This is my attempt to come up with a mechanical trading system for the mini-Russell. I would like to get some opinions if you think this will work going forward. I backtested ~2 yrs. of data then optimized. Obviously I don't want to reveal the details but want to know if my thought process is valid for those that have had similar attempts to build a trading system? If I trade it "live" - what size account would you use? Other thoughts/ideas? Thanks! See 12 page .PDF
Derfermark did you take into account the slippage per trade? i notice you only included commission of $6 RT but not slippage..
Yes and No - I took it into account but didn't include any slippage for a couple reasons. 1) All entries are on Stop Limit orders and I'm only trading 1 contract (initially) so I will get filled at my price 98% of time (from experience). 2) all exits are with Limit orders and my backtesting assumed that if price did not go at least 1 tick THRU my limit price, I did not get filled. I should tell you some raw numbers too, during the ~2 yr. period, 376 trades were made, 235/141 Winners/Losers = 62.5%, Avg. Win $236, Avg. Loss -$257
(1 + 236/257) * 62.5% - 1 = 0.1989 This is your system's mathematical outcome, or expectation. Not strong, but for ER2 maybe not very weak either. I have no average number to offer here.
Well, considering the small drawdown shown on the hypothetical system, you don't have much to lose. On the other hand, with an average profit barely above $50/trade on the *optimized* version, I don't think the results are very impressive. It really depends on your current capital and time constraints as to whether this is worth pursuing.
Thanks risktaker - I agree results aren't that impressive but I've been trading for several years and am convinced that "slow and steady" is the way to go. If my system works somewhat like the backtested period, I could double my account every year... (famous last words - lol) What good intraday systems have you seen that have impressed you? What were DDs, win ratio, etc.?
Will you consider testing with older data to see how negatively (or positively) the optimization effect the overall results? This looks like a good system based on the equity curve shown.. anyhow, wish you the best trading this system
I hesitate to throw a cold towel on the party. Displaying the equity curve is fine. Congratulations. As to whether the system will "work" let me point out that the conditions that you found (that you coded for entry and exit) may reflect a statistical edge, OR they may reflect a seasonal tendency (what I call a historical tendency). If it is the first, I suggest you hire an accountant and lawyer ASAP, so you will be prepared to manage your profits. If it is the second, well, no one can tell you how long the party will last. I do wish you the best of luck, as it is not easy to find a viable edge. Steve
Hi DerferMark, I did a lot of systemdevelopement the last months and here are my 2 cents... First, without declaring your rules, it would be of interest on what timeframe (chart) and if trend or contratrend action your system works, an equitycurcve alone is to less to give an answer... I guess you did choose the ER, cause he did the best intraday trends of all popular US indices, at the moment... Maybe this most trendy behavior can/will change, and goes sometimes more to the noisy site of the ES. So I would test your system also on the ES, your system should at least trade at breakeven on the ES and/or the NQ, if you will get a negative equity with all vars. you should expect some trouble or at least a much more deeper drawdown soon... From your description I think you work with a timeframe little bit longer than 5 Min. maybe 10-15 Min, 30 Min. seems to long... Next you go with the trend, cause you did choose the ER... Next you work with Indicators, may be simply with MA`s... I found all Indicator systems fall apart soon, cause the nature of those instruments. If market behavior changes, the smoothing factor of the indicators make them fail, there should at least some adaption(of the parameters) to marketaction like volatility, velocity of moves, cyclelength etc. i did not found a good solution here... Some more thougts about indicators of me: Some example: Let`s take any 1 hour of trading, if you use a longer moving average(50 bars or something...) you end up with the same readings at the end of the hour if the market did: a smooth steady increase a very sharp spike upside, followed by decline a stronger smooth increase with some sharper decline the last 1/4 strong increase with medium decline in the middle and further little increase etc, etc.... Indicators does not describe marketaction !!! With always the same reading in the MA, the marketaction was quite different and the consequences are quite different. A better way would be to describe the market in timewindows of 30-60 min. and describe in some other way the pure priceaction in every timewindow to find some statistical edge. Indicators are like a shotgun, but we need a laserpointer precision gun to meet our goals... Uhh i`m now to far from your question... just my 2 cents... Michael