Systematic Performance Stats
- Thread starter gjmason
- Start date
I was fine tuning my emini trading system last night and programmed it into Wealth-Lab to get feedback. I was very surprised to find with a few minor changes my system went from nicely profitable to very profitable. I'm using 15 minute bars, but any length of time works, just 15 seems optimum. I haven't figured out how to get in futures mode, so I loaded the INX history from 12/06/2001 to today, using a constant $100,000 investment to simulate the value of two ES contracts. I put $150,000 into the account to get it off the ground in case of drawdowns.
Here is a screen shot of the performance report from 12/06/01 to 06/26/2002. The system uses a trailing stop system and goes both directions. Sorry, I'm not going to disclose the rules but thought it might be interesting to those who think mechanical trading systems don't have potential.
Here is a screen shot of the performance report from 12/06/01 to 06/26/2002. The system uses a trailing stop system and goes both directions. Sorry, I'm not going to disclose the rules but thought it might be interesting to those who think mechanical trading systems don't have potential.
Hi. Those are impressive results. Took a while for me to fully understand the stats as I don't utilize Wealth-lab.
2 ?'s ... did you account for commission/slippage (if so, how much).
FWIW, I would also run the data over the actual contract vs. using the cash index. I have found disparity in backtesting systems on the SP cash vs. futures.
If you'd like, drop me a PM and I'll email you the continuous emini contract for the past year (let me know what time interval).
Good luck!
Rich
2 ?'s ... did you account for commission/slippage (if so, how much).
FWIW, I would also run the data over the actual contract vs. using the cash index. I have found disparity in backtesting systems on the SP cash vs. futures.
If you'd like, drop me a PM and I'll email you the continuous emini contract for the past year (let me know what time interval).
Good luck!
Rich
These are impressive charts as well as a pretty capable looking tester! I've looked at the demos of NeoTicker, AmiBroker, and Wealth Lab to make comparisons. (I'm using TS2000i for trading and testing currently).
If you had the time, could you create a thread under Software and describe your experiences with WLD2? I, for one, would be very interested in hearing about them as I am considering that software package.
I agree with rcreal's suggestion that you do your testing on ES, or at least SPY. There are many systems that perform well on an index but poorly on the real instrument. *Particularly* a system that uses stops to enter or exit.
I'm not that sure about the system's position sizing, but the average trade seems pretty low. If the system is trading $100k worth of the index and getting an average trade of $100, you may be looking at results that you could not achieve in real trading. The high number of trades adds to that problem. The system trades 15-20 times per day on a 15m chart. One of the best ways to get a supersmooth equity curve is to have a large number of trades with small profits per trade. But those results rarely translate to your account when live trading begins. I've got the account statements to prove it :eek:
You're on a good track though. Just be sure to test with ES and have a healthy dose of slippage/commish tossed in.
-Jeff
I'm not that sure about the system's position sizing, but the average trade seems pretty low. If the system is trading $100k worth of the index and getting an average trade of $100, you may be looking at results that you could not achieve in real trading. The high number of trades adds to that problem. The system trades 15-20 times per day on a 15m chart. One of the best ways to get a supersmooth equity curve is to have a large number of trades with small profits per trade. But those results rarely translate to your account when live trading begins. I've got the account statements to prove it :eek:
You're on a good track though. Just be sure to test with ES and have a healthy dose of slippage/commish tossed in.
-Jeff
Hi Dave, Wealth-Lab does keep getting better and better. Unfortunately I'm not one who uses even 1% of its potential, so I'm the wrong one to discuss anything, other than to say it's an awesome program. I'm sure you've looked at their website, but http://www.wealth-lab.com/cgi-bin/WealthLab.DLL/getpage?page=Articles.htm has some good examples of what you can do.
It does have realtime capability now, but haven't worked with it yet.
Rcreal and Jeffm I agree with your points on the small average gain per trade, and using the actual contracts. I may take you up on the continous contract Rcreal, I'll let you know.
I've refined the commission and ticks I give up to more accurately reflect the true costs. I'll post those results later tonight. From experience, my costs per contract are $12.50 in + $12.50 out + $4.80 round turn or $29.80 per trade per contract. I've experienced no slippage the last month on under 4 contracts using stops (99% of time get my stop price).
The system has many scratch trades for small gains and losses, thus the high # of trades, with a few large point winners that make the money. Also unaccounted for are 1) not following the system 2) entry errors 3) bad ticks on charts 4) Globex/broker downtime 5) stupidity.
My goal is to get a consistent approach and try to make it more profitable over time. The backtesting helps me psychologically weather the down periods with no profits, and stick with the methodology.
I'll post more screen shots tonight, at least to give people a look at Wealth-Lab's different reports.
I've refined the commission and ticks I give up to more accurately reflect the true costs. I'll post those results later tonight. From experience, my costs per contract are $12.50 in + $12.50 out + $4.80 round turn or $29.80 per trade per contract. I've experienced no slippage the last month on under 4 contracts using stops (99% of time get my stop price).
The system has many scratch trades for small gains and losses, thus the high # of trades, with a few large point winners that make the money. Also unaccounted for are 1) not following the system 2) entry errors 3) bad ticks on charts 4) Globex/broker downtime 5) stupidity.
My goal is to get a consistent approach and try to make it more profitable over time. The backtesting helps me psychologically weather the down periods with no profits, and stick with the methodology.
I'll post more screen shots tonight, at least to give people a look at Wealth-Lab's different reports.