That's a valid question. Posted at 2am after a long day of trading and night of busting my butt at work. This is the chart I meant to post. JJ
again without explaination..... btw many people, including me do not open xls attachments due to virus risks. a jpg screenshot would be nice.
One thing I may want to consider.. is days like this if by 9:30 central it done moved this far.. that perhaps I should tell my bot to not initiate any trades.. since it seems like right now is normally when it will end up getting into a trade and getting itself into a bind.... but no trades have been taken yet.. will just wait and see for now
shrug ... the xls is the explanation. But i agree with you about not opening foreign source spreadsheets, so I'll see if i can't get a screen shot ... just for you walter. JJ
The gif won't load (too big), and while I'm pretty decent at graphics, I don't have any graphics programs on here, so: All Trades Day $ Amount 1 $1,450 2 $40 3 ($344) 4 $95 5 ($35) 6 ($349) 7 ($229) 8 ($263) 9 $170 10 $1 11 ($768) Sum Total ($232) Wins 5 Losses 6 Max Win $1,450 Max Loss ($768) Avg Win $351 Avg Loss ($331) Consecutive Win 2 Consecutive Loss 4 Man, that pasted almost perfectly!
Nah, that actually is a great question. Because within that question lies the answer to really good trading and system design, automated or not. There is no righttime to be designing a system, or trading the markets. The markets will always deliver profits to the well informed and skillfull trader, if they can automated their system to some degree, so be it. Regards, JJ
I'll explain my thinking, since I tend to infer based on the results, I think other traders do to, but that is not necessarily the case. The system had a good day which did in fact turn out to be an outlier (extreme value in one direction away from the mean average). After that the system did not trade well in either trend days or what are called chop days, with isolated days of small-to-medium profit followed by multiple days of mid-sized consecutive losses, followed by yesterday's blow-out losses. Regards, hope the information is taken with the best intention, because that's how it is presented. As bolter mentioned, many traders have gone through this process (I had six months of losses, followed by two years of system design, redesign and retool). Regards, JJ edit: Notice the Risk:Reward ratio is now pretty close to 1:1 ... not good.
Frosty, you might want to do the following test to see how well your simulation really approximates real-money trading: Run your live system on real money and run the exact same system on your paper account for exactly the same period. You might be surprised to find an unexpected difference between the "theoretical" results on the paper account and the empirical results on the real-money account. This would most likely be a hidden frictional cost that you could eventually measure and then add in to your commission/slippage structure but could also be due to some trades filling or not filling as expected. HANG IN THERE--set maximum limits by which you'll decide that your system is wrong or broken, and then give it free rein to run. Meddling or second-guessing changes your system's results--it's like the Heisenberg Uncertainty Principle applied to automated trading systems.
Agree, looking for differences between your backtest and live trading results is crucial. But, if this uncovers nothing, I can honestly say that you cant determine a damn thing from just 11 trades. Thats just noise.
I can say with a great deal of certainty that my live results and backtested are identical... the whole time the application is running its recording every tick..... I then use this for backtesting after.. every one of these days it has been live I ran the backtest portion on it at the end of teh day...and the results are the same.... so I am confident that there are no errors there.