1) In my case I am starting with ES and 1 car for bare testing purposes. I thought about scaling strategies to perhaps test on an 'avg' fill price but as I said I am not familiar with future specific strategies so would like to hear suggestions.
I am sure you will find some good advice here : How to Benefit from Commodity Price Volatility - Commodities Investing Seminar - April 21-22, London http://www.edhec-risk.com/AIeducati...nt Seminars/Miffre_Till_Commodities_Investing
Sometimes I think people get caught up in theory too much. Trade your system with one contract and see how it compares to the simulated results. It's the easiest and most effective way to find out what you want to know. If you're doing a 100 trades in a session and your system has a positive expection, the losses on any given day should be marginal. If the system tests well with one contract, add a couple more and see if it still holds up. Runningbear
RunningBear, I expect your right, it just seemed to me that futures traders would have a general rule of thumb or technique to account for these trades? Here is example of the differences, 1) Top/Bottom Tick Removed 2) All Trades
Not so bad since the EC with top/bottom removed is almost flat( if commissions are taken into account of course).
Yes, 2.40/car was included. My guess I could assume the actual P&L is somewhere in between these two curves. However now I have no resolution on drawdowns and runs. I guess I can include some random number generator and perform some monte carlo analysis but perhaps RunningBear had it best, just trade 1 car and see what happens. Anyone want to make some money?
I have struggled with this issue for a few years myself. The only way I was able to create a very conservative backtest with limit order fills is to ensure that the bar that fills the order actual punches through the order level by at least one tick. I really think TS should include this option for limit orders in the strategy options. In reality, a given percentage of limit orders will fill when touched but not pierced. However, the assumption that 100% of all orders are filled by simply touching is entirely faulty. My only solution was to build my own quick and dirty backtest platform in C++, with TS continuous contract data exported to text file. I included the fill if touch / fill if pierced option, and my e-mini day-trading results were substantially different between the two options. rt
RT, Luckily WealthLab offers a function for futures to ensure a trade through. As for TS, there is an ADE indicator that a user developed to 'peek' ahead one bar. Its a little clunky and I'm not 100% confident in the results but its better then nothing. As such, have you developed High Frequency MR strats that perform well with 1 tick trade-throughs?? The idea of selling strength and buying weakness seems robust to me but it simply doesn't test well when excluded top/bottom ticks.
knocks, what about trying your strategy on YM or something that has mostly similar behavior to ES but where the tick size is smaller, so that the removal of top/bottom ticks is not as restrictive.