employee Retirement Income Security Act of 1974, as amended, individual retirement account or individual retirement annuity, or give advice in a fiduciary capacity with respect to the materials presented herein. If you are an individual retirement or other investor, contact your financial advisor or other fiduciary unrelated to Quantopian about whether any given investment idea, strategy, product or service described herein may be appropriate for your circumstances. All investments involve risk, including loss of principal. Quantopian makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances.

165 responses
Gravatar avatar Stephen Hanly 2 days ago
Why not charge us for live trading instead of shutting it down?

Gravatar avatar Tyler Wilson 2 days ago
I would pay

Gravatar avatar Addison Dubay 2 days ago
Of course all of us would be willing to pay for the service, and we would be super excited if you actually accepted our offer. The reality is that your business model has changed (now that you have profitable algorithms to sell to investors and all). I plea with you to please give us a further sunset date. 9/29 is way too soon for many of us to learn a whole new system.

Gravatar avatar tipu sultan 2 days ago
I'm disheartened by this as well. Being able to back test on the same platform you're trading on is pretty crucial.

Gravatar avatar Nicholas Johnson 2 days ago
Will this affect over trading at all? Anyway to integrate notifications if buys and sells a stock

Gravatar avatar Minh 2 days ago
I've suspected this would happen sooner or later. Letting people invest their own money algorithmically for free was too good to be true. It was truly a revolutionary feature. Sad to see it go but thanks for the ride.

Gravatar avatar jacob shrum 2 days ago
would pay

Gravatar avatar Mat 2 days ago
I spent a lot of time learning your platform for live trading my algos and you won't let enough time to find an alternative platform. Why not keeping a paid service ?

Gravatar avatar Chris Venne 2 days ago
So much for leveling the playing field. What a massive disappointment.

Gravatar avatar Igor Ushakov 2 days ago
There are other better tools for running backtests. Other features I'm not so interested in. I came here and invested time into learning your tools just for RH integration. Don't see the need sticking around anymore because this was the only feature that won me over.

Gravatar avatar Costantino 2 days ago
at least do you will continue to provide to data (prices and fundamentals) to zipline-live? It must not be free, you could change a resonable fee and I'm sure a lot of us will subscribe the data feed.

Gravatar avatar Russell Nibbelink 2 days ago
I really feel like you are underestimating the impact that this feature had on your customers. For me this was the bread and butter of your business. This was the disruptor, its a shame that you don't continue. Even though there were some problems, it never seemed too bad. I was VERY happy with the service and feel very let down.

Gravatar avatar Thomas Havens 2 days ago
I also would pay for this. I have developed a lot here and 30 days for change is unreasonable.

This is they most hypocritical and sad thing I've seen. Let me pay you 5 bucks a month so you can maintain the same features. Is this about supporting the little guys or pleasing the big money? There is no way it is prohibitive to allow us users to stay in this community if we are willing to pay.

One note on backtesting vs a live trade.... Your slippage models are not great. Backtesting is almost deceiving, using this system just to backtest is dumb. I run a live algorithm and i would rather have spent my time elsewhere if i knew Quantopian would drop all moral and ethical obligations when they saw a little cash...

Sold out... real sad... Hope this keeps you up at night in your penthouse guys....

Gravatar avatar jacob shrum 2 days ago
Lashing out will not solve this problem.

Gravatar avatar Eric Cheshier 2 days ago
Extremely disappointing news. Hundreds of hours spent building algos on your platform and then you pull the plug with a month of notice?

Gravatar avatar Spencer Current 2 days ago
This hurts.

Gravatar avatar Warren Harding 2 days ago
I think it would be better if Quantopian moved to a 'rent-a-server' model, that's what Quantconnect is doing last I checked. No, I'm not associated with Quantconnect. This is going to really hurt the community in my opinion. What else will code written on Quantopian be good for except the contest and Quantopian allocations? If you have an algo that isn't quite good enough to get an allocation, or isn't what they are looking for, but is good enough to trade you'll be out of luck I guess. Betting it all on getting an allocation is a too much of a gamble in my opinion. I'll probably finish up the contest entries I'm working on and move on for the most part. Porting back to C# is going to be a pain. Oh well, C# is my native language....

Gravatar avatar Addison Dubay 2 days ago
You are right Jacob, lashing out will not solve the problem. If you identify that the only problem is that they are shutting down the live trading service. Now, if you consider that they are hardly giving any notice before shutting it down as a problem, that is worth lashing out over. Many of us make secondary income running our algorithms, and ~40 days is simply not enough time to shift to a different service (if one even exists).

Gravatar avatar Thomas Havens 2 days ago
If you run a long only strategy with RH... you are completely out on your own... Quantopian already has the RH API developed as they were granted access.

Regarding using IB...Pipeline isn't incorporated yet... I don't know anyone that runs an algo that doesn't incorporate pipeline filtering... clocks ticking... Im sorry to "lash out". But this is the time to stand up for values. What is happening here is not about resources. Its nieve to think that.. this thread has over 300 views and most people would be willing to pay 10-30$ a month to keep in this game...This is how the rich stay rich. see life for what it is.

Gravatar avatar jacob shrum 2 days ago
Removing live trading will KILL this community. People aren't going to devote hundreds of hours to be able to execute a great backtest. People get into this out of passion, they stay in because they get paid for following their passion.

Moving to a paid service would provide a passive income stream for Quantopian - And enable us to continue benefiting from the passive income stream Quantopian has allowed us to create for ourselves.

People aren't going to continue to write algorithms with the hopes that you will give them an allocation - Anyone who can get an allocation doesn't need quantopian when they weigh the risk to the reward (spend hours upon hours writing an algorithm that most likely won't be accepted, and, if it is, will yield a very small profit given the current allocation sizes and commissions)

Gravatar avatar James Barker 2 days ago
@Warren QuantConnect has full python support these days. Seems like a solid alternative.

Gravatar avatar Daniel Koh 2 days ago
Oh well... back to trading manually until I find something else to migrate to!

Gravatar avatar tipu sultan 2 days ago
@jacob you are correct. I spent a hundred or so hours developing what I've done because of the promise on actually integrating it into the market. Back testing here and then moving the logic to another platform simply doesn't work.

Gravatar avatar Tsyphur Terrell 2 days ago
What is the point of Quantopian if you can't execute live? I'm new to the service, and it makes no logical sense to learn the system only to have to translate the code to another source. Sounds like this company is collapsing.

Gravatar avatar Sofyan Saputra 2 days ago
Live trading on the same backtester platform is one of the major draws of Quantopian. The majority of sophisticated traders will be less motivated to develop on the platform if there is no reward of being able to actually trade what you just created. We could pay for the service, but I don't think Quantopian is fully focused on enhancing the platform and trading experience. Instead they are more focused on cultivating the hedge fund and recruitment. Paying for a second rate service would not be the ideal scenario. Additionally, Q may find out that recruitment may take a hit without live trading.

Quantopian is still a good backtesting tool and the Pipeline API is extremely powerful, but the live trading was always unstable. This may be a good opportunity for people to start experimenting with the IB API and hosting their own live algorithms or experimenting with other platforms such as Quantconnect, which is more focused on the technology. Feel free to reach out if you are looking to do the same.

Gravatar avatar Tyler Wilson 2 days ago
Will the paper trading with Quantopian still work?

Gravatar avatar Addison Dubay 2 days ago
I recommend you all go leave a review on their Facebook page as well.

Gravatar avatar jacob shrum 2 days ago
I recommend you do something more productive than leaving a review on a page that very few people have ever visited/will visit. Instead of acting on your emotions, act on logic and utility. Spend that time solving whatever problems that this has caused for you. Complaining will not solve your problems - it will waste your time.

Gravatar avatar Viridian Hawk 2 days ago
It is unfortunate to see this go. I mentioned this before, but I think the live-trading element was an incredibly clever loss-leader for Quantopian to attract and grow their userbase. It's what brought me here.

There are some really intelligent and knowledgeable professors of economics and those types participating in the Q Open. I on the other hand don't come from that background. So I gotta ask myself, is it worth it spending months on end researching and exploring potential alpha sources on the Q platform when my only chance of getting anything out of it means I need to beat experts in the field (in addition to a whole lot of luck)? The value proposition for developing on the Q platform is a bit more bleak without the live-trading element. Put another way, with this change I'm more likely to make money developing my code on another platform where I can live-trade it than I have any chance of getting anything out of Q Open or Q Fund. Those efforts are no longer unified.

On the other hand, once upon a time I was the CTO of a venture-funded start-up, so I know all about the importance/pressure to keep a singular focus and not spread yourself too thin. I could see this going either way -- I guess time will tell whether Quantopian just shot themselves in the foot.

I haven't used Quantiacs, but what does Quantopian have now to differentiate themselves from Quantiacs? With this change it would appear Quantopian has less of a moat, less value proposition over Quantiacs.

So, y'know, I will keep on working on ideas for the Q Open and Q Fund -- the research environment here is amazing -- but now my efforts will be divided instead of unified between two platforms, with putting food on the table (so to speak, via live-trading) necessitating more emphasis -- for now.

Thoughts:

It would be amazing if we could get access to that prime brokerage through the Quantopian platform. :) Problem solved, huh?

There is another platform that supports live-trading: Quantconnect, but they appear to be running on computers from the 80s... ran out of memory trying to backtest their demo "Buy-and-hold SPY" algorithm. Also, they don't support Robinhood.

I'm sitting here on an computer with 64gb ram, 6 cpu cores, and a recent graphics card. It's been kind of comical developing on a super slow cloud platform. I'd love to be running my backtests and live trading locally. But....

There's the zipline-live project. They don't have fundamentals or pipeline or Robinhood integration. Briefly glossing over it I didn't even see any mention of data.history(). In addition they warned me it would be moderately difficult to get it up and running. Not encouraging so far...

It's not hard to build a live-trading platform. Problem is the data -- it's so massive and expensive $$$$$. Perhaps there'd be enough community interest to build out a "quant collective" platform that solves these issues?

Gravatar avatar kamran sokhanvari 2 days ago
Fawce, Very disappointed to see this.

I do believe the real trading integration was a major driver for joining this platform and wanting to get the algorithms to actually work with the intent of allocating ones funds to it. I have invested over 18 months of work in this platform and although I was concerned about the future commitment to this feature I felt your mission for wanting to maintain a level playing field for the smaller guys would win out. This move significantly reduces the value of using this platform specially now we know that at any time you guys can decide to take something else out. Who knows ...

The end of life notice is too short and not respectful of the time of the community members who have to actually figure something else out. Please at least reconsider that while we all figure out what to do !!!

Gravatar avatar Addison Dubay 2 days ago
If it buys us 90 days instead of 40 days, it's not time wasted. -and they do have a decent following on there.

Gravatar avatar Spencer Current 2 days ago
This is truly unfortunate though. I guess, the service was free to begin with so who am I to complain right? Regardless, I know a lot of people put hundreds of hours into their algorithms including myself. We're now faced with porting to another site to risk them eventually doing the same thing and ending service, or creating your own platform which would take hundreds of additional hours. You gave some kids some really cool toys that opened their mind to all possibilities and opportunities of the world, only to take it away after they've seen a glimpse of what they can't have.

Gravatar avatar QQQ VVV 2 days ago
Noooooooooooooo. I am just starting...

Gravatar avatar luc prieur 2 days ago
This is disappointing. I can't imagine that maintaining an API link to IB be that difficult or expensive. It is working now and only requires some debugging to move forward (I can only guess). This effort is most probably minuscule compared to the effort of the community to migrate to another platform.

Gravatar avatar Tyler Wilson 2 days ago
It is working now
That is a stretch. Not sure I would necessarily call it working.

Gravatar avatar Thomas Havens 2 days ago
If the issues were actually about maintaining the API and not a conflict of interest with their new rich friends then i would assume that they would plan for the full development/porting of zipline live/piepline and all the promises on that web page to github before announcing a cut off date... Just as a gesture of good faith. Seems like there isn't any good faith in this rushed withdraw from their values and mission statement.

Gravatar avatar Fabien V 2 days ago
I would happily pay a reasonnable monthly fee (10s of $ not 100s)

I wonder if not allowing auto trading is going to drive some people off the site that would otherwise also compete and thus lower the level of the competition or are those different categories of people? I for one was more interested in my own trading but I was on the side developping multistrategy algo that I was hoping to present in the contest at some point...
However the prospect of winning any material money in the contest is not the driving force as it remains tough to win as far as I can tell... now I ifnd myself googling for alternative to Q for my own trading... turns out there are not too many good solution out there!
QC is a pain to develop with, you can sorta use collective2 api but its pricey for little added value, I have not tried quantiacs or couldtrader , has anyone good alternative to suggest for my home-grown algos?

Gravatar avatar Simon Thornington 2 days ago
There is no doubt that this is a devastating loss. There aren't (that I am aware) any other places which provided enough cross-sectional (fundamental) data to do stock screens, integrated seamlessly into both a freeform research environment, algo paper trading, and live trading. Although I was forced to shut off my algos for work last November, I benefited greatly from the workflow, and I will miss it.

That said, this should come as a surprise to very few people. It was clear that despite all their nudging, many people who were live trading their own accounts were not trading the sorts of algos that Quantopian was interested in financing. In their business model, that makes them strictly parasitic. Perhaps they paid their way in terms of contribution to the "community", or perhaps not. I suspect that the forums/community add little in the way of concrete long-short algos which are suitable for their fund. In fact, I suspect that many of the people who are able and willing to write algos which cannot be traded personally yet fit the mold for an allocation do so quietly and with little interaction in public.

I hope people have some sympathy for how difficult a decision this must have been for them. At the end of the day, they have to keep the lights on, and they have employees whose livelihoods depend on the business being a success. Personally I think the "Quantopian execs have turned into plutocrats in penthouses" angle is totally wrong. They are running a business in a tough year for the industry, and just trying to move forward with a focused plan. Live IB trading sounds like it's been a bit of a mess for the last few months, and probably quite a stressful distraction.

Still a terrible shame, though. Especially the data... I was never much a fan of the Python...

Gravatar avatar Roberto M 2 days ago
Very disappointing news... It will make no sense to spend hours in this community anymore if the perspective to run your own algo will not be there anymore! Very sad. Looks like it is dying... That is what big money does - it is hard to conciliate open source, community driven development, and venture capitalists.

Gravatar avatar Kevin n 2 days ago
I am still a bit confused. Can you still do live paper trading on Quantopian's platform?

Gravatar avatar Thomas Havens 2 days ago
I agree with you Simon, However if your mission statement is " Level wall streets playing field" and then you sell out to wall street...

I think that Mike Dent, Dan Dunn, John Fawcett, Jean Bredeche and the rest of the board owe us an explanation.

Thats some deceitful stuff right there. I developed a long only RH algorithm... but I'm in the same boat i understand that hedging and any market is a better play against risk. Eventually i will adapt my method to short on similar principals. However they only way that would ever re-enter into qunatopians acceptable competition algorithms is if i can continue to develop on a very similar platform.

Gravatar avatar Burrito Dan 2 days ago
I am sad, as I trade real money, and haven't got a plan B for someone who can do it, even for a simple strategy. However, running my own company, I realise that commercial realism prevails, and they would have agonised over this one.

@Kevin - Live paper trading is not being axed, from my read, just the broker integration.

I don't need anything other than daily price data and stuff I can get from Quandl, so probably a hosted version of zipline would suffice. Then I would not need to recode. I agree with the comments that $10s but not $100s of dollars per month for the platform is comfortable.

Gravatar avatar Blue Seahawk 2 days ago
Even at $150/mo it is nothing vs fund mega millions, there aren't a huge number doing live trading, although a subscription would reduce the volume level over algos focused on only 1 or a few stocks. 78 messages talking XIV, VXX etc in August alone come to mind. I can't know how many were just here using a free service without understanding the deal or were (or would become) also interested in the fund or contest, maybe even most, I can't know, but that type of thing was viewed as a distraction. Also the login issue complaints that were appearing unresolvable disappear. I wonder how many understand the program or sort of view it as a birthright utility.

Since Jan 2014, I lost some money on IB, have an RH algo that has tripled in 9 months and is running ~75% or so of parallel paper trading. 4M in ~5 years in a backtest so it looked great, had hopes for my future in it. 2678 algos stored locally (many are modifications of each other), major investment of life, time. My RH code was the best return so far even tho the time invested was relatively tiny, for example ...

Today 17 tabs open running variations of code toward potential contest entries or fund candidates (just finding it really tough at 10M), so even though the fund has been my main focus, on this sudden loss of real money trading, can't process it, except maybe with rough thoughts like wondering whether freeloaders ruined some things, I don't want to characterize them that way, I'd like to think all users arrive with an innate sense of fair exchange, reality, yeah maybe not so much or I don't know.

Gravatar avatar Herman Stanly 2 days ago
For some slower strategies (monthly or quarterly), one can still do it manually by live paper trading I assume? So not a total loss considering we'll still have access to a lot of fundamental data.

And I do see why they may have wanted or needed to make this move. But I would have thought a migration to github for a lot of the code that they don't need anymore would have been a nice gesture, and one that wouldn't have taken too much time or support?

Still though... I think they missed a key element, many of us came to this platform to use live trading first, then we started experimenting with developing market neutral strategies. It wasn't the other way around. So by now discontinuing that service they will shut off a key driver of user growth and therefore of viable algorithm development growth.

Gravatar avatar Addison Dubay 2 days ago
Simon, I agree with your statement. I would self-identify as a parasite to their business model to a degree. However, I did try to contribute to the forums and answer questions when I had time. It's unlikely, but possible, that my contributions aided someone that wrote a contest winner, or a helped a future contest winner. Further, I was only a parasite to the point that I still had a lot to learn to develop a strategy, and write an algorithm that would be worthy of, and eligible to enter into the contest, but I had a desire to do so. That being said, they are providing extremely short notice at the expense of those who helped build their business.

Gravatar avatar Thomas Havens 2 days ago
also note that IB doesn't even integrate with Turbotax. like now i need to find a tax accountant to do that too or generate 1000000's of pages of PDF and submit them online through their CSV generator thing....

This is just classic wall street, evil big money fk the little man wall street. I came here cus of the community. Turns out they wanted to fleece a few of our ideas and leave us out to dry after group think was over... Wouldn't be surprised if they start steeling our code content or looking up high returns and copying code at this point... Did you read the terms of service?

Gravatar avatar Jacob Blanchard 2 days ago
well this is bullshit, we spent month working on our algorithm

Gravatar avatar jacob shrum 2 days ago
It is amazing how people can complain about a free service. Grow up.

Gravatar avatar Burrito Dan 2 days ago
'Tis better to have loved and lost than never to have loved at all.

Gravatar avatar Mustafa Tambawalla 2 days ago
Definitely a bummer. Might have to reconsider how badly I actually wanna learn this stuff considering the fact I'm definitely not gonna win a contest any time soon.

Gravatar avatar jacob shrum 2 days ago
From Jared Board (QuantConnect Owner)

As a guesture to welcome all Quantopian users the new baseline free
tier is 8GB Ram allocation! Enjoy!
Gravatar avatar Daniel Koh 2 days ago
@jacob do we need a special invite? I'm joining QuantConnect

Gravatar avatar Luca 2 days ago
I would love to be able to read between the lines of this announcement. There might be different reasons for this decision. Quantopian knows this is going to be a though hit to the community and if they are willing to accept it either Quantopian business is going so well they can focus solely on what they need without caring about the community (I doubt it) or, more realistically, they have limited resources and they have to focus only on what is essential to their business because they are facing dire straits. If so, I wonder if the next step is Quantopian shutting down. I know there have been so many announcements about $250M of Investment and Multi-Million Dollar Allocations so that everything seems fine, but you never know what really happens behind the door.

Gravatar avatar jacob shrum 2 days ago
@luca: we can speculate till we are blue in the face - it does us no good.

Gravatar avatar Thomas Havens 2 days ago
Jacob,

are you quite literally the machine.

Free services value customers all the time...

Gravatar avatar Serge d'Adesky 2 days ago
I agree with the gist of the comments made here:

1) there is no reason Q should support live trading algorithms that don't contribute to Q's bottom line for free. Startup businesses are tough and burn money fast. Q is just as susceptible to this as the next outfit, and it is completely understandable that they must operate in a cash flow positive way.

however why not offer a subscription fee that includes the cost of supporting logins, code changes, etc?
By doing so Q is amplifying its reach into the world of quants, and possibly involving that rare programmer who may contribute to their best
algo solutions. Moreover, Q is thereby remaining much more VISIBLE to quants and financeers worldwide. For example, while laboring to develop algo worthy of winning one of the prizes, I was always amenable to the idea of running real money through a proven Q app and paying the 2%/20% fees inherent in that. Granted I am a small player, managing a measly $7 million in capital, but multiplay that by several 100 other RIA's lurking in the background, and you are talking real money.

Of course I can keep programming away for the sake of the simulated trading environment. Except that I would be hesitant to do that, as we know how different real live trading can be from simulations.

I cannot imagine that that would take more than the efforts of 1 full time technical person at Q. Assuming a salary of 80k a year, and dividing that
by 5000 plus users, it seems the cost of that would be relatively trivial. Add to that of course the costs of the data feed. How about it Q? What is a reasonable cost for such a monthly service?
to permit (e.g extra 30 days seems reasonable) migrations to other platforms / languages that
permit migration to other interfaces/solutions ks to IB and Robinhood. Thousands of programmers have contributed months, and in my case years of their programming time helping Q members achieve this record-setting trading environment. They should not be hung out to dry.

I trust the Q team is not out to harm the thousands of programmers who believed in their model and gave it their all. So I am anxious to hear in the next few days what path to a better transition they are able to offer that provides a "win/win" for all concerned.

Gravatar avatar Luca 2 days ago
@jacob shrum I am speculating because I really like Quantopian. The choices they made in developing their platform make happy both the software engineer and the scientist that are in me. It's hard to find something similar. So I'd really like to know if the next step is closing down Quantopian.

Gravatar avatar jacob shrum 2 days ago
@luca, they will not publicly admit that they are failing - so you won't get an answer. I understand your passion, but only time will reveal the answers you are looking for.

Gravatar avatar Luca 2 days ago
@jacob, you are right, but at the same time I like to hear what other people think. It doesn't change much, I still have to make my own choices while I wait for the answers but it's interesting to hear other point of views.

Gravatar avatar Burrito Dan 2 days ago
@Luca I suspect they are NOT going under, given they raised $25m in November:

https://www.crunchbase.com/organization/quantopian#/entity

https://www.sec.gov/Archives/edgar/data/1589635/000123191916000067/xslFormDX01/primary_doc.xml

From Andreessen Horowitz. They are top tier. Basically Quantopian has won.

Gravatar avatar Burrito Dan 2 days ago
In fact, why not, here's a quote from Ben Horowitz himself, from his book (a must read if you run your own startup):

“Every time I read a management or self-help book, I find myself saying, “That’s fine, but that wasn’t really the hard thing about the situation.” The hard thing isn’t setting a big, hairy, audacious goal. The hard thing is laying people off when you miss the big goal. The hard thing isn’t hiring great people. The hard thing is when those “great people” develop a sense of entitlement and start demanding unreasonable things. The hard thing isn’t setting up an organizational chart. The hard thing is getting people to communicate within the organization that you just designed. The hard thing isn’t dreaming big. The hard thing is waking up in the middle of the night in a cold sweat when the dream turns into a nightmare.” ― Ben Horowitz, The Hard Thing About Hard Things: Building a Business When There Are No Easy Answers

Gravatar avatar Luke Izlar 2 days ago
The main thing I'm worried about with quantopian is thus: How do they expect to draw enough users to keep coming up with good algorithms for their investors when there is almost 0 incentive to do so?

Prior to this, you had a crowd of people live trading and backtesting in order to live trade, sharing their knowledge and strategies which helped the people making algo's for the fund. Now you will only have contest algorithms, but barely any users left to collaborate with and draw from. I've been at this a year and I know there is hardly any chance of getting ranked in the contest or getting picked for an allocation due to their very institutionally driven rules so why would I even bother attempting? Why would anyone other than a few ultra-knowledgable and ultra-dedicated people even attempt it? Surely quantopian as a fund will suffer.

Gravatar avatar Mustafa Tambawalla 2 days ago
So could someone summarize quant connect real quick? Want to know if it's worth switching over.

Gravatar avatar Thomas Havens 2 days ago
Jacob, You really think they are failing to make ends meat? like wouldn't a subscription base allow them a solid market free capital stream? additionally rationally thinking the highest i would pay a month is likely 30-40$ Something like that would reduce the noise and logistical concerns.

Regarding the original posting " We will be making it easier to learn about quantitative finance, to research strategies, and perhaps get an allocation" - J.Fawcett - I spent 6 months developing something... When i started live trading i needed to re-learn everything about the market and do it fast cut i was loosing money. If you remove this feature the quality of your base will change.

@John Fawcett - Charge me homie.

Gravatar avatar Luke Izlar 2 days ago
I will only say this to all the other users with me: lets jump start zipline live project and make it what quantopian was, is, and should be. It's ALMOST READY. we can get it going in a month and not lose any progress!

Gravatar avatar Burrito Dan 2 days ago
@Mustafa - this thread is a good one for your question: https://www.quantopian.com/posts/in-light-of-quantopian-shutting-down-live-trading-what-would-be-the-alternative-option

Gravatar avatar Viridian Hawk 2 days ago
Realizing how difficult it is to replicate what Quantopian has been offering us for free... all that data... newfound appreciation for it. :)

I have a suggestion for people who are panicking. If your algo isn't terribly time-sensitive (which I assume most here aren't), why not just paper-trade your algo and use something like Monkeyscript to scrape the trades as they happen 15-minutes delayed (perhaps you can even compensate for it), and execute those trades via whatever brokerage API endpoints you wish to use?

Gravatar avatar Manfred Bastero 2 days ago
And just like that, once again, the majority is left holding the bag. Quantopian used us to get to where they need to be. Lesson learned!

Gravatar avatar Stephen Hanly 2 days ago
I don't understand why Quantopian wouldn't have looked to sell or partner with another company to use their live trading feature and support it? Namely, wouldn't Robinhood find this aligns well with their business model? Maybe Quantopian is hoping to get us riled up and therefore use that as a bargaining chip to help sell something they've already developed? Maybe QuantConnect would be interested in purchasing this and/or integrating it into their platform?

If you do the math, there are probably somewhere on the order of 10,000 users that would be interested in paying $200 a year? That's $2M a year of revenue that could be paying out right now. And presumably the base of users has been growing and would continue to grow.

Gravatar avatar Doug Wood 2 days ago
1) I would pay an asset based fee (tiered, or flat + tiered) but the 1 million dollar restriction has to go.
2) Only a 5 week heads up is kind of rough...

Seems like providing the platform as a [paid] service is a bigger market opportunity than crowdsourcing alpha models for a hedge fund. Have you considered spinning this out as a separate business?

Gravatar avatar Luca 2 days ago
@Burrito Dan. I like your quote, but what is your view then? Isn't this decision affecting its user base and so hurting its business? If Quantopian doesn't have financial problem, why taking the risk of ruining their business with this drastic decision? I am asking for the sake of entertainment ;)

Gravatar avatar Viridian Hawk 2 days ago
@Thomas Havens - The Robinhood API is not officially public, but it's also no secret. It's the same endpoints that the app uses. Just google it -- it's really easy to use.

Gravatar avatar Burrito Dan 2 days ago
@Luca OK then:

“IF YOU ARE GOING TO EAT SHIT, DON’T NIBBLE” ― Ben Horowitz, The Hard Thing About Hard Things: Building a Business When There Are No Easy Answers

Gravatar avatar Delman Woodrum 2 days ago
Might be time for Quantopian to update their community tagline.

Our community — 100,000 members and growing — ranges from seasoned algorithmic traders to aspiring quants. We help each other with code problems and discuss ideas in algorithmic trading.
I expect a huge majority of the community is going to move on to other platforms that allow live trading, seeing as how, at least for me, the live trading ease-of-use was the solitary reason for using this platform.

Gravatar avatar Alexander Katyk 2 days ago
Support the move - I think institutional clients would be more comfortable too that the code they essentially allocate to can't be easily used to fund another account. The more clients the better experience for users like me :)

Gravatar avatar Dan Dunn 2 days ago
Thank you all for replying. Your passion shines through, and I appreciate it.

Several of the replies we got today asked about whether it would help if we started charging for the service. Unfortunately, that wouldn’t solve the problem. One of the big reasons we’re making this choice is so that we can focus on our business: helping thousands of people each month learn about quantitative trading, teaching them about statistics and signals and coding, and rewarding the best of their ideas with prizes and allocations. We can do that more, and better, if we improve our focus.

Some people ventured estimates of the size of the personal trading business. We only have a few hundred people trading real money. If we started charging $20/month (suggested earlier in the thread), and every single person agreed to pay, it would only be a fraction of the total cost. I just shared a longer version of this answer here.

Several replies were about alternatives, and there were several reasonable ideas kicked around. I suggest looking at http://www.zipline-live.io/. Here at Quantopian we are making continuous investments in zipline, and zipline-live can naturally piggy-back on those improvements. We have seen the power of open source at Quantopian. I am hopeful that the time and energy that people have poured into their algorithms can spill over into zipline-live. With your excitement and skill, zipline-live might be the replacement that so many people here are looking for. The maintainers of zipline-live can better answer questions about how to integrate data sources, add pipeline, and important details like those.

A few people asked about what Quantopian does other than brokerage integrations. Quantopian is the biggest community of quants around, and most people come here to learn. We provide tutorials, lectures, tools, and tons of data. For some people who use Quantopian, it’s about working on an idea that might someday get an allocation. I write this paragraph to extend perspective, not as a solution.

I want to close by repeating fawce’s thanks from before. You’ve all pushed us to make Quantopian a great platform, and you’re still pushing us. Thank you, all, very much. Ending this feature is very hard for us. Many of you had kind words for us in this thread, and understood how hard this is for us, and I thank you for that empathy. For many, we know we’re disappointing you, and we are very sorry.

Gravatar avatar Burrito Dan 2 days ago
@delman they have plenty of data on their users behaviour, and I'm sure those of us who live trade are in a minority they can afford to lose

Gravatar avatar Thomas Havens 2 days ago
@Dan Dunn

Thanks a lot for the reply and confirming low numbers of successful traders. i assume that also means that algorithm allocation isn't actually impossible...

1) When will pipeline and the equivalent of the bare bones Quantopian be online?
2) Robinhood API needs to be part of this build as the code is out there and robhinhood is much better for learning than IB. ( Tax integration, free, app ect.... hedging is an important tool but over mosts heads lets be honest)
3) if you are truly dedicated to the learning of this filed you must understand that working for self reward is the primary motivation... This does directly contradict your statement about wanting to be a learning platform. Khan academy statistics is more than anyone really needs to know . You don't need to re-teach that stuff. your value is in your platform.

Gravatar avatar Kory Hoang 2 days ago
@All,

I am saddened by this news. I am going to assume that this decision is final and there's no going back.

With that said, if anyone has interest in migrating your Python codes to TradeStation or MultiCharts, please email me at: [email protected]

I highly suggest checking out EasyLanguage for TradeStation and MultiCharts. You can do a lot of the things you do here on Quantopian with EasyLanguage and it is very easy to learn (obviously).

I also suggest checking out QuantConnect and Quantiacs.

Gravatar avatar Eric Novinson 2 days ago
The tutorials are still useful to me anyway, and at least zipline is open source. I wasn't actually live trading but was considering it if I found something reliable. Wasn't able to download the quandl data into zipline but maybe someone will put up a tutorial. Would like the option to trade potentially profitable algorithms myself, otherwise it feels like the focus is performing spec work for institutional investors.

Gravatar avatar Andrew Bannerman 2 days ago
I just started to use Quantopian in the past few days and aiming to write my first algo for live deployment. I use R and external data sources and I was just thinking 'today' of the value of learning to write 'algos' here. With Robinhood, it meant there was an edge for the little guy and strategies not scalable to institutional level or strategies where commissions would erode the edge. All those were possible here, the community seemed good too. However, losing the live trading is a missing chain in the link imho. I am quite surprised that the number of people actually trading live algos is so little, very surprised actually. If it dosnt make economical sense, then heck I can see why you have to do it. It is a true danger now that you may lose many users as they may write scripts in R, python or any other external software / scripting / programming. The major win here or the major win 'was' that you could write a script and back test and utilize the same rules live. Excellent..... Very disappointed here, however, maybe something open source like this is an option? who knows but looks like crawling back into my cave. I hear zorro trader has direct broker connections, have to learn their language but might be worth a look.

Gravatar avatar Mustafa Tambawalla 2 days ago
Last thing I'll say here before I go on with my business, but I believe that your quote "Quantopian inspires talented people everywhere to write investment algorithms" no longer stands true as there is very little incentive for low capital and less experienced traders to actually use this website. Our algorithms serve for very little personal gain after the removal of live trading and the only gain we could possibly seek to make requires your validation, where facts show the chances of actually winning a contest are abysmal, even with an impressive algorithm. Not everyone is capable of developing groundbreaking algorithms, which you guys are indirectly forcing, but everyone is capable of learning something new.

Gravatar avatar Grant Kiehne 2 days ago
My basic assessment here is that Quantopian gave brokerage integrations a try and it didn't work out. After 3.5 years of working it pretty hard, publicizing a lot, and even after offering totally free trading, out of 100,000 registered users, only a few hundred are impacted by this change. If retail algorithmic trading were a way to print money, then I would have expected much greater adoption. It would seem that the market just isn't there. It was a pretty expensive way to get the answer, but I guess that's for the Quantopian VC's to decide if it was worth it.

It is a minor disappointment for me, since I do not trade with Quantopian (nor anywhere else, at this point). I was hoping to make a little money, either with the contest or by getting an allocation, and then put it toward an algorithm.

It would be interesting to know how Quantopian real-money users fared over the last 3.5 years. It may be that Quantopian, on average, is doing users a favor with this change. Are there data that could be shared?

Gravatar avatar Timothy Bhattacharyya 2 days ago
@fawce This is a terrible decision. The 99% of users who don't get an allocation will leave the platform after a few months. Who's going to start working on a great futures algo unless there is the option to trade real money on your own (if you don't get an allocation?).

There's no going back here. Once you shut down the option of live trading, users will migrate to other platforms and never return. I second Viridian Hawk's suggestion. Makes us move the prime broker and charge a reasonably small monthly fee.

Gravatar avatar Peter Bakker 2 days ago
My 2cts: People are motivated by different things, but greed is a very important one. Applause another one. To be honest: I'm motivated by greed, and in my mind, the chance to make money with developing an algo for the contest or fund is rather small, and the potential earnings are meagre, so my best bet was to trade my own algos. I trade depending on the day between 80 and 110k, and my performance is above market with lower risk (known risk aka exposure).

Quantopian would like you to develop high capacity (millions/billions throughput) market neutral, dollar neutral algorithms using datasets other then price/volume, and it happens that if you trade those algo's with sub 500K, you probably get eaten by the transaction costs and slippage we have on IB and RH. Hence those algo's are not interesting for me now as I dont trade that much money freely

Hence for me it's easy: I'm here for my own gain, and my own gain is developing algo's and deploying real live algo's that fit my risk profile. So now this unfortunate step makes me think: why would I stay, why would I contribute? Almost no reason to do so except for some research. I might still share algo's but probably more in the form of notebooks as proof of concept...

What now? @Lecoque shared his libraries, there is IBridgePy.com and there is Zipline-Live that probably have the least amount of work for algo's that have a known set of assets as those algo's don't need pipeline. FYI: IB has a limit of 100 assets where you can get realtime quotes for (or pay for more) so algo's that have less then 100 assets, they are fine.
Or you can go for the other platforms, but they seem to be less advanced and I don't feel like rewriting 100's of Algo's

But this is also a blessing in disguise. The broker integration did not have futures, did not have options and I would love to trade those as well.

Bottomline for me: Fawce, Dan, Delaney: thanks for the ride, I guess this was hard to decide but Quantopian is not democratising the finance industry anymore.... unfortunately.

The thing that is hardest to swallow is the time we get to find another solution: it is way too short and it shows that you guys actually do not respect nor value the community as you claim to do. PLEASE GIVE US TO END OCTOBER!

I feel a bit betrayed and sad, as I believed in your mission... I guess everybody needs a sucker punch now and then

I'll share a few more algo's for the benefit of the community but for the rest: I'm out, no more hours on the forums, no more helping people with Q worthy algo's
I rather put my energy in zipline-live or other initiatives

Gravatar avatar Viridian Hawk 2 days ago
It would be interesting to know how Quantopian real-money users faired
over the last 3.5 years. It may be that Quantopian, on average, is
doing users a favor with this change. Are there data that could be
shared?
@Grant -- coming from a start-up background myself I can tell you that if we had success stories on our platform we did our best to publicize them. Since they haven't I think your suspicion may be correct. :)

I think the reality is -- and you see this in the contest as well -- this stuff is f---ing hard. It's surprisingly hard to have a hedged, consistent, low volatility algo that beats the S&P500. I mean we're at a disadvantage compared to Wall Street, but look even at hedge funds -- they're struggling. Often these funds are losing money right now in a more or less straightforward bull market. As soon as you remove "market risk," "sector risk," etc. you just replace it with alpha-factor risk. That one is really hard to overcome -- an accidental bias or market dynamics shift and all that hedging doesn't save you.

So I think people must give up when they can't find anything that backtests consistently well. Or if they do find something they give up as soon as it fails miserably out-of-sample. Proper statistical rigour is really really hard. Or on the flip side, somebody with healthy levels of skepticism may lack confidence to live-trade their algo. And once you factor in slippage, commissions, margin fees, and taxes, well... buy-and-hold VOO or VT might just be the better option.

I'm surprised though. I have seen some really promising looking algorithms in these forums (and I assume there are a lot of people who wouldn't dream of posting their most promising algos -- I know I haven't). But you see some of these great algos and they've been cloned hundreds of times. I would have thought there'd be an army of lurkers somewhere live-trading these. I'm surprised to hear there are only a couple hundred. Maybe 200 is the total intersection of world population, python programmers, interested in the stock market, and clever. :)

Namely, wouldn't Robinhood find this aligns well with their business
model?
@Stephen Hanly -- to the contrary, my experience was that Robinhood was adamantly against in helping me get a second account running for use with Quantopian. I mean here I am trying to deposit a hundred thou into their platform, and they're just like no. Don't know what's up with Robinhood, but I did not get a good vibe from them.

Gravatar avatar Rolando Retana 2 days ago
I would definetly pay for the live trading if that was an option. I have invested alo, ALOT of days work on this tool, it would be totally dissapointing to see the tool go :(

Anyway, I will probably just move to quandconnect and start recoding everything there.

Gravatar avatar Grant Kiehne 2 days ago
Maybe 200 is the total intersection of world population, python programmers, interested in the stock market, and clever. :)
Interesting. But those are just the attributes required for successful Q hedge fund algo writers. I'd think that in order to attract prospective fund authors, it would have been a plus to offer the trading platform, but I guess it didn't work out that way. There had been so much interest in futures, then they come, and poof! Odd. I'd think it would be in the elemental nature of serious traders to want to use the platform with their own capital, and also try their hands at getting a Q fund allocation. It seemed like a winning combination. Overall, given only 200 real-money traders, it suggests that the present pool of serious Q fund algo authors is rather small, but it is not clear that dropping broker integrations will help the numbers grow.

Gravatar avatar Aqua Rooster 2 days ago
I might get some backlash for these comments I think this is a good move.

Quantopian has positioned itself as a crowd sourced hedge fund and moved away from the broker model. Why would a hedge fund provide brokerage services?

Secondly it is not difficult to get an allocation and the allocations are not 500k but millions of dollars which means a decent share of the profits. I hope that now that they are more focused we will see better opportunities for some of us who are keen on making a career in quant finance by introducing new datasets, improving speed of backtests and more tutorials on clustering etc.

Gravatar avatar L. Williams 2 days ago
Why do I feel like this isn't adding up? The entire notion of algorithmic trading is to be able to utilize the algorithm. What drove people to learn the platform in the first place was the ability to identify strategies where they could invest their money, and have a computer, which is far more disciplined, execute a specific strategy. This is beyond stupid, and no rational business person in their right mind would make such a decision, unless there was something else behind the scenes that we're not being told. I see this as akin to the pattern day trading rule. They say it's designed to help the average consumer not get taken advantage of, but the reality is it's designed to limit average consumers ability to profit from the market, because any idiot with half a brain can beat the average annual 6% return. You guys totally sold us out! I wouldn't be surprised if the other sites were next. "Level Playing Field", yeah right. You know the day you set out to do that you would be a target, and instead of standing up for the little guy, you totally sold us out. So how about you change the "Level playing field" part to "Playing field", cause that's exactly what this is now. One big joke. The question is, did they pay you to sell out, or threaten you.

Gravatar avatar Miles Adkins Yesterday
More liberals complaining about free stuff getting taken away from them.

Gravatar avatar Takis Mercouris Yesterday
I haven't been able to develop an algorithm that would receive an allocation, but thanks to the community ideas discussed, I have been able to develop an algorithm that I have been profitably trading since last December, while I continue to investigate new ideas. What attracted me to Quantopian in the first place was the complete integration between backtesting and live trading via IB -- Had the IB not been present, I would have looked for other options or stayed with R...
Joining Quantopian just for the hope of getting an allocation would have only attracted good quants, but then the fora would be thiner with fewer ideas and contributions -- this as others before me said may end up reducing the uptake in the quants Quantopian says it wants to attract.

I would also happily pay $20-30/month for the ability to trade a couple algorithms integrated with IB, as is today.

Gravatar avatar Peter Bakker Yesterday
@Aqua Rooster: "Secondly it is not difficult to get an allocation " ... I beg to differ, the threshold is pretty high and although I wrote 100's of original algo's, none were selected, but maybe you can help in the true spirit of communty: can you share an algo that has the basis of something that could attract an allocation? There have been only a few allocations and there are 100's of live traders and 1000's of tinkerers... the % of success is pretty small

@Miles Adkins: If you have nothing to add to the discussion please go [email protected] somewhere else

Gravatar avatar Aqua Rooster Yesterday
@Peter, I get the feeling that they are looking for ultra low risk algorithms. Something that won't lose more than 2-3% over past 10 years.

Gravatar avatar Tony Li Yesterday
Please keep live trading feature and consider to monthly charge .

Gravatar avatar Marc DeBoer Yesterday
Damn it's been real

Gravatar avatar Tim Decker Yesterday
Very disappointing decision.

Quantitative processes, analytic thinking, pipeline programming, etc are incomplete without an interface to the markets. From open outcry to HFT all strategies/algorithms must be able to meet the final dynamic, implementation.

"No operation extends with any certainty beyond the first encounter with the main body of the enemy" - Helmuth Von Moltke

Definitely a challenging decision for all those who have invested time learning the specifics of the Quantopian system with the full intention of transforming learning into live trading.

Gravatar avatar Andrew Kunkel Yesterday
For those of us that algo trade with Robinhood, can someone provide an alternative algo trading platform we can use and provide step-by-step instructions on how to get it to work with our existing algos?

Gravatar avatar Rahul Patharkar Yesterday
A monthly fee or even a percentage of our profits would acceptable if live trading with Robinhood was possible. I am in the process of learning right now but definitely want to put my money into any algorithm I develop that does well. The criteria for getting an allocation for your algorithm is very restrictive (although I see why Quantopian would want it that way). For the moment I would be happy with any algorithm with decent alpha. A longer term goal would be to develop an algorithm that could get an allocation. I don' think there is much in it for me anymore if I can't try to profit with my own money until I figure out how to make an algorithm that gets funded. I am realistic in that it will take me a while to become an algorithmic investing genius. I can't risk my time and effort to something that has such an uncertain return. I make decent profits in manual investing. I believe in the long run I could make an algorithm that fits Quantopian's requirements, however, why risk trusting them? This is a major change in their promise. How do you even know they don't use some people's algorithms without paying them? They have to do what is good for their business but we have to do the same. There is simply too much risk for any beginner to commit time to Quantopian now.

Gravatar avatar Tony Chang Yesterday
So if i write an algorithm under Quantopian that is not qualified for allocation and i want to use my own real money to trade on such algorithm, what choice do I have going forward ?

Gravatar avatar Thomas Chang Yesterday
@Kory Hoang
Seems you have experience on many other platforms and software. Could you please open a new post and talk about it? In this way all of interesting user here can put their question there.

Gravatar avatar Mattias Lamotte Yesterday
While I am also very disappointed in Quantopian's decision, I think we all have to accept that Quantopian is a financial enterprise. Having 300-400 accounts at USD 30 per month just doesn't make sense for any business, especially if there's no real growth relative to the effort and resources put into it (just take a look at how much a good programmer is paid in Boston).

I agree that 5.5 weeks isn't enough time for most of us to properly test migrated code (I would have thought that a minimum of 2 weeks of paper trading and 2 weeks of trading with small money would be required before safely redeploying full capital). I have played around with Zipline-live and I have installed it with little difficulty on Mac OS X. Looking at Github, main limitations currently seem to be:

limited to having no pipeline (a workaround is in the works but it may not be ready by October 1st)
limited to 100 tickers (I believe you can pay IB to increase your data priviliges)
multi-account support doesn't seem to be there yet (this is a problem for those with F accounts)
Anyhow, I think most us need to move past the bitterness displayed in some of these posts, and find the best way to move forward in a productive manner. Zipline-live is the clear way forward. We should be in able to get something that can run our existing code with minimal changes. This would allow us to continue using the Quantopian Backtester to test new strategies, paper trade them for a few weeks on Quantopian, and then redeploy them with minimal coding changes on local instances of zipline-live.

Of course, running a cloud-based solution was super convenient, and problems were quickly dealt with by Quantopian's team. We will now have to double check that our connectivity and algos are running properly every time the market opens. Frankly, this is not a big deal.

I run 2 real money IB accounts traded though Quantopian. My plan is to buy 2 mini PCs for USD 100 each (https://www.amazon.com/Azulle-Quantum-Access-Windows-storage/dp/B00X4O6GRK/ref=sr_1_1?ie=UTF8&qid=1503469611&sr=8-1&keywords=azulle) and install python2/virtualenv/zipline-live on them. These mini PCs take no space, consume virtually no energy and they seem relatively stable (they are slow and have very slow write speeds, but that's a trade-off I can live with). If such a solution works, it will be relatively inexpensive and hassle-free to run your own algos with zipline-live...

Gravatar avatar Tony Chang Yesterday
@Mattias Lamotte , sorry I may be asking a silly question. Does zipline-live provide real-time market data (the same as Quantopian) ? The reason why i am asking is : I try to open a new account at interactive broker and i know that we need to pay for market data fee in case i need real time market data. Without paying such fee my stock quote will be in "delay" mode. So i am just wondering how zipline-live can provide real-time market data ? Also, what is the coverage at zipline-live ? Only US stock ? How about CMEGROUP, currency, gold/oil ..etc ? Thanks.

Gravatar avatar Mattias Lamotte Yesterday
@Tony Chang Zipline is the library on which the Quantopian backtester is based.

Like Quantopian, Zipline-live doesn't provide any real-time data. Historical daily data can be pulled from a variety of free sources (quandl, yahoo finance, google finance etc), Real-time data needs to be bought from the broker in any case (no free lunch).

While I am not aware of the existing capabilities (I believe the first iteration will be limited to US stocks), I believe that zipline-live could easily be configured to work with any financial product. I suggest you go explore the Github pages of zipline-live to get a better sense of capabilities.

Gravatar avatar Luca Yesterday
@Mattias Lamotte, I agree with you that Zipline-live is a way forward, at least that would be my preferred way to go. Unfortunately it doesn't have all the features available on Quantopian (yet?) nor we have a way to make use of Quantopian data with it. Those are the main issues in my opinion. Having to host our own Zipline-live server is just annoying but not a major problem.

Gravatar avatar Tony Chang Yesterday
For example, Does zipline-live support morningstar stuff ? Or the US500 stocks ?

Gravatar avatar Mattias Lamotte Yesterday
@Tony Chang and @Luca, it looks like all the morningstar stuff (fundamental data and all Q500US and Q1500US universes) are not going to work on zipline-live. Looking at the Github pages, they are working on a pipeline workaround where you will be able to download all the tickers for a given exchange, and then set filter parameters (e.g.: moving average of dollar volume) to filter out illiquid names from the pipeline output dataframe. I'm not sure this will be ready by October 1st, and in any event it will represent a change in code logic that will require a period of testing/debugging to adapt to zipline-live.

Interactive Brokers do have APIs to call some fundamental data (https://interactivebrokers.github.io/tws-api/reuters_fundamentals.html#gsc.tab=0) but I don't know if historical fundamental data is available.

Gravatar avatar Greg Bylenok Yesterday
I am relatively new to the Quantopian platform but am not sure I will continue at this point. I don't fall into those couple hundred people actually risking their own capital, but I certainly identify with them. I always saw live-trading as a stepping-stone on the path to achieving an allocation. What if I develop an algorithm that doesn't scale well or doesn't meet Quantopian's narrow risk parameters? To Quantopian, it's useless but to me, it's potentially supremely valuable. Oh well. I guess I should go build skills elsewhere and then return here when I feel I can compete on the long-short, market-neutral playing field.

Gravatar avatar DS 501st Yesterday
Very frustrated. Just started a month ago. Hopefully I'll be able to make the transition to zipline-live, but without RH integration it's fairly useless to me as-is, so I'm hoping a LOT of angry people jump into development and speed it forward as quickly as possible. I would also have supported paying Quantopian, but I guess it isn't worth their time.

It would be really, really great would be if Quantopian could detail further what they meant by "We've talked to the project leaders, and we've agreed to support the future development of zipline-live." I'd honestly like to see a commitment from Quantopian to help zipline-live get as many of Quantopian 's features as possible operational ready by the time Quantopian shuts down brokerage integrations. If you can do that, it'd be a relatively seamless transition and zipline-live developers could add additional features after that on their own in the future based on the support that Quantopian provided in getting them up to speed.

And frankly, a ~30-day shutdown window is a bit short. Not that I expect anyone (in any industry, ever, anywhere) to admit mistakes or change decisions based on user input, but Quantopian should seriously consider extending that to give zipline-live more time to rapidly develop, especially if Quantopian weren't actually serious about supporting their development.

Gravatar avatar Grant Kiehne Yesterday
@ Fawce, Dan Dunn -

Presumably, you are still using a feed from NxCore, and will continue to use it for the Q fund? Do you see any path to supporting the crowd with the tools to create a zipline-live compatible OHLCV minute bar feed? Would you be willing to open-source your injestor architecture and code? Or maybe even supply the feed (with a volume discount passed on to the crowd)?

I'm just wondering if there is any path at all for your abandoned crowd to obtain their own 1:1 compatible data, so that Quantopian research/backtesting/paper trading could be used, and then the algo ported over to zipline-live?

By the way, although I'm sure there were compelling reasons for the short notice, 30 days notice is pretty short for such a major move. Without additional background, I'm compelled to preach. In the business world, you'll have trouble if you continue to apply this approach to stakeholders (e.g. Q fund "managers," VCs, Point 72, prospective customers, vendors, etc.). You'll get a bad reputation and folks won't want to do business with you. Let's say you allocate $50 million to a user algo, things are going great, and then the author decides to pull the plug, with only 30 days notice? You'd likely not want to work with that guy any more. It does feel like certain elements of the Golden Rule were overlooked here.

On a separate note, it is perplexing that the retail trading industry is huge, yet there is no market for enabling retail traders to do a bang-up job of it. I think this was the vision expressed in Fawce's Quantopian Manifesto. Build it and they will come. Make the market. Maybe the market is primarily gambling? The idea of thoroughly researching a strategy, coding, debugging, and deploying, doing out-of-sample paper trading, etc. just doesn't work at the retail level? I'd think that IB and RH would be super disappointed, but maybe they already understand that the upside potential is just not there.

Gravatar avatar luc prieur Yesterday
I suggest a techincal solution such that users can continue trading with Q. If Q could implement a new trading option, in which case the Q servers would establish a connection with a software client residing on the users PC. That client would in turn be connected to the users IB TWS. The IB TWS has a well documented API. The software client would act as a bridge between the Q servers and IB TWS. The Q servers would send the purchase requests as calculated by the algo running on the Q server and receive back portfolio information.

I hope this can help.

Thanks

Luc

Gravatar avatar John O'Leary Yesterday
For any IB users who are disappointed by this news, please check out the platform IBridgePy. It is a platform that (judging by the name) bridges IB with a python program that you build, compile, and run on your own machine. This eliminates a lot of Quantopian's fundamental problems and gives you much tighter control of what goes on in your algorithms.

Also, it has a "run like Quantopian" mode that uses Quantopian's clock rules and whatnot. The main advantage of this is being able to write and backtest code in Quantopian's web client, then copy/paste the code into IBridgePy.

It is still fairly new and in active development, so it probably is not without its bugs. But honestly, it's the next best thing in my opinion. Only big drawback is that you need a dedicated, on 24/7, machine to run it if you want to live trade.

Gravatar avatar luc prieur Yesterday
John, the client software I am suggesting actually would use the IBridgePy python library for its implementation. It would need a feed from the Q servers to get the orders (i.e. the result of the algo). IBridgePY is the python library that implement the IB TWS API.

Gravatar avatar Tony Chang Yesterday
So can somebody provide a comparison in between IBridgePy and zipline-live ? Is IBridgePy provided by IB ? Which tool is "more long-term" ? I do not want to spend another time cycle for learning yet another platform (assuming that i need to say goodbye to quantopian).

Gravatar avatar luc prieur Yesterday
Maybe I am mistaken for the name of python API library to to IB TWS, maybe it is not Ibridgepy.

Gravatar avatar John O'Leary Yesterday
Luc,

I don't really see the point of maintaining a connection with Q anymore. I always assumed the recommended method for using IB as a live trading platform was to maintain two connected IB accounts, one as the "live trader" (which handles the money) and one as the "observer" (which merely exists so that you can logon and use your IB account to its full advantage). If you don't have an observer, you can't logon to IB for any reason without disconnecting your live trader.

Quantopian is (was) able to act as the observer, but it was very limited in terms of what it could look at, in my opinion. IBridgePy is a very, very good solution as long as you are willing to put in the extra work.

Gravatar avatar John O'Leary Yesterday
Tony,

I don't know much about zipline-live, but in the small snapshot I've got of it from this thread, it seems much more limited than IBridgePy.

IBridgePy will let you look at any ticker (stocks, futures, options, forex, whatever) that your IB account has permission to see, and lets you trade using whatever method you want, provided IB TWS has the capability to do so (which it should, because it is a very powerful platform).

There are almost no limitations to what you can do when it comes to IBridgePy. None that I can think of off the top of my head, anyway.

Gravatar avatar Peter Bakker Yesterday
I played with IBridgePy and it seems to work. I could query my accounts and could buy a share. They implemented all main functions from Quantopian but as extra they can have a faster cycle. 1 sec instead of 1minute. Next step for me is to port an algo and take over the trading. I’ll keep you posted.

Gravatar avatar Tony Chang Yesterday
Hi John O'Leary , thank for your speedy response. I just do a quick search on IBridgePy and immediately realize at least on thing it wins Quantopian - that the Quantopian schedule_function (which is limited to US stock trading hour) is no longer an issue at IBridgePy ^_^. This is important in case we trade thing..like CMEGROUP future such as gold / oil. Now i have 2 more questions. As you say, IBridgePy allows us to trade ANY product that IB provides right...so can i trade Hong Kong stocks (or even HK index future / option) using IBridgePy ? Also, do you know whether i can get delayed market data free of charge (or i must subscribe first even for accessing delayed data) ?
Thanks.

Gravatar avatar John O'Leary Yesterday
Also, I've had very good success with using IBridgePy to grab historical price data (any timeframe, any resolution) that can then be used for backtesting on your own backtesting platform. Again, this applies to any stock, future, option, or forex. It's important to note that this data comes directly from IB, not from some database with potentially incorrect data. This was one of the biggest limitations of Q, in my opinion. The ability to export data to CSV is huge.

Gravatar avatar John O'Leary Yesterday
Tony,

You need to check your IB permissions to see what you have access to trade. If you have the permission to trade and view real-time data for the HK stock/option/future exchange, there shouldn't be anything stopping you from using IBridgePy to live trade on them.

And to your second question, I don't believe so. Trying to access data (even if it's delayed) that you don't have permission to trade has resulted in some kind of "Lacking permission to view this data" error. There may be a workaround I'm unaware of, however.

Gravatar avatar Stephen Hanly Yesterday
Has Quantopian considered selling the backend software that supported the live trading piece? I wonder if there is a business opportunity to build a Quantopian as a marketplace? I had asked Dann a while ago about buying the backend software that integrates with Quantopian and individual brokerage accounts. I’m still interested in this. I would think that Quantopian’s investors would be interested in getting something for this software/interface that they’ve already developed?

https://www.quantopian.com/posts/quantopian-as-a-marketplace

Gravatar avatar ryan clous Yesterday Edit Delete
I'm studying right now, and I have spent a year developing on your platform. I believe it is great and provides exactly what we need for live trading. I believe the quantopian team is making the wrong decision here in shutting down live trading instead of monetizing their service. I believe many people would pay for broker integration on this platform. I will post a petition and speak with someone in quantopian to see if it is a monetary issue, so if possible we can start a kickstarter/gofundme of sorts and see if they will implement a subscription based platform for us to use. Another possible route is to add a donate option on their website.

There are many avenues they could of taken to continue the service, and it makes me wonder why they haven't utilized those options. If they shut it down for good, it makes me wonder what their intentions are. Meanwhile, I will look into other alternatives.

Gravatar avatar Todd Wood Yesterday
@ Stephen, I replied to your market place idea or some variation thereof. https://www.quantopian.com/posts/quantopian-as-a-marketplace

It's doable if the "trading middleware" between Q and the broker(s) moves to an institutional model of charging 1/2 penny or a penny per share. The number of live traders becomes irrelevant, but the volume matters. This is what anyone pays to clear through institutional firms like Goldman, Cantor Fitzgerald etc. It's a sliding scale based on the client volume and is rolled into the trade cost.

One would have to understand the volume behind ~200 traders currently using the platform, but it's definitely possible.

It would take considerable effort to get Zipline-Live to that stage, but perhaps Q would partner in such a venture with the contribution of existing code base for live trading in return for a split of revenue. It doesn't make sense to completely drop something if there is a chance to receive some benefit by licensing it out and not have to expend additional resources. I think there are some options to explore.

Gravatar avatar John O'Leary Yesterday
How common is it for hedge funds to be completely algorithm-dependent? Is it fair to describe Quantopian as a crowd-sourced hedge fund?

Surely there must be other similar business models that could succeed without the same approval process that Quantopian requires. I would be very, very interested in keeping track of some of these younger "crowd-sourced" hedge funds and seeing what comes of them.

Gravatar avatar Grant Kiehne Yesterday
Is it fair to describe Quantopian as a crowd-sourced hedge fund?
In my opinion, no. Yes, the licensed algos come from a global crowd, but the business is not crowd-based as a cooperative (e.g. https://en.wikipedia.org/wiki/Cooperative). The crowd has no power, no right to information, no nothing. By design, this is how Quantopian is set up and I suspect all other so-called "crowd-sourced" funds are the same. Until somebody actually sets one up that has a fundamentally different business model, I wouldn't expect a different outcome. I don't fault the Quantopian team and their backers. They are clever, and know that they could do things differently. I have to wonder, though, if they fully appreciate the engagement they could have under a true crowd-sourced (i.e. crowd-owned and crowd-controlled) collective effort. Leveling Wall Street needs to start with the fundamentals of the business that intends to do the leveling. One draw to Quantopian for me was the hope of not working for The Man (I do that already), but that isn't working out so well. Everything seems strongly vectored in the opposite direction.

Gravatar avatar Garrett Brigman Yesterday
Would Pay. Big disappointment.

Quantopian Goal = Investor Client Goal

Broker Integrations = Community Goal

Investor Client Goal > Community Goal

Quantopian Goal = $ NOT Community.

Gravatar avatar John O'Leary Yesterday
Grant,

Good points. From what I could gather by looking at other similar ventures, there is some serious red tape in regards to the world of hedge funds and how they take and disperse money to their investors, especially if most of their clients are small-time. Makes a more discretely-managed fund hard to create. "Leveling Wall Street" isn't going to happen unless the regulations change.

Even so, I'm confident that some new business will emerge that will make better use of crowd-sourcing algorithms for profit, with hopefully a more attractive start than Quantopian. I agree with Grant, I don't like the direction this is going.

Disclaimer: I don't doubt that Quantopian is going to be wildly successful. It's just not what I'm looking for anymore.

Gravatar avatar Guy Fleury Yesterday
The decision Quantopian made is a very difficult one. But, I think, they had to look at the hard numbers and where they wanted to go. The numbers were not viable. So, they took a step back, paused. They might reconsider later, or permanently accept the change to their orientation. They too have to evolve, do the best they can for their people.

I came back to Quantopian last year for three things. To have access to minute US stock market data at low cost, find people who shared my interest: designing worthwhile automated stock trading strategies, and prove to myself that I could reach higher long term performance levels by increasing trading activity using my methdology.

We all know it takes a lot to design interesting strategies. When you throw them in the data fire, they do not all survive. And whatever strategy is designed, it has no chance of going live unless it has passed our individual acid tests. These all seem to end up with words like: profit, edge, alpha, or this no good gizmo thingy.

We have a research platform, did I say it was free, that allows us to test whatever we have in mind. We have access to extensive historical databases. We can code, within limits, billions upon billions of possible trading scenarios.

I will stick around. Continue to use what is made available. Develop strategies with high potential, put them to the test. Port the EOD scenarios I have developed elsewhere to the minute level. Let those concepts prove themselves, and then find ways to port them elsewhere.

What I want is the methodology, the trading procedures to be implemented. Even if I have to convert them to manual afterwards, at least I would have proven to myself, using the Quantopian environment, that it would have worked in the past for an extensive period of time giving me enough confidence in my work to carry it forward to its logical conclusion. Even if it is on another machine using another programming language.

So, yes, I will stick around, for the time being.

Gravatar avatar Kern Winn Yesterday
Disappointing doesn't really even begin to cover it..

Quantopian is essentially now just going to be in the business of harvesting peoples' good work for free (via "contests") and making money off of it. Nice thinking.

Gravatar avatar John O'Leary Yesterday
We have a research platform, did I say it was free, that allows us to test whatever we have in mind. We have access to extensive historical databases. We can code, within limits, billions upon billions of possible trading scenarios.
Guy,

While this is true, it does not need to be exclusive to Quantopian. The only advantage Quantopian has now is that it's easier to utilize than other methods. If you're willing to invest time and effort, platforms like Zipline-live and IBridgePy can offer the same exact functionality and more, with the added benefit of actually still being able to live trade, all for free. And if you ask me, the limitations of Quantopian's platform are too restrictive anyway.

The learning curve is high, but that's the price that you need to pay. Otherwise, you'll need to be shelling out actual money for a decent platform that can perform research, backtest, and live trade.

Gravatar avatar Andrew Kunkel Yesterday
If anyone is interested, I'm working with some people on potentially developing an alternative.

Send me an email at [email protected] or message me here if you're interested in being part of it or contributing to the development in some fashion. Thanks!

Gravatar avatar Daniel Koh Yesterday
If anyone knows of some economical feed for minutely/secondly prices, I think some of us including myself can build something interesting (BYO-VM, options trading).

Gravatar avatar Christopher Bohrer Yesterday
This sucks

Gravatar avatar Thomas Havens Yesterday
@Dan Dunn

I want Zipelinelive to work.

https://www.quantopian.com/data?type=free - This is a nice list.

example:

from quantopian.pipeline import Pipeline
from quantopian.pipeline.data.builtin import USEquityPricing
from quantopian.pipeline.filters.morningstar import Q1500US

Will this example work in zipeline? What data types from the "free" data will be ported over for the new open source project? Is there any path for users who use the paid data to incorporate it in the Zipelinelive? I think its important to scope the branch to the larger community so we can start to plan.

Zipelinelive may be a win for us individual traders in the long run. It will provide a stable platform where the rug can not be pulled out from under us. Im sorry about being upset with management yesterday. I have spent too much energy and time here to stop now. " better to have loved and lost than never loved at all" - @ Burrito Dan

Gravatar avatar L. Williams Yesterday
@Daniel Koh

I dabbled with QuantConnect, but didn't really care for their platform, but they have the granularity you're talking about. Quantopian was much better on all counts with what it offered, with the exception of not having second and tick data. They integrate with IB as well, but not with Robinhood, so you will need a much larger sum to begin trading, though I'm not sure about some of the other brokers. Personally, I'm going to go with SureTrader and simply trade the market manually. As far as live trading goes, none of this make business sense. When you have a product that people want, you sell it. You don't destroy it, unless you have another motive behind what you're doing. I bet you wouldn't have to dig very deep to find out the truth, but I could spend that time coding with NinjaTrader or something else. Quantopian id dead, suspiciously just like Yahoo's historical API. I guess all those sick people who wanted to monopolize data that should be free, and sell it to the public for a profit, actually won over. ;) Let's all sell our souls for jelly rolls.

Gravatar avatar Thomas Havens Yesterday
"simply trade the market manually" - no fking thanks man....

Gravatar avatar Tim Decker Yesterday
Quantopian Team,

Can we get clarity from Quantopian about what exactly will remain, that is publicly available/exposed, after this change? Some questions to answer (I'm certain this is not an exhaustive list):

Will backtesting tools and notebooks still exist?
Will there still be the same(as of today) data available in backtesting and notebooks?
Will Quantopian Live paper trading still exist?
Are there planned changes to remove capability from the Quantopian API (that is not related to live trading with a broker)?
Is all that will remain a 'community' (similar to Stackoverflow) which is merely a forum rather than a forum AND toolkit?
There is some FUD due to the announcement without the clarity of what will still be available to community members.

Gravatar avatar Serge d'Adesky Yesterday
Does Quantiacs permit live trading links?

Gravatar avatar Thomas Chang Yesterday
Hi @John,

You said "The main advantage of this is being able to write and backtest code in Quantopian's web client, then copy/paste the code into IBridgePy.". Are you sure one can also use something like the pipeline by QuantOpian there? Besides, I use quite often or quite a lot of the Built-in or CustomFactor in my algos. I fear that by IBridgePy you can't do that.

Gravatar avatar Daniel Koh Yesterday
Actually it'll still work out for those of us who do live trades, if we can get a trade notification email (so that we can trade manually) instead of actually executing a trade through a broker.

Gravatar avatar Thomas Chang Yesterday
@Tom,

I've heard that by Quantiacs you can not trade equuities but futures, right?

Gravatar avatar Addison Dubay Yesterday
@ Daniel,

Except, it will be delayed 15 minutes.

Gravatar avatar Serge d'Adesky Yesterday
Daniel,

Will they permit an email out? I thought they had that portion of software suppressed to allow any outside signalling.

Gravatar avatar Andrew Hamlet Yesterday
Disappointed by this decision. Will be spending less time on the platform now that broker integration is no longer supported. I imagine others will be too....

Gravatar avatar Lucas Lee Yesterday
Very very disappointed, I do have live algos running, and I pretty happy with them.
Maybe we have to switch to something else eventually, so does anyone know which one is better:
IBridgePy, zipline-live, QuantConnect ?

Gravatar avatar Daniel Koh Yesterday
@Serge

I do not know what guards they have in place in terms of outside messaging, but an engineer will probably need to implement something if they choose to take this route of emailing trades.

Gravatar avatar Lucas Lee Yesterday
Correct me if I am wrong. I think the only api we could use to communicate with outside is fetch_csv, and fetch_csv could only be used in initialize. Quantpian does this for safety purpose?

Gravatar avatar Eric Bell Yesterday
Oh well. I switched to Quantconnect. I can backtest 20 times on futures, equity, forex, and options before I get one backtest in QuantoPain. They allow so many brokers for a small monthly fee, and the great news is you can develop your strategies in C# and Python for free. They only charge like $10 a month for live trading.

You guys inflate your sharpe ratios on backtests to make it look like you can make a bunch of returns. There is no leverage control, and the backtest never included interest on margin. I mean the competition used to have 3x Leverage, I don't think brokers allow anything over 2x Leverage. Beta 0 doesn't make sense, why would I want to be market neutral when the market has mostly gone up over the last 80 years.

Anyways, switch to QuantConnect!!! You won't regret it!

Gravatar avatar Kern Winn Yesterday
What becomes of the datasets? Will they be included in zipline-live, or can they be accessed via IBridgePy, Quantiacs and so on?

I feel like Quantopian just committed e-suicide here, and everyone's scrambling to work out what we're going to do next.

Gravatar avatar Viridian Hawk Yesterday
@Lucas Lee, @Daniel Koh, @Serge d'Adesky Just write a greesemonkey script to monitor the live-trade panel and send you emails --- or better yet, code it to place the trades for you via your broker's API endpoints. Why would Quantopian add functionality like this for you? They've expressed their motivations already. They don't want to have to support features that distract from their goals.

Gravatar avatar Lucas Lee Yesterday
@Viridian Hawk , thank you for your reply. It's the last thing I might do if I cannot find a better solution.
I spent many hours working on my algos, I don't want to spend extra time to port my code to quantconnect, or setup and maintain a new system for my live trading. Unfortunately, it seems that I have to accept the change.

Gravatar avatar Viridian Hawk Yesterday
I think everybody who is posting about all the extra things Quantopian should do to create complex hacks, stopgaps and workarounds are missing the point. (Except me when I suggested Q give us access to the prime broker they use, because that would be the most awesome solution.) They're not going to spend their time and energy developing a workaround new from scratch that would be harder than simply maintaining the current system. Why would they do that? They are moving on from this so that they can stop spending time on it, not so that they can spend more time on it.

Quantopian clearly wants a singular focus on the Q Fund here. They see everything else as a distraction. They're trying to redefine their userbase and community along those lines.

I think everybody who is being left out of that should just check out QuantConnect. Sure it's not a drop-in replacement, but I think for most of the strategies people have been running here, you can replicate the exact same logic there. Perhaps they can make a deal with Robinhood sooner rather than later and get that integration going. They are also powered by open source, so if anybody wants to contribute a Robinhood integration, that might be all that's needed to get it going.

zipline-live is less promising. Depending on your strategy and if you can get the data needed for it, maybe you can get it to work for you. If your strategy simply works based on daily resolution OHLC bars, it could be fine.

@Daniel Koh, you could write a Greasemonkey script to do that for you from the paper trade console.
@Tim Decker, I think they answered at least half of those questions already. Everything pertaining to the contest and the fund remains.
@Stephen Hanly, I don't think the live-trading part is that hard. Sure, it's a hassle to maintain if the broker keeps breaking things, but getting Q's current code base for that wouldn't help with that issue anyways. The hard part to replicate is the data that drives the algorithms here. That's an engineering nightmare and prohibitively expensive. (I also assume that those integrations also involved business deals, which the source code isn't going to help you with.) Here is the IB API: https://www.interactivebrokers.com/en/index.php?f=5041 Here is an unofficial Python API for Robinhood: https://github.com/Jamonek/Robinhood

It is a shame Quantopian didn't spin the live-trading aspect out into a different website -- Powered by the same backend, but with different interfaces and different communities. Lets face it -- live-trading on Q was just kind of tacked on. There's so much room for innovation and improvement -- but it was never their focus so it was never properly nurtured. I think that could explain why after a couple years there weren't more than a couple hundred active live-trade users.

Gravatar avatar Viridian Hawk Yesterday
Hey, everybody. There's a lot of passion here. I just posted a new thread with some questions, partially out of my own curiosity, but also because I thought it might be useful for everybody to be able to tell their own story, and work from there to find solutions:

https://www.quantopian.com/posts/questionnaire-for-quantopian-live-brokerage-traders

Gravatar avatar L. Williams Yesterday
Install Kali Linux. Reverse engineer the Robin Hood platform. Install bluestacks and a sniffer and use the SSL keys to decrypt the protocol stream. This will tell you how orders are placed. Recreate the classes to provide the same functionality we got from Quantopian, minus all the hocus pocus. You could even improve the granularity. Use Quantopian for backtesting your strategy. Use your own code for placing and managing orders. Think about it. What sort of things or we looking to do?

Create Limit and Market Orders
Cancel Orders
Create Stop Orders
Create Stop Limit Orders
Monitor Price

You could code this in a month or less, and give it way for free just to piss people off. I had started down that road when I created Vicentex.com. Then all of a sudden, upon code complete, Yahoo discontinued its historical API and made it impossible to use YQL. That's how I ended up here. Fact is, these guys work really hard to keep us in the dark, because the truth is, this is a multi-billion dollar business. Think about it. You invest your money with a broker, and you're lucky if you get ten percent, but we all know that we can blow that out of the water now don't we? So who needs brokers if we all start down this path? This is about money. Huge firms, who shall remain nameless, are replacing humans with machines as we speak, but these guys don't want us getting rich off our own intellect. Oh no. That would actually "Level" the playing field. Unfortunately for them, they can't stop us. And no, we're not going to build algorithms for millionaires and hope they fund us. Let them build their own. Better yet, tell them go find a good broker. If we can't profit from our own hard, work, why should we contribute to income inequality without an opportunity to capitalize? Sorry, gotta go blue pill on this one. Your matrix has nothing offer.

Gravatar avatar Luke Izlar Yesterday
L. Williams, there is already a robinhood API found out by users that uses python to place orders. The issue as you said is data. I wonder if Quantopian would be willing to supply us data for zipline live as a measure of good faith.

Gravatar avatar Tony Chang Yesterday
@ Eric Bell, 3 months ago when i started to learn algo trade, i made a comparsion on both quantopian and quantconnect. Eventually i picked quantopian because i have zero knowledge on C# "class" (and so C# is too difficult for me to learn). Correct me if i am wrong. I believe python is more "conventional" and use the normal programming logic. C# is something hard for beginners, especially everything is in "class-style" and the programming logic flow is not "sequential" (although i agree that it is even more powerful if we fully understand C#). However, i believe most algo traders are not "IT guys". We want to pay focus on algo rather than learning programming language. That is, if both python and C# can do the same thing, i prefere to learn python because it is much easier to learn.

So do you feel learning C# is difficult ? any guidance on such learning ? i am now struggling in between http://www.ibridgepy.com/ and quantconnect.

Gravatar avatar L. Williams Yesterday
That would certainly be an offering worth of repentance, but what would be in it for them? That way to a man's heart is not through his conscience Luke. It's through his desire. They would never acquiesce to such noble endeavors. Of course, you could use their platform to feed yours, but alas, I've said too much already. The force is not with us on this one.

Gravatar avatar Chris Venne Yesterday
@Luke Izla, Could you share any information/source of the Robinhood API that uses python to place orders which you mention above?
Thanks

Gravatar avatar L. Williams Yesterday
Here's a proposal community. The API that exists allows you to pull data in "from" a spreadsheet via a URL. If the wonderful folks at Quantopian could spend a small amount of time extending this functionality to allow us to export our pipeline results by posting TO a URL, we could leverage the existing platform by importing the output into our own, and utilizing the data wherever we like. We could run this daily, to extract our securities, import them into our newly created portfolio management platform, and execute trades, however we like. This way Quantopian doesn't have to worry about bugs in their code, and we can leverage much of the existing algorithms, that we've painfully created for "leveling the playing field." What say ye gentlemen? This would help you keep your stellar reputation in tact, and your sins would all be forgiven.

Gravatar avatar Minh Yesterday
That is never going to happen. Quantopian provides us with neatly packaged data all for free because we have to use their interface. They're not about to let users export their very expensive data for free.

Gravatar avatar Eudis Duran Yesterday
Are you serious? Why would I continue to use this platform with no live trading support? huge disappointment. moving on to the next best thing.

Gravatar avatar Viridian Hawk Yesterday
@Tony Chang -- My experience with Python is limited to Quantopian, but I will agree, as it is typically used here it's linear, or what is called "Functional Programming." While this is simpler to learn, the other common technique "Object Oriented Programming" is considered the best practice. It's harder to wrap your head around, but in the long run it's cleaner and more maintainable. Fear not though, QuantConnect also supports Python, and their platform is pretty similar to Quantopian, and you can code in whatever style you prefer.
@L. Williams -- nobody need go through all that trouble reverse engineering the Robinhood API. It's well documented if you search Google for the Robinhood API, it's on GitHub. On your other point, just like you can't buy an mp3 or a movie and then offer it for free streaming on your website, Quantopian can't let us stream or download for our own private use the data they're paying for. Also, why would they? If all you need is daily data, you can get it free from Quandl and load it into zipline.
@Chris Venne -- I already posted the link above.

Gravatar avatar Dan Dunn 18 minutes ago
Thanks again, all, for the replies. The follow-up points I made yesterday are relevant to some of the replies that came in recently: the importance of focus for Quantopian, the fact we didn't find a viable business in the broker integration, and, most importantly, our wish that we could support this and our thanks for the community's contributions. I won't repeat those points here at further length.

Several of the comments today talked about how the interest in trading personal capital might translate into algorithms that get an allocation. Like many of you who posted here, we thought that people who were intrigued by trading their own capital would go on to write algorithms that get an allocation. Unfortunately, the data doesn't confirm that. When we tried to figure out why, our conclusion was that the problems are too different. Personal traders are often trying to capture market effects, like beta-to-spy, that aren't highly valued by institutional investors. Personal investors are also optimizing for their account size. When trading personal capital, people generally write long-only or long-biased algos, and they choose strategies and assets that don't scale well. None of these things are "bad" - they're quite reasonable. But they are an aspect of our decision to end brokerage integrations.

I also wanted to confirm that sharing the live data off the Quantopian platform isn't something we can do. We license the data from other parties, and that license is limited to the Quantopian platform. Of course, much of the data we use is available for free, like Quandl and daily price/volume bar data.

A note on moderation: I've had to make a few moderation decisions on posts that violate our terms of use. Please remember that disagreement is fine, but civil tone is important. I'm also moving this post off the "featured" tab - at 160+ replies, it's become unwieldy.

DISCLAIMER

Gravatar avatar Dias Karbayev 9 minutes ago
Zipline-live is in alpha stage!
How can Quantopian recommend it as an alternative? Unless Quantopian cooperates with them and assists on the development, which I think is highly unlikely.
May be I will have to get back to using IB API directly as I used to do 3 years ago, given that now IB has a native Python implementation of the API. It is also worth considering deploying algos on amazon servers manually. By the way, QuantConnect is $20/mo (not $10 for the Prime plan).
I still will use Quantopian for backtesting, at least for some time.