This is excellent! Thanks!

@Seong - You probably meant > 0, not == 0, in line 264, correct?

bravo! forced my hand now... i gotta use this! amazing.

First of all, great partnership, very exciting and definitely something I'll be trying out.

I presume this means you'll move the commission structure to 0 in the contest instead of the current default content commission which is now even further from matching reality?

Some questions:

• Does Robinhood have a minimum balance?
  
• Do they transact fractional shares, or only whole shares?
  
• Presumably, they have to conform to the same $25K minimum for an account that'll support shorting and clearing trades immediately, correct?
  

@Kevin - Robinhood does not allow shorting; Quantopian requires at least one short position if you want to be invested overnight.

https://support.robinhood.com/hc/en-us/articles/202844889-Account-Requirements

To apply for a Robinhood account, you must:
be 18 years or older (custodial accounts are coming soon)
have a valid Social Security Number (not a Taxpayer Identification Number)
have a legal U.S. residential address
be a U.S. citizen or U.S. permanent resident or have a valid U.S. visa

Australians interested in Robinhood may signup for our waitlist here.

We hope to announce expansion plans for more countries this year. Please note that we are still in the early phases of launching internationally and it may take several months before we officially launch.

This is terrific! I have spent a few months away from Quantopian, but this has given me a real reason to start back in again.

I know that Robinhood started out only supporting Long positions, but I also know they have been testing margin accounts for several months that can short stocks. Does anyone know if they are offering margin accounts for use with Quantopian?

I'd also like to know more about the "day-trader" rule that requires 25k.

So glad to hear the positive feedback.

Margin trading in Robinhood is currently blocked in Quantopian. We're going to start working on supporting this soon. We kept it out of scope at first to reach this initial milestone.

Generally, we'll refer you to Robinhood for specific questions about accounts and their policies: https://support.robinhood.com/hc/en-us

The Reg-T pattern day trader rule applies to all US persons; it won't be any different with Robinhood.

Per https://support.robinhood.com/hc/en-us/articles/203640049-Account-Types:

Robinhood currently supports individual cash accounts. There is no account minimum for cash accounts.

Regarding settlement, https://support.robinhood.com/hc/en-us/articles/203585565-Stock-Settlement:

The stock settlement time frame is the trade date plus three business days (T+3). This means when a trade is executed, the brokerage firm must deliver the stock or cash no later than three business days after the trade date.

But it sounds like it transactions could clear faster ("...no later than...")? I guess I don't understand. Is there are roomful of accountants in green eye shades reviewing the transactions? Wouldn't the transaction and all associated accounting be instantaneous?

Also, in https://support.robinhood.com/hc/en-us/articles/203585005-Day-Trading, I see:

Please note that there is no limit on Day Trades in a cash account.

So, if the stock/cash from a trade is delivered immediately (which apparently it could be), then it sounds like day trading is possible without $25K and a margin account? Or is it uncommon for transactions to go through immediately?

I haven't found anything on fractional shares yet. I gather that trading in something like AAPL at $119 per share isn't gonna work so well with $500 in capital.

Any plans at Quantopian to support the T+3 rule for backtesting? We have the set_long_only trading guard, but nothing to delay cash appearing after a trade.

The Robinhood framework algo added to the docs this week provides code for simulating this in backtesting and then for handling the situation in live trading: https://www.quantopian.com/help/#sample-robinhood

Which brokers allows you to buy/sell fractional shares ? the only securities that allows to do so , are mutual funds. The only way you can acquire fractional shares is if the stocks is going through a stock split, DRIP, and other corp. action. , you cannot acquire them from the market ( right ?? )

Correct, you cannot trade fractional shares in stocks.

http://www.wsj.com/articles/SB10001424052702303657404576366051874343550

https://www.betterment.com/resources/investment-strategy/fractional-shares-diversification-portfolio-investing/

"Betterment customers' trades and resulting positions can be fractional—down to 1/1,000,000th of a share."

so essentially , share builder operates as a mutual fund.

To answer Grant's question about turn-around times: From my personal experience with Robinhood, it has ALWAYS taken T+3 days before I can use the cash again (they call it "buying power" in the app).

Perhaps there's different rules here in Canada. With my Broker , I do not have to wait 3 days for the trade to settle before buying another stock . I can buy a stock today , sell it tomorrow, and use the proceed to buy another stock . HOWEVER, if i want to make a withdrawal from my account , in that case i'll to wait till the trade is settled ( t + 3) before the withdrawal request is process .

I better check with IB Canada if this apply to them as well, before going live in a few weeks.

Was that a cash account or a margin account? Margin accounts you can buy and sell freely. Cash accounts have to settle, in my experience.

it's with a cash account , TFSA ( Tax Free Saving Account...almost like an IRA) with no margin. We have some clients day trading in them.

@Lionel, I think US is no different. Buy, Sell, Buy is ok - you just can't do the second sell until the prior sell is settled.

This assumes you already had the funds to purchase to begin with in both cases.

For example see: https://research.scottrade.com/knowledgecenter/Public/education/Article?docId=4a6477b554ba4533a91ea5835c5aa4e2
See point 4 under "Example – Free Riding using unsettled funds"

Are futures and options available too ?

@ Ken, I just confirmed the information with a collegue of mine. In Cash account in Canada and most European countries, clients can day trade , no need to wait T + 3 to place another order. In the US this possible only in margin account and if you have $25,000. LikeI mentioned above , I have placed multiple buy and sell orders on the same day. in my TFSA ( no margin CASH account) in the past.

I see Tristan's response above, "it has ALWAYS taken T+3 days before I can use the cash again" but I still don't understand why? Is it a legal requirement that clearing the transaction is delayed by 3 days? It is mysterious, because I would think that there would be a retail market to clear immediately, so that the cash would be available for trading, no? In other words, if there are no legal impediments, why isn't a broker offering immediate clearing on cash accounts?

Interesting Lionel, I thought I was confirming what you said, but good for you all non-US folks to have better deals - didn't realize settlement rules were different too. Makes sense, since it's an SEC thing. Guess YMMV depending upon your country.

Looking at https://research.scottrade.com/knowledgecenter/Public/education/article?docid=d1113a4f4cb847adbe9d32837739cde8&lang=EN :

The process of completing an order is called clearance and settlement. By law, the final transfer of stock ownership must be completed within three business days of the trade. The transfer happens in three steps. First, the number of shares bought and sold is confirmed to be the same. Second, the seller must be credited with payment from money that is transferred out of the buyer's account. Third, the shares must be transferred from seller to buyer.

It is not clear why all of this couldn't happen instantaneously. It's all electronic, right? Or have some steps been left out?

Brokers take three days to settle trades because they are allowed to. They're making money on the float.

There are some trades that need some time to settle, so the regulations need to allow for those, and since distinguishing the ones that require more time from the ones that don't isn't straightforward, the regulations allow the same settlement time for all of them.

I suppose a broker could differentiate themselves with faster settlement times, but would that give them enough extra business to make up for the lost float? Apparently most brokers don't think so.

@Grant,

Trade settlement is a bit more complex than the Scottrade post which ignores the affirmation step in the trade confirmation, affirmation, settlement process. Essentially, you have at least three parties for every trade; the buyer, the seller and the bank (custodian). All trades on most exchanges require settlement through a depository (e.g. DTCC) and not all market participants are DTCC members; certainly retail investors like you and me are not. Many brokers such as IB may use a separate clearing broker to settle trades and it is possible that the counter-party to your trade has both a DTCC and non-DTCC member in the settlement process. The 3 day clearing rule is a DTCC requirement designed to meet not only the highly automated trade settlements from sophisticated member firms, but also the less automated and even manual process of trade settlement of retail investors using street name, non depository certificates.

It is somewhat frustrating how little progress has been made in same day settlements of equities (not a problem for treasuries or other investments), but all in all it should not be a factor in rational (!!!!!) trading, since brokerage accounts generally use accrual (trade date) accounting and the actual cash settlement is hidden behind your account margin buffer. However, for the highly levered, high frequency wannabe's - good luck!

--EDIT
Robinhood is not a DTCC member, AFAIK, and as others pointed out, only offer cash accounts and not margin accounts.

@Jonathan Kamens who wrote:

Brokers take three days to settle trades because they are allowed to. They're making money on the float.

I don't think this is an accurate statement. First, float in a low interest rate world and with a fed window rate less than 1% (0.75%) leaves little float to be had. Second, brokers tend to have equal amount of DVP (delivery Vs payment) and RVP (receipts Vs payment) that cancel their cash balances. Last, banks are the ones who in prior interest rate environments made money from float, but that is history, old history.

Finally, the brokers have some say as DTCC members, but not the final say in the T+3 settlement process. The T+3 requirement comes from DTCC and not individual brokers, who offer margin trading to their clients to remove this impediment.

One last thought, that may be hard to follow.

The latest and greatest in trade settlement innovation is seen with Patrick Byrne (CEO of Overstock.com) who launched t0, where the trade is the settlement using an implementation of the Bitcoin blockchain. I was a huge user of IB more than a decade ago when their API was innovative and accessible to retail investors. In 2015, this is old news and hardly innovative. Perhaps, if Quantopian was partnering with t0, that would be cool, but partnering with brokers who offer "free commissions" is old school and highly, umm should I say, predatory.

LTBBW

Incredible. Now there's no excuse for me to not test algo trading in an expense free environment. Thank you.

Sally,

Not sure I get the "predatory" part. Please elaborate. Are there some hidden costs?

Grant

Grant,

Saw this in "quora" when questioning what the bid/ask spreads would be in Robinhood

Its just not possible to trade, even as an amateur, if the money isn't immediately available to cycle back into another trade. Third, the UI is confusing and not at all intuitive. Fourth, no limit orders. Only market. Yesterday, I had a market order offer price from Robinhood that was a full dollar below market price. Think about that, there's a $1 spread from buy to sell on a completely liquid security like a Netflix, Amazon, Facebook, etc. Its just inexcusable.
Blockquote
Don't know how accurate that is as I do not have a Robinhood account.

But since you cannot place a "limit" order I would NOT want to trade anything but the most liquid stocks in this fashion, and still would be leery. I certainly would not want to day trade even if I was willing to commit the + 25k needed for the SEC requirement.

From https://support.robinhood.com/hc/en-us/articles/205002845-Order-Types-

Robinhood supports market orders, limit orders, stop limit orders, and stop orders. Certain orders may be entered as good for the day or good till canceled (GTC).

There is no free lunch. I would not trust any broker who claims a free service in executing trades - it makes me wonder how they get paid. If you can't understand how someone gets paid, it is best not to do business with them. Many brokers get paid for order flow, by taking a large bid/ask spread or by crossing orders all of which dwarf commission costs. Quantopian claims:

This partnership has made the process of algorithmic trading, from start-to-finish, completely free.

worries me.

Frankly, I don't really care if Q makes their money by charging for seminars or by peddling data sources. Buy when they promote brokers, they should disclose any revenue sharing arrangements with the community. Full disclosure as they say. Anyways, whenever a broker promises free trades, try to understand how they get paid.

From Robinhood

Execution

In certain Internet forums, a minority have wondered if Robinhood marks up orders intentionally to make money on the bid-ask spread. We want to reassure the community that we do not mark up orders to make money. In fact, this behavior is strictly prohibited. The SEC’s Regulation NMS requires that brokerages guarantee the NBBO (National Best Bid and Offer).

For information on how we make money, please click here.

Robinhood makes money in many of the same ways as traditional online brokerages. These include:

• Collecting interest from customers who choose to upgrade to a margin account. We are testing margin in beta and will offer margin accounts later this year.

• Accruing interest from customers’ uninvested cash balances. It is important to note that our customers are not charged.

Robinhood has raised $66 million from NEA, Index Ventures, Ribbit Capital, Vaizra Investments, Google Ventures, Andreessen Horowitz, Social Leverage, Box CEO Aaron Levie, Path CEO Dave Morin, Jared Leto, Snoop Dogg, Linkin Park, and Nas, which gives us the freedom to focus on building an outstanding experience rather than short-term profits.

So I guess they don't markup the bid/ask spread

Well, I suppose Q could launch identical real-money algos with IB and Robinhood, and compare the results. I think this is aimed at tinkerers like me, who could put up $500 in capital to see if a few bucks could be made. The account minimum and $1 per trade cost at IB are non-starters, so it's cool that Q has pulled this off. Their initial effort with E-Trade didn't really make sense, since the per-trade costs were so high.

Market makers (HFT) pay good money for uninformed order flow, they could be taking some of those payments too.

This is awesome.

Let's be honest Robinhood, your model is payment for order flow as Simon pointed out. I find it more than a little disingenuous that they don't just disclose that up front. It's a bit insulting to our intelligence, really, that they expect us to believe they're going to make any appreciable amount of money off the accrued interest from investor's uninvested balances. By law they can't invest that money in anything that yields much more than 1%, which means for every $1M in revenue they get from this method they need $100M in uninvested cash lying about. Given the valuation implied by their investments, and the cost to run the company, they're going to need revenues of at least $100M/year. Which would mean they would need $10B in uninvested assets lying around. Not $10B in assets under management, $10B uninvested just sitting in customer accounts. Margin revenue is a little more lucrative, although one would think that the typical margin customer doesn't have much overlap with the customer who would find Robinhood attractive. In the end I guarantee that if you saw their pitch deck you'd see both those items contributing less than 25% of their projected revenue (love to see them prove me wrong with a copy of their deck though!). The rest has to come from somewhere, and that somewhere is payment for order flow. Which translates directly to worse execution than direct to exchange orders. The magnitude of which is probably small enough not to matter for a small buy and hold account of the type Robinhood's original business model is supposed to appeal to, but might make a big difference to a Quantopian based algo that trades several times a day. I like Grant's ideas, I might put a couple K into matching accounts just to test it out.

I think your skepticism is healthy and I'd be interested in the results of such a side-by-side test. That said, I'll refer you to Robinhood's own support article, updated as recently as Thursday (i.e. it is presumably up to date) which states the following on the topic of Payment for Order Flow (PFOF):

Prior to launching the product, we had some language saying that we
intended to generate revenue through PFOF. The reason we removed this
was because currently we execute all of our orders through our
clearing partner APEX Clearing; as a result, Robinhood does not
directly receive PFOF and any revenue we indirectly generate from it
is negligible. If we were ever to receive it directly, or if it
becomes a significant source of revenue, we will reintroduce that
language into the FAQ.

https://support.robinhood.com/hc/en-us/articles/204336175-Order-Routing-and-Exchanges

Thank you Josh!

Can you tell us if Quantopian and Robinhood have any revenue sharing agreement?

Josh,
Thanks for that, although if you read the full link it's less than fully open and honest about how PFOF works. They claim PFOF has no impact on the price a customer obtains, which just isn't true. It also looks like they aren't taking advantage of dark pools, which you can do with IB (and it hurts to say something good about IB!). Based on my fill experience with them and the price improvement over "NBBO" I regularly see through dark pool routing, I'd guess that this alone will hurt Robinhood's execution comparison. I'll stand by my assertion that while they're living off their funding round now, they can't continue to exist without revenue and they only viable place to get that at this point is PFOF. As to Sally's question, what revenue could Robinhood be sharing with Quantopian since they don't have any revenue!

Indeed, when I read that language, my first thought was that they struck a deal with Apex such that they get cost-free execution from Apex (under certain conditions which ensure that orders can be internalized and routed to the highest-rebate venues, such as by disallowing directed orders), and Apex takes the risk and collects the PFOF. Robinhood doesn't have to declare that they "directly" receive PFOF themselves.

None of which is nefarious in and of itself, in the same way that HFT is not nefarious, contrary to what Michael Lewis and IEX would have you believe. Uninformed flow is valuable, because with it, market-makers can earn the spread on market orders without being exposed to undue adverse selection risk. That flow is worth some fraction of the spread + rebate, which they can kick back. If Robinhood's flow becomes more toxic through their opening up the API to informed algorithmic traders, this may change the viability of their business model, but I doubt that will happen.

Hi Josh (you seem to be the moderator here),

Any possibility of Q/Robinhood doing a little real-money study to address some of the questions above? It seems like some real-money algos could be devised and run on both IB and Robinhood. You have the technical horsepower at Q and it wouldn't take that much capital. Or maybe you already did this in evaluating whether to integrate with Robinhood?

And you have some clever managers and backers. Someone must have done a head-scratch and said "Hmm? We're in business. Robinhood is in business. We both need to make money. How is this gonna work?" The headline "Zero Commission Algorithmic Trading" is enticing, but what's the business model? Or is the idea that you'll attract folks who will eventually write algos for the Q hedge fund?

Grant

Oh wow, this is very cool and unexpected to come so soon! Last I saw people were hacking their way into running a virtual android OS with RobinHood installed and a pretty hacky python package to do this.

Great partnership. Surprised to see the dissension in the crowd. Quantopian just doubled their available brokers. That's a great thing. Keep it up. I agree with some others that I'd really like to see t0 next.

This is AWESOME! I've been waiting for this for so long and it's finally HERE! Good job Quantopian!

yooo hoooo!

Has anyone deployed this on RH yet? Not sure how they link up when they only ask for an account name?

Just now, when I opted to Live Trade with my Robinhood account, a new tab opened in the browser and I was prompted for my Robinhood account name and password.

Thanks Robert. Was the account name your account# as listed in your monthly statements?

No, it was my Robinhood username.

Thanks. I got it setup now. Chris

Very good!

As Robert said, it's the username you use to login to your account. i.e. the same prompt as when you login to your Robinhood account on an iPhone (not sure about androids).

In live trading with Robinhood, is the amount of unsettled funds reported back to the Quantopian algo every minute?

Indirectly, yes.

For cash accounts, you can compute that number by asking for

unsettled_funds = context.portfolio.cash - context.account.settled_cash

Does anyone have comparison data regarding execution quality for Robinhood vs IB?
Assuming one isn't interested in non-equity securities of course.

Thanks Seong,

When does Robinhood set context.portfolio.cash & context.account.settled_cash? Would they change as soon as an order gets filled (either partially or fully) and then get reported back to the algo immediately (at the next whole minute)?

When does Robinhood start the T+3 "clock" to determine when the funds would be settled from a transaction? Is it typically/always a close-of-business transaction, or would one expect unsettled_funds to be freed up during the trading day?

Grant

Hi Grant,

I'm not entirely sure about the T+3 'clock' but as for the first question, yes, the algo would know that cash and settled_cash has changed. From my experience, the change happens in tandem with the order (which is almost immediately).

Also, a quick update: A common behavior among brokerages is to convert market orders into a limit order with some percent buffer to protect the investor by preventing an order from being executed at some wildly different price. In this case, Robinhood does the same with a 5% increase. So your market order for share XYZ at last traded price $100 will be submitted as a limit order with $100 * 1.05 = $105. It's important to insure that this final limit order price is under your buying power.

The sample algorithm using order_for_robinhood has been modified to include a 5% buffer and for your convenience is attached to this reply and updated to the original post.

Happy to answer any questions

Thanks Seong,

Note that your posted backtest immediately above is not showing. Everything is blank.

Regarding the T+3 timing, do unsettled funds become available during the trading day? For example, if my algo ties up funds at noon on Monday, would they be available at noon on Thursday? Or at the open on Friday? According to http://www.sec.gov/investor/pubs/tplus3.htm, it sounds like the unsettled funds will be freed up at the open on Friday, worst-case, but maybe Robinhood moves the money faster?

It's important to insure that this final limit order price is under your buying power.

Are you saying that if my algo attempts to buy 1 share of XYZ with a market order and my buying power is $100, Robinhood will reject the order? But if my buying power is $105, Robinhood will accept the order and convert it to a limit order?

According to https://support.robinhood.com/hc/en-us/articles/205002845-Order-Types-, Robinhood supports other order types. So how is the buying power considered for those? Taking your example, say XYZ last traded at $100 and I submit a limit order at $110, but my buying power is $105. Would Robinhood reject the order? Or would they shift the limit to $105?

Are orders submitted to Robinhood automatically cancelled at the end of the trading day (as they are with IB)? According to https://support.robinhood.com/hc/en-us/articles/205002845-Order-Types- :

Robinhood supports market orders, limit orders, stop limit orders, and stop orders. Certain orders may be entered as good for the day or good till canceled (GTC).

So, through Quantopian, is it possible to enter a GTC order to Robinhood? Or are you automatically entering all orders as good for the day?

What is Quantopian's interpretation of https://support.robinhood.com/hc/en-us/articles/202853769-How-Robinhood-Makes-Money ? Are they telling the whole story? Kevin Q.'s comments above seem valid. One concern I have is that unless there is a path for Robinhood (and Quantopian, for that matter) to become a viable business, it could be another dot-com bust. What's your confidence that they know what they are doing?

Great news! Does it mean that contest comissions will be set to 0?

Hey everyone!

I have had an account from Robinhood for quite some time now, and although I'm new to Quantopian I can at least answer some basic questions.

Grant: Funds do not become available during the day - they become available at open of the appropriate business day. So if you close a trade at noon on Monday, you will have the funds settled when the market opens on Friday. I personally would not suggest sending market orders to Robinhood, because I found the slippage to be unpleasant.

Seong: settled_cash at all time should be equivalent to our buying power, correct?

Andrew - I have also noticed that I get filled at worse values than I was expecting on my market orders. I have started to use limit orders at the current price and that seems to be working fine. (Note: I do not know if this problem is worse than any other broker when using market orders)

When deployed to live trading the last line throws an error message. Should i take out the last line of code about "leverage" if i just have a regular cash account?

Hi Matt,

Do you mean the record(context.account.leverage) is throwing an error for you? I believe I've seen that happen in cash accounts with no funds, is that the same in your case?

@Seong, yep, lack of cash could definitely be my problem. :) Thank you

Wanted to see if i could deploy this algo, never done it before so I was happy I got it to somewhat work.

@Andrew Nelson,

Yes while you're deployed Live, settled_cash should be equal to buying_power at all times on Robinhood cash accounts.

However in backtesting, settled_cash == context.portfolio.cash while buying_power == Infinity you can find out more information here (https://www.quantopian.com/help#api-account)

@Matt

Great to hear. I'm happy to help if you have other questions

have any one tried using this algorithm for robinhood? I can't seem to get it through backtesting as it is throwing a runtime error.

Hi Sam,

Please try cloning the backtest again. For your convenience, I've attached it here.

Robinhood has raised funds from:

Jared Leto, Snoop Dogg, Linkin Park, and Nas

Not only is it cheap, it's also funded by some of the coolest investors there are.

Does anyone know what kind of interest Robinhood charges on short positions?

Thanks,
-Paul Devlin

Turns out they don't allow short sales yet. Here's the reply I got when I asked them the very same question about short interest:

Thanks for reaching out!

At this time, we do not offer a short selling feature, but we will add this functionality in the future. To learn more about order types we do support, please see our FAQ: https://support.robinhood.com/hc/en-us/articles/205002845-Order-types-

Sincerely,
The Robinhood Team
robinhood.com

Interesting. I'm only concerned because their website made it sound like interest on short positions is going to be their biggest source of income. I guess it doesn't necessarily have to cost more than just plain commissions though.

They've gotta make money somehow, I'm guessing the plan is to build up the customer base as fast as possible with the VC funds then start getting payment for order flow and jacking up the short borrow fees.

The latest version exits with a runtime error when using small capital amount ~ $10k

Has anyone used this for live trading yet?

Hi Sam,

We're currently running this algorithm live with real money. I've tested the latest version with ~$10k and it runs without errors for me.

Which backtest periods were you using?

Seong

It is only about 2% per year thru 2015, benchmark is 14% per year.

A few folks have been asking about Limit, Stop, and StopLimit Orders with the order_for_robinhood method. Here's an updated source code for just that.

All you'll need to do is pass in the order style to order_for_robinhood: e.g. order_for_robinhood(context, symbol('AAPL'), .5, LimitOrder(100.0))

Hi can we place both buy and sell limit orders in the code for all the stocks we want to purchase or that we are exposed to? ?What is the code and what line is the code placed in line 202 to buy limit order and line 288 to sell limit order ? thanks

Hi Bo,

Would love to help you out. Could you clarify what you're trying to do? Are you trying to place sell limit orders for ALL stocks that you currently hold?

Hi Seong,
correct I wanted to know how to code buy and sell limit orders for entire portfolio or all stocks that I will purchase and then sell on Robinhood. Thank you for your help.

Hi Bo,

You can find ways to submit buy limit orders in the open_new_positions method.

or security in context.increase_exposure:  
        if security in data:  
            o_id = order_for_robinhood(context,  
                                       security,  
                                       context.weight,  
                                       LimitOrder(data[security].price))  

As for selling all positions, you can go through your existing portfolio positions and exit them one by one like:

for stock in context.portfolio.positions:  
    if context.portfolio.positions[stock].amount != 0:  
        order_for_robinhood(context, stock,  0.0, LimitOrder(limit_price))  

Are orders placed with Robinhood good til cancelled? Or will they be cancelled automatically at the end of the trading day?

For Quantopian live trading, they are good till the end of the trading day.

Hey guys, I'm sure this probably has been answered previously, but Robinhood is for long-only...correct?

Thanks!

Correct.

For those looking to simulate a T+3 Cash settlement date after a sale in their backtests. You can check out the framework I'm attaching here. It basically looks to see if 3 business days have passed (does not include holidays).

Hope it helps.

For those looking to simulate a T+3 Cash settlement date after a sale in their backtests. You can check out the framework I'm attaching here. It basically looks to see if 3 business days have passed (does not include holidays).

Hope it helps.

Hi Seong,

Glancing over the code, it appears that a sale of any amount will trigger a 3-day black-out period, or am I missing something? Isn't it that only the proceeds from the sale aren't available for 3 days?

Grant

To Grant, that is my understanding as well. if there is any sale, it prohibits any buy even if there is cash available for the transaction

Fractional shares when???

@ Jeff,

I'm not sure I follow. Say I used $1,000 of $10,000 in cash to buy some stock, leaving $9,000 in cash. I then sell the shares (which I could do at any time), and end up with $10,100 in cash ($100 profit). My understanding is that the $1,100 from the sale would not be available for 3 days, but the $9,000 would be unencumbered. So, I don't see how the code Seong posted would track the $1,100 separately? Instead, if I understand correctly, effectively settled_cash (buying_power) goes to zero for three days?

@ Seong,

Have you asked Robinhood why they wait 3 days? Perhaps you could ask them to explain on this forum, since my understanding is that they could transact faster from a legal standpoint.

Hi Seong,

Any response to my question above regarding your code to handle the T+3 rule? To do backtesting, what's the right way to handle it (as I mentioned, your code seems to block all trading for 3 days, instead of using settled_cash (buying_power))?

Also, I understand Q has an algo running with Robinhood. Will you be publishing the results?

Grant

From what I understand Grant Kiehne is that Robin Hood makes it's money from funds not currently invested. However, there is now Robin Hood instant.

Question:

Is there a list of methods specific to robinhood anywhere? Or are they the same as the rest of Quantopian? So far, I only know of "order_for_robinhood". Also, do I need to specify that my environment is robinhood?

Is it generally true that order_target_percent() won't work under Robinhood & T+3 for re-balancing/re-allocating a portfolio? For example, say the algo does:

for stock in context.stocks:  
    order_target_percent(stock,weight[stock])  

Then, the code runs a week later with different weights. Will the portfolio not get re-allocated because of the T+3 rule?

Also, what's the story with Robinhood Instant?

Is it generally true that order_target_percent() won't work under Robinhood & T+3 for re-balancing/re-allocating a portfolio? For example, say the algo does:

for stock in context.stocks:  
    order_target_percent(stock,weight[stock])  

Then, the code runs a week later with different weights. Will the portfolio not get re-allocated because of the T+3 rule?

Also, what's the story with Robinhood Instant?

Grant,

For Robinhood Instant questions, you will have to reach out to Robinhood's Support Team!

As for the order_target_percent method, it's tricky. Order_for_robinhood was created as a fool proof way for you to not create any orders as long as there are unsettled funds in your account AND to only trade on the stocks that belong in the algorithm (ie. so it won't mess with your existing positions).

We also added functionality to prohibit sales when there are any unsettled funds (once again because we wanted the sample algorithm to be as fool-proof as possible) but there is room for adding enhancements like trading on available cash and so forth. All are welcome to add that functionality and would like to see what others come up with (I'd be happy to error-proof the code as well)!

Specifically for order_target_percent, it uses the whole portfolio value (including unsettled funds) to create percentage weights. So if you still have unsettled funds in your account, order_target_percent will still take that into consideration while creating orders which may end up in unexpected behavior. If you do not have any unsettled_funds (ie. you wait 3 days after a sale and check in your code that context.portfolio.cash == context.account.settled_cash , order_target_percent should work as expected.

Jared,

The Robinhood specific methods you are looking for are found in the sample algorithm above and on this thread. I'm happy to clarify any questions you might have on them.

Thanks,
Seong

Am I wrong or the handling of T+3 rule (and so settled_cash vs unsettled_cash cash) applies to IB cash accounts too?
Just trying to understand why this very important detail has been so discussed with Robinhood integration and not so much when IB integration was announced. Maybe the majority of people have a margin account with IB?

Yes, this same issue should affect IB cash accounts. I would be working on a solution myself, except that my IB cash retirement accounts are all denominated in CAD and therefore unusable for a different reason!

But yes to your second question too, I would assume that the vast majority of IB traders using Quantopian both have margin accounts, as well as more than $25k total capital with them, due to the Reg-T day trading rules.

Simon.

Thanks Simon. You also raised another issue I didn't think about, the retirement accounts. That's a pity, I won't be able to use mine either(at least not via Q).

I always develop a buy event that is 3 days later then the sell event and pre-check for overspending... annoying but needed for live trading in cash accounts

Thanks Seong,

Well, supposing you know something about Robinhood Instant, presumably it would be supported by Quantopian? As best I can tell, it is kinda-sorta like margin, except only up to $1000 covering the 3-day clearing period, without shorting. I also saw that traders are still subject to rules against pattern day trading. And anti money laundering rules somehow need to be considered, as well. It does sound like a recipe for disaster. Let's see, a billion smart phones, times $1000 per phone, is a trillion dollars...what could go wrong?

Regarding the T+3 business, how do mutual funds deal with it? Do they have to wait 3 days, too? Or does it only apply to the little guy?

Can I suggest Quantopian to consider adding TradeZero to its available brokers?

I have just read about it and it seems appealing for non-US citizens
http://www.tradingschools.org/reviews/trade-zero/

Seong,

Also, presumably Q does not intend to profit from its relationship with Robinhood, since your main business is the hedge fund. Is this correct?

Grant

Hello all,

I am trying to modify the algorithm so instead of aiming for 20% for each asset whose short SMA > long SMA, it divides the cash between all assets with short SMA > long SMA. Ex if 3 out of 5 are OK, it invests 33% on each and 0% in cash instead of 20% on each and 40 % cash.

From what I unsterdand, the algorithm never rebalances positions since we call order_for_robinhood only on the reduce or increase exposure lists.
In these list we never consider position where short SMA > long SMA but are invested (we only consider those whose amount is 0, confer line 243)
However I thought it was supposed to since in order_for_robinhood we check
If position is within 1% of target weight (line 166).

I am a bit confused ... I am missing something?

Thank-you for your help! I added the modifiied algorithm, I but #Modification comments everywhere I changed the algorithm.

N.B.: I tried to call order_for_rebinhood for every asset instead of only assets in increase or reduce exposure lists, but I got an error :
AttributeError: 'NoneType' object has no attribute 'amount'
... USER ALGORITHM:308, in open_new_positions
(get_order(o_id).amount, security.symbol))

Hi Nicolas,

You are getting an error on line 306 because get_order(o_id) is returning None because order_for_robinhood didn't actually need to order anything (ie. it didn't find a 1% difference and returned None). In this case, you'd want to do something like:

if get_order(o_id):  
                log.info("Ordering %s shares of %s" %  
                     (get_order(o_id).amount, security.symbol))  

However, I'm not quite sure what your other issue was. Would you mind clarifying what behavior you're expecting and how it's different from what you're getting?

Thanks

Hello Seong Lee,

Thank-you for your help, I think I misunderstood how the algorithm works. I thought that we were supposed to rebalance position every month, and only if we were more than 1% out of range. However, the algorithm is not suppose to rebalance, and the 1% check is only a check.

I have two questions:
1) Why do I have very different results with daily data ? I though the algorithm should work with daily data. I even get a negative position on IEF ETF :(

2) Let say you have $10 000 that are invested at some point on 4 out of 5 EFT, with 80% of your cash (assume you are max leverage is 100% to simplify), if assets goes up let say 10%, you will have $8 000*1.1= $8 800 in ETF and $ 2000 in cash. Total is $10 800.
If the fifth one goes to the increase_exposure list, the algorithm will try to buy 20% of portfolio value which is $2 160. As you only have $2 000 in cash, the order can't be fulfilled without leverage. Either I missed something, or the algorithm is not made to manage this specific case.
In my opinion, you either need to rebalance each month, or you have to check if there is enough available cash to fulfill the order.

I may very well have missed something in the algorithm! (in the end I try to adapt the algorithm so be invested at 100% on every securities where short SMA > long SMA, that is why I need to rebalance, and why this question came up)

Regards,

Nicolas

Hi Guys, as this is my first post here I would like to say what an amazing platform! Excellent work.
I am new to Quantopian and I am currently experimenting with this 'Mebane Faber’s Tactical Asset Allocation' model. For the most part it works very well and i like the concept, mainly because it is able to stop loss at critical points, a big pain for investors and hedge funds as investors withdraw their funds. Where I think there is a drawback is at the beginning of the implementation of the model. Let me explain:
First I opted not to trade until the desired condition is reached (i.e there will be no trades until the 20 day average exceeds the 200 day) (context.first_trade = False #(was True)).
since the algo needs to look back 200 days, I shouldn't have been able to run it from the beginning of time (Jan 3 2002) as data are not available prior to that date, but I can. Therefore I think that this algo begins trading as follows: first 20 days does not trade at all because the 20 and the 200 day average is the same as both are less than 20. thereafter if 21>20 it trades then 22>20 trades and so on).
Even if I start trading from a different date(i.e Jan 1 2005) it starts trading from that point on, meaning in the beginning of the algo there is never trades after the 20 days pass it will trade based on the conditions I gave above.
If possible can you correct this bug?

Thanks for your help

Cheers

Does Robinhood support MOC orders?

Quick update:

Robinhood has just announced Robinhood Instant. This removes the need for cash settlement dates and T+3 waiting periods. So if you have Robinhood Instant, you'll no longer need to check for unsettled cash.

You can find out more information here (http://blog.robinhood.com/news/2016/2/22/introducing-robinhood-instant)

And Hernan,

While I'm not quite sure about Robinhood, Quantopian does not support MOC orders at the moment!

Thanks,
Seong

Hey all,

So I deployed a Robinhood algo before trading yesterday (2/22/16). It was not running this morning when I checked around 10:30 E.T. Had to sign back in to Robinhood before it resumed trading the algo. I am wondering if this happened to anyone else? I am guessing the deployment of Robinhood Instant may have created a one time instance where user accounts were logged out?

Seong, in the blog post that you linked it states that in order to obtain Robinhood Instant, one has to share their referral link and hope that they are invited to Robinhood Instant. Is this correct? Even for Quantopian users?

You're pending invite if you already have a Robinhood account. Referrals move you further ahead in the invite line. There's a post on your Robinhood homepage that will show you your current position in the invite queue.

Seong, I think you still need to check for unsettled cash. Yes, you can now purchase with unsettled cash; however, Robinhood Instant isn't a margin account and you can't sell a security purchased with unsettled cash until after the settlement date or you'll have committed a free ride.

Thanks for the update Seong! I am very impressed and surprised that Robinhood is able to do this.

Once your account has been activated for Robinhood Instant, you can buy and sell stocks (long-only) without waiting the 3-days.

Things to keep in mind:

1) Always use limit orders. I am not sure if this is unique to Robinhood, but when I have done market orders I always get a price that is worse than the current price. I have not used limit orders within Quantopian yet, but I think it is possible.

2) Robinhood is currently long only. Short sales will require a margin account, which I think Robinhood is beta testing. Anyone know what Robinhood's margin rates are?

3) The standard SEC day-trading rules still apply. I think the Robinhood app will have optional safeguards to prevent you from triggering this rule, but not sure about the API that Quantopian uses.

Hi Seong, how can we adapt the Mebane Faber's "Global Tactical Asset
Allocation" (GTAA) algo with Robinhood instant? thanks

Hi Ioannis,

I've made a few instructions including some warnings here! (https://www.quantopian.com/posts/robinhood-live-trading-update-stop-waiting-3-days-with-robinhood-instant)

Seong

I tried to sign up for Robinhood, and got the message:

Sorry!

At this time, Robinhood is only available on iOS and Android. You can
learn more about supported devices at the Robinhood Help Center.

Believe it or not, I don't actually have a smart phone (and if I did, why would I want to use it for trading, anyway?). You might want to talk with Robinhood to see if they could support accounts that don't require an app on a smart phone. Why would I want an app with Quantopian/Robinhood? Perhaps there is a work-around?

Grant,

I have never done it, but I bet you could just install an Android OS on your PC and get your account that way.

I actually run a Android emulator on a system at work, and I place all my orders through it that way when I'm in the shop.

Is it possible to run this algorithm only on available cash & on selected securities alone? I do not want to exercise all the long-term stocks that I have plans to hold for more than 10 years.

Bump to Karthick's Question. I am wanting to accomplish the same thing.

@Karthick,
If you have a GTC manually placed limit order on a security, Quantopian orders for that security are rejected (I think by Robinhood).
In regards to cash available, you can even limit the algo to only using a certain percent of it.

Bump to Karthick's Question. Wondering the same thing.

Chris, Karthik: Clone the current algorithm. Its supposed to do this already. Only assets in context.assets are traded, so make sure any you don't want to trade are not in that list.

Is robinhood integration available only for hosted zipline?

So can I still live trade with Robinhood on Quantopian?

The Robinhood tutorial is down on youtube mentioned in the first post.

Is there another place I can find information for live trading?

Thanks,

Dan

So can I still live trade with Robinhood on Quantopian?

No. See https://www.quantopian.com/posts/phasing-out-brokerage-integrations.

I'd heard that some folks had figured out how to do custom implementations of http://www.zipline-live.io/, but I'm not aware of anywhere one can do Robinhood-Quantopian trading on a platform like Quantopian had offered.

I'd heard that some folks had figured out how to do custom implementations of http://www.zipline-live.io/, but I'm not aware of anywhere one can do Robinhood-Quantopian trading on a platform like Quantopian had offered.

Been working on one since Q killed off live trading. Its already working internally. Hoping to get a rollout to the public this year.

Hi, I am new comer to Quantopian and I could not open the youtube link for deploying live algorithm to robinhood. Also I would like to know if I want to trade a stock when the price fluctuates more than a certain percentage(buy if increase and sell if decrease), how can I code that?

Have you had a look at the tutorials? They have examples.

Hello! This thread is awesome. I am also new here. Can we talk about your accumulated results since Day One? How have your successes been so far with x number of algorithms?

Where is the question going at? How would the information help you?