Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
What do you guys think about concurrently running multiple automated strategies that have a very low or negative correlation?
If you measure the performance of each strategy individually, and pay no attention to the impact it has with/against the other strategies, but instead analyze the trades from a correlating standpoint (MultiCharts is good for this), I think there can be a lot of value added here.
If you have a basket of say five strategies the idea is you trade a few different markets (doesn't have to be 5, you can have more than one strategy per market for hedging) and each strategy has its own unique signals that are unlikely to signal the same entry type (long or short) on the same type of stock (ie: index, oil, currency), then I think the results can be far better than trying to selectively choose the perfect strategy for the perfect market and limit yourself to just one strategy per market, or even just one strategy in general.
I'm not very good at these things, it is a new concept for me, so I can only speak in general terms that have some common sense value. Anyone have specific experience trading non-correlating strategies and can offer up some pointers?
I agree 100% and it is what I have been working on... a portfolio of strategies diversified over:
1. Trade time horizon (intraday, daily, weekly, monthly)
2. Trade type (momentum, mean reversion, market microstructure)
3. Asset type (equities, interest rates, commodities, currencies)
What will hurt when I go live with the idea is the capital requirements. With a small account you can "play" with the ES, CL, GC, ZN, or whatever rocks your boat. But when you look at holding (and rolling as required) futures contracts on a weekly or monthly basis the capital requirements go up exponentially (at least by my risk model). For that reason my weekly/monthly strategies are executed with ETFs (unleveraged) and not futures contracts right now. Pretty boring, actually. Throw a bit of Kelly in there I'll confide I'm 100% in weekly/monthly right now.... as none of my intraday or daily strats has a positive expectancy at the moment. So the Kelly bet is 0 for me now intraday.
Google "mean variance portfolio optimization". Markowitz is a name you will come across.
I'll dig through my PDF library and see what I have. With R and/or Matlab you can do some portfolio-level optimizations of the modeled backtest results of individual strategies. That would give you a theoretical weighting of the strategies in the portfolio from a risk/reward basis.
FWIW one thing I'm not that happy with in NT7 is that while I can backest a basket of strategies, you can't get a basket-level equity curve/Sharpe ratio/max drawdown etc etc. That would simplify my life considerably.
I'm not good in this neither, but a good (the best in C2 imvho) automated strategy hosted by C2 ( VT26 - a futures trading system on Collective2 ) is using this idea: the same strategy with different parameters (I think...), and 6 different uncorrelated contracts.
Wow great stuff in there... I will have to read this several times, but at first glance I am really liking the risk of ruin (ROR) separated into # of operations, and the monte carlo sim on the entire portfolio.
He also makes an interesting argument about using a daily penalty to account for slippage and commission, instead of a per-trade penalty. His argument is the latter produces an statistically incorrect equity curve, but I am not following his reasoning on that.
No secret sauce in there... but the $100K suggested miminum just for this 7 instrument strategy is what I was talking about. A portfolio of say, five strategies, starts to add up!
So your saying that using Kelly you can't find a mathematically sound risk/reward for intraday trading, which is why you need larger time frames and therefore more capital due to the increased volatility of a multi-day trade?
I find that interesting. Wouldn't it mean that, in general, there is not enough diversification occurring on a daily basis to be profitable, and that you need multiple days in order for the markets to have moved far enough apart from each other for a profitable opportunity to occur? That in itself doesn't sound right to me, just looking at say CL vs ES vs 6E and their daily ATR and moves.
I'll read your PDF next MX. Hopefully there is minimal math
After you have some setups and you are disciplined, I'm not so I don't trade discretionary, what is the MOST important issue to be profitable?
I will answer this question after I get answers to another question.
Can it be profitable to trade a …
.
I was even more bold and said that if you trade several setups, you will be better off even with a loosing setup. I wanted to prove this point, but got some stupid posts so I left it.
Thats why I'm against scaling. In the best scenario (both PT give positive outcome on their own) they have very strong correlation.