Don't have a ClariFI can't afford it. But I do have GAT's fabulous pysystemtrade. Modified it and ran it on equities. Don't look great. That's why I'm asking if any one experienced would like to share. My one guess is that SPY doesn't do good with trend following and a lot of individual stocks will be highly correlated to it.
I use momentum on SPY, according to Gary Antonacci's system here. http://www.optimalmomentum.com/gem_trackrecord.html What do you guys think of this system?
Don't have much experience with it, but found this on GAT's site, there are also comments and implementation of this system on quantopian. Hi Rob, have you heard of the book "Dual Momentum"? It proposes a simple system which you rebalance once a month, and the 40 year backtest is outstanding, with an avg annual return of 17.43%, standard deviation of 12.64%, sharpe of .87, and max drawdown of -23%. The system is: if the last 12 months of returns on the SP500 are greater than TBILLs, then put 100% of your account in whichever has had the higher 12 month return: SP500 or an all-world-excluding-US ETF. If not, then put 100% of your account in an Aggregate Bond fund ETF. What are your thoughts on this as a 5-minute-per-month simple system? Rob Carver24 July 2017 at 08:40 It isn't a bad idea (any system is better than none, and it's simple enough that it's unlikely to be overfitted). But I don't like the idea of putting 100% into any asset because it shows a confidence in your forecasts you are unlikely to have, and the trading costs could be considerable. Also investors with enough capital ought to be diversifying beyond this simple two asset portfolio. Coincidentally these kinds of subjects is covered in great length in my new book - out in September! (where I also use 12 month momentum as an example of a simple forecasting signal, but consider diversification, how much it is safe to tweak the strategic allocation, and trading costs)
Over at Alpha Architect they do something similar, but with a larger basket of instruments, thus creating more diversification: https://alphaarchitect.com/2014/12/02/the-robust-asset-allocation-raa-index/ Each month they post on their blog which asset classes should be used. For example: https://alphaarchitect.com/2018/06/01/diy-asset-allocation-weights-june-2018/
A lot of Meb Faber type strategies are successful in backtest primarily, if not entirely, because of 2008. They're really based on a single observation.
I did the keynote at this event and a Q&A session on trading system design Presentation is here. Video here (I start rambling on at 5:06:30, and then again from 8:23:20 for the Q&A). Enjoy GAT
For anyone who is using GAT's awesome pysystemtrade: I don't understand why the sign of the carry forecast is different between Soybean and Livecow. Here is a list of prices for both, and the corresponding carry forecast. If the price of the further contract is higher than the price of the nearer contract, shouldn't the carry forecast be negative? The expectation is that the price of the further contract will tend to move to the price of the nearer contract over time. Can anyone explain why I am getting these forecasts. Are you guys getting similar forecasts for these instruments? Thanks.
Hi, truetype. I agree that Faber-like momentum strategies shine when the market crashes, but look at Gary's performance over the past 40 years. The average outperformance vs the benchmark is a huge 6% per year. It's hard to argue that that is just luck over such a long backtest period.
There was a bug in the carry calculations, did you pull the latest code? This was what I posted earlier: I traced the method, it seems that raw_carry isn't called by the carry rule in futuresconfig.yaml due to the way it is defined carry2 calls only rawdata.daily_annualised_roll() futuresconfig.yaml carry: function: systems.provided.futures_chapter15.rules.carry2 data: - "rawdata.daily_annualised_roll" When I changed the config to data: - "rawdata.raw_carry" the raw carry forecasts look more correct (ie comparable across instruments)