Nice! What are the effects on max Drawdown and average p/l per trade?
The effects are quite big. But that’s probably because the strategy itself is not very satisfying. Although that is the interesting part I think, quite often we take something that might look good in backtest, but it doesn’t perform very good in live trading. Maybe this could be a way to make a bad or mediocre strategy perform better.
The screenshot shows the original to the right and the other to the left.
The black line is the average (20 periods) of the blue line.
…and that is the bit where you curve fit your simulated trading filter to work with how your back test winning and losing trades were distributed (in both location and size of win or loss) but in the future your trades could be distributed very differently and a 20 period average of equity could just see you hitting more losers than winners and all the big losers and miss all the big winners.
Historical averages of equity curves are just the same as historical averages of price – pick the right period and they look great…. but the future is not the past.
Yes, wise words as always Mr. Vonasi, I completely agree.
Now I haven’t tested anything else than 20 though, it’s a number I often use in order to determine a shorter average curve.
But as I stated before in the topic, I also think this is curve fitting, but I can’t help finding it interesting to check out for myself and that it perhaps can be used in some circumstance in the future.
LinkParticipant
Senior
And the Code to “my sim stat V2”, please?
Sorry, but the code is so badly written it would make all the coding pro’s here bleed through their eyes. The exit is also limited to a normal trailing stop, so it’s not very versatile.
But I will post it as soon as I have developed it further. 🙂
This idea from ‘better systems trader’ podcast sounds interesting.
A moving average on an equity curve. If avg below equity curve, turn off algo and start stimulated trading. Turn back on when moving average above equity curve.
Would coding this be possible on PRT?
Yes that is the basic simulated trading idea and exactly what is shown in this post by Mattias:
Simulated trading
…but as said before it is very easy to curve fit an average to anything – including a historical equity curve. The future trades will be distributed differently to the past trades and you can just as easily find that you have chosen the worst possible period for your average going forward and not the best.
Sorry I got excited with the idea and posted immediately.
I should of checked the previous posts.
Can I request the code for this idea please.
I have no idea how to even start
Have you read through the posts in this topic? There’s lots of information. E.g., Paul wrote an excellent code that you can build on. Good luck!