Switching between systems using equity curve feedback

Sergio77 said:
Yes, life must be easy. If I need to do everything in Python the game is lost. A trader must concentrate on trading and not on programming.
More...
As I get older, I prefer to concentrate more on not trading and more on automating, just tired of sitting and watching little marks on a screen, make enough bucks and let computers find the rest.
abattia said:
I once looked at a similar mechanical concept, and found the following issues (similar to those of a lagging indicator!):
1) The time lag between start of drawdown and the switching off of the underperforming system; you watch lots of equity destroyed before the underperforming system gets switched off. Conversely, you miss many wining trades before an off system switches back on.
2) “Switching Chop”; just after the underperformer is switched off, performance improves again. After a few successful trades, the off system is switched back on, and the next trades are all losers. In the end, you participate in all the losers and miss all the winners! It happened to me!
In my view, instead of the above it is best to pick systems with win rates and drawdowns you can live with. And then just take all the set-ups. In my experience, you come out ahead.
More...
I have developed some 35 systems and have eight of them automated so far but on first weekend of each month, I look at equity curves of all systems and I seek a bit different on those I am not following/trading, I am looking for the worst of them in equity curves, I have designed what average mean drawdowns expected and when this point is reached, volume is added near expected drawdown lows, so many more percentage points lower more contracts added, BUT never more than 8% of overall original position. For me, this has worked out and tested better than finding which are strongest as often times those will go other direction. Matter of fact, I backtested mean for making new equity highs and when these areas are made, volume starts cutting back at increments of 10%. One huge reason people lose so much money at highs as they trading largest amount of shares or contracts.
 
Handle123 said:
As I get older, I prefer to concentrate more on not trading and more on automating, just tired of sitting and watching little marks on a screen, make enough bucks and let computers find the rest.

I have developed some 35 systems and have eight of them automated so far but on first weekend of each month, I look at equity curves of all systems and I seek a bit different on those I am not following/trading, I am looking for the worst of them in equity curves, I have designed what average mean drawdowns expected and when this point is reached, volume is added near expected drawdown lows, so many more percentage points lower more contracts added, BUT never more than 8% of overall original position. For me, this has worked out and tested better than finding which are strongest as often times those will go other direction. Matter of fact, I backtested mean for making new equity highs and when these areas are made, volume starts cutting back at increments of 10%. One huge reason people lose so much money at highs as they trading largest amount of shares or contracts.
More...

You can test this concept in TradersStudio
 
Ok, but TradersStudio can be extended with COM DLL's as well as our scripts, so it would be cheaper to use it to test somethings you would need to write from scratch.
 
Notes on Equity Curve Feedback
1) Classic moving average crossover of equity requires trades to have dependencies.
2) If trades to not have dependencies you can use a partial martingale with a 4 sigma turn off.
 
You must log in or register to reply here.
Back
Top