Finished up backtesting today of a system that I had hoped would look better than the results actually show. Simulated P/L curve attached...Backtesting is somewhat useless many argue since it's all history and doesn't guarantee anything going forward. However I have found that if results are good for a statistically significant period of time in varying market conditions, it helps in trading that system as you have much more confidence and are less likely to deviate from the rules or change things on the fly.
Needless to say, this system doesn't hold much promise. Back to the drawing board.
| RESULTS FOR DAY | |
|---|---|
| Contracts: | 0 |
Posts
I've been in the software development field for 10 years but never got into developing trading systems; I don't know, I just don't trust automated systems, especially for retail traders. They might work for large hedge funds that have low transaction costs and can make 1 tick, and just repeat the process thousands of times, but I think learning how to read price action is the key to discretionary trading for the retail trader, paying retail commissions.
ReplyDeleteBut in the end, no right or wrong way to trade. P&L is all that really matters :)
MBA,
ReplyDeleteJust some helpful advice...reduce your account size, limit the number of contracts you can trade, set a max loss limit, and most of all, keep at it!
:o)
E-Mini, No plans to develop any automated systems. I'd rather mechanically follow a system and have control over any anomalies that arise. But I've never really found one I like. I literally have hundreds of P/L charts of systems I've tested in my files. Maybe I should give up trying to find one I like but I'm too stubborn.
ReplyDeletedt233, The way I've traded the past week my account size IS reduced! :(
MBA - I have read your posts on ET for quite a while & just found the blog. I trade stocks, some options, mostly commodities (both outrights & spreads) and am getting into E-minis Intra-Day more now.
ReplyDeleteI had an idea while out with a group of friends, one of which always makes the wrong decisions and I figured I could do fine in life by reversing everything he did. It sounds strange but....I now incorporate that concept into my trading system testing.
You can even have a profitable system by saying something like 'If the 6 month Avg. Return from Open to Close is > 0, then go long MOO & Close MOC'.
You may be able to do the same thing with your system, based on your equity curve. It looks like there may be some dependence between trades within your trade results.
I would be interested to see what you ended up with if you traded the signal as is if the Avg. P/L of the past 50 trades is Positive & Faded the signal if it is Negative.
Also, your method may only be profitable under certain conditions so your new goal could be methods to identify when those conditions are present and only trade during those times, or at the very least, modify your position sizing to take advantage of those times when the ideal conditions present itself.
Regards,
Eric
Eric, good idea and I have often incorporated fading signals in my backtesting when the P/L slope is clearly down. I've also toyed with trading the equity curve of a system, for example only placing trades when a moving avg. cross shows equity rising. The risk is in over optimizing the system which is a real concern. I'll let you know how things look if I test out your idea. Thanks.
ReplyDelete