At the request of A.D.. testing his system on a portfolio from 19 American futures for 15 of recent years.
Basically, in Wealth-Lab you can do everything the same as in TradingBlox, no worse. Since in TV I code much worse than in WL, then the main code in WL from the TradeStation language was written in ten minutes, and the rest of the time I had to tinker so that after the take profit a new position was not immediately opened — I spent another fifteen minutes on it. Compare trades with TradeStation — it all fits. :)
Well in general, so — a portfolio of 19 futures. Slippage two ticks for entry and two ticks for exit. Commission $2,40 on contract. Risk in every trade 1% from the deposit. Reinvestment test. Test with standard parameters (1.3; 50)Here are the results for each futures separately.:
This is equity for 15 years and drawdown. We see that the system has just come out of a three-year drawdown:
Here you can see that the drawdowns lasted and 300 days, And 500, and the last one is more 700 days and reached 50%.
These are the main indicators and coefficients:
These are the yields by years — there were four unprofitable years:
Let's move on to genetic optimization:
Here optimization results are sorted by Recovery Factor:
Here sorted by Profit Factor:
Here for profit (or by average annual return):
Parameter selection, certainly, the case is complex and controversial — one likes to have a smaller drawdown, another to get more profitable trades, the third so that there were fewer unprofitable years, the fourth so that the total profit is larger and so on…… you can't please everyone. :)
All in all, I would leave the standard parameters and do not trust optimization — yesterday some parameters fit, but this does not mean that tomorrow they will be lucky too. It seems to me that this is a matter of chance. Standard parameters are logical, and this is the main thing. Only 1,3 I would replace with 1,5 to make it even more logical and universal :)
