"The Greatest Dichotomy"?

James,

Yes, declining rates are at the core of my thoughts on the current macro environment. I absolutely believe that the current run up in stocks is attributable to the low rate environment. I make adjustments to my models based on my perception of what "natural interest rates" should be, which is basically pricing treasuries using traditional credit analysis.

Simon Thornington

I mean, it's the Felder Report. Has it ever been anything but negative on stocks?

The Felder Report notes that the supply of bonds has been diminishing. Does anyone know of a datasource for new bond issuance that can be pulled into a Q algo? Probably need to utilize fetcher, but hoping there is something I missed that is built in.

Rising interest rates are easy to test. You can get monthly interest rates going back to 1853 and total return stock market prices going back to 1871 very easily. But if you want daily or minutely prices you are going to be out of luck.

Almost without exception participants on this forum are fooling themselves and if they trade based on their blinkered approach will eventually pay the consequences.

@ Anthony,

I was hoping for a data source on new bond issuance, not interest rates....

EDIT: I also don't really see any "approach" established on this thread, so I am confused as to how a "blinkered approach" is even part of the discussion. This thread seems to me as if it is more of a macro discussion, attempting to gain insight into the current dichotomy that exists between the empirical relationship between stocks and bonds, and the conflicting behavior we have seen in recent weeks.

Yep, that too....its all there

Forget about Quantopian for this sort of stuff. You don't need that sort of complexity. I started with a simple home made Excel spreadsheet to plan out wha I wanted to do and have now programmed it into Python/ Pandas.

Pre 1900 dates can be a bit of a challenge!

Three Centuries of Macroeconomic Data in the United Kingdom

Probably the biggest source of amusement is looking at stock market monthly prices from 1929 to 1935. Test that out with some of these nutcase systems people produce.

Nassem Taleb's style of writing may not be to everyone's taste but he sure as hell has the correct long view.

Thanks Anthony, but I believe you may have misread my question....

I specified that I was seeking a way to pull data direct into Q (avoiding fetcher). The idea was to be able run an algo that allows me to keep a close eye on future bond issuances without having to manually update a csv file every time FRED updates its data weekly.

Regarding how all this relates to a strategy....I still remain confused.

I remember all those abtruse questions we had to answer as history undergraduates at Oxford back in the 1970s: what is the point in studying history? Those who do not study history are doomed to make the mistakes of the past over and over again. Sure, markets change, technology improves but if you want to take the long view you don't want to look through a microscope. You want to stand way back and look through a telescope at the dim and distant past to see the same dull cycles and errors repeating themselves again and again.

Ecclesiastes 1:9
What has been will be again,
what has been done will be done again;
there is nothing new under the sun.

I also don't really see any "approach" established on this thread, so I am confused as to how a "blinkered approach" is even part of the discussion. This thread seems to me as if it is more of a macro discussion, attempting to gain insight into the current dichotomy that exists between the empirical relationship between stocks and bonds, and the conflicting behavior we have seen in recent weeks.

The blinkered approach comment is not directed at this thread or at you personally. It is directed at the entire community of retail and professional traders trying to shoot the lights out and laughably using 2 year back tests. It is directed at the whole Quantopian community. It is not meant as a criticism although it is inevitably that too.

It is just saying that people are fooling themselves by using absurdly short time frames and ridiculously shallow history.

It is like an ant standing at the tip of a blade of grass trying to understand the universe.

Please don't imagine I am claiming any Nostrodamus like powers: I am merely saying relying on the sort of analysis and back testing I have seen so far on Quantopian is really not going to help anyone in the long term.

I know it all sounds so pompous, rude, arrogant and counter productive. It is not meant to be. It is just that the VAST majority of investors and traders are heading for the cliff like lemmings.

Especially those who use leverage.

Frank
Yeah, I know what you were asking re Q. Its just that I don't think Q is the right framework for this sort of deep history stuff. I guess if you want to pull that sort of data into Q going forward they will have to amend the code at their end.

My partner and I took a good look at Zipline a while back. Unfortunately it changes so rapidly and is so complex it was like running in front of an express train.

Which is why we took the view we would develop our own software for our own simple purposes.

In a way Q is an uncomfortable mix - it almost seems to be aiming at HFT and yet of course will never be able to compete at that level.

And I must not forget I speak from the perspective of a long term investor. With, frankly, no earthly need for tick data or the vast majority of the bells and whistles attached to Q.

The longer I study the markets the more the quantitative approach amuses me. It is a godsend at one level - thank god for the index trackers. Down with the stockpickers.

At the other end people still imagine that probabilistic quantitative methods and AI can predict the future and reap untold riches. It very much reminds me of the alchemists of old.

But let's not forget we now realise the alchemists ambition is theoretically achievable as we now realise.

Maybe the universe is deterministic. Maybe therefore inputting the right variables will one day be able to predict the markets. It is right to carry on the research. But until and unless that day comes people are going to waste and awful lot of their hard earned cash in search of Eldorado.

Read Voltaire. Candide found the right answer.

I agree...All studies conducted using the dataset(s) of Quantopian should always be considered within the context of a longer horizon . Does your empirical view of the data provide any insights into the current high correlation between stocks and bonds, as well as the negative interest rates currently established in a significant amount of western economies?

Sadly, no. My apparent arrogance merely hides the fact that I have more or less given up and admitted that despite working in the markets since 1987 I know no more than the Quantopian neophyte just leaving college.

How about a personal opinion on the likely effect of negative interest rates on long term economic growth?

Beats me Frank. I simply lay my investment bread on the water and hope.

Are you Merrill Lynch by the way?

...not Merrill

I am always looking for participants on the Quantopian platform that have an investment thesis, or some type of approach using older data that requires further study. Other than the stock market data from 1929-1935, is there any other historical data that you currently find interesting in the context of the markets today?

I'll have a think. I am wholly momentum based I fear, hence do not much care for causation and correlation since I am not at all too sure there are any reliable and durable relationships out there on which to base an investment strategy. Hence partly my cynicism for AI.

Simon Thornington

http://imgur.com/a/Nw3UB

Alan Coppola

@Frank,
In reference to your technical question above:

I specified that I was seeking a way to pull data direct into Q
(avoiding fetcher). The idea was to be able run an algo that allows me
to keep a close eye on future bond issuances without having to
manually update a csv file every time FRED updates its data weekly.

I believe the answer is no. The only external links are fetcher and data coming in from approved vendors.
Confirming links and potential work-arounds:
https://www.quantopian.com/posts/accessing-external-data
https://www.quantopian.com/posts/can-the-fetcher-be-used-for-minute-level-data
https://www.quantopian.com/posts/external-data-api-and-files-at-quantopian

alan

Thanks Alan. I'm actually going to start work on a python script that pulls FRED data and appends to a google sheet that is pulled into fetcher. Not excited about work required, but hopefully I can find some shortcuts in github or elsewhere. I don't feel as if I will ever be comfortable with an algo until it is monitoring macro trends. I also am very eager to start dissecting the the Fed balance sheet. Excess bank reserves, declining velocity of money, and all these other oddities in the monetary system being caused by QE are very interesting. I once took a course on the Money and Banking, and a more sophisticated model for the money multiplier was developed and used to interpret the last liquidity crisis. I am starting to think that the money multiplier as a result of QE should be studied in a segregated manner. Old Economy multiplier and New Economy multiplier. Basically try to build a counterfactual as if QE had never occurred, and then compare that against reality.

Alan Coppola

https://research.stlouisfed.org/docs/api/fred/

@Frank, sounds like a good cause.
Looks like they have an API:

and some third party clients in python on GitHub...I've never used any of them though.

Python
◾https://github.com/avelkoski/FRB ◾ https://github.com/mortada/fredapi ◾ https://github.com/zachwill/fred

Good hunting!
alan