I rabidly disagree with this statement. As far as I saw (I may have missed something though) you didn't account for any slippage or commissions in your testing. Commissions of $5 should be assumed per r/t, and slippage is present in all trading systems, and here's why...... Your entry is to buy at a breakout of the opening range. If the range is 1204.75-01.25 (the 60 min OR in the ES) then you will buy at 05 stop and sell at 01 stop. Anything short of this is not a breakout. Now, you obviously don't need to use stop orders, but here are the two likely fill scenarios: A) You place a buy stop at 05. The market hits 05 and you get a fill at 05.25 ($12.50 slippage). B) The market hits 05 and you decide to be patient and use a limit at 05. However, since this is a breakout system and you want to see immediate positive movement, you do not get a fill; the market shoots to 05.75 in about three seconds. Sure, you can get in at exactly your entry value, but it won't happen all of the time, and the times you don't get your fill because of a runaway market are usually the times this strategy makes its money. Therfore, 2 ticks slippage is the bare minimum in any systems test, and three ticks is more safe; remember, we're not trying to make the most impressive backtest gross profit numbers, we're trying to quantify robustness.
Agreed. People are fooled into thinking there is something special about the high OR low being so often established early in the day, when in fact a random walk would produce the same results. In other words, after price has wandered up a bit, it is unlikely to make it all the way back to its lows, but there's no edge in that fact - it's still just wandering around from the current point forward. (I'm not saying the market is random, just that the data presented does not prove otherwise.)
I do agree that the profit per trade for ES over the last 2 years was not very impressive - but this is using a system developed on a different market and the same system is considerably more profitable on other markets over the period. The point of testing on other markets is to ensure robustness - i.e. that the system doesn't produce a loss. I trade ES almost every day using stop-limit orders to enter the market and I cannot remember the last time I got filled at anything other than my stop price. The limit price is set a couple of points out for security only.
First i would like to thank sidinUK for the wonderful post. Second, if anyone wants to read Gary Smith's book "Live the dream", he/she could see that thursdays were poor days during his systems testing from 1986-1994.. I do not know how it tested from 1994-2002..But i suggest thursdays weren't good . Of course, this is curve fitting IMO, but i do believe that trading sidinUK's system will be a profitable endeavour. If anyone wants to use this system in real time though, i would suggest that he/she double the amount of the drawdown, just in case ) As for slippage in ES- i do not think that you can have slippage of 2-3 ticks on no more than 100 contracts.. If any of you guys can handle more than 100 contracts congratulations to you ))
dbs119, I'm glad you like the article. I didn't know about the book you mention - I'll have to have a look for that. Maybe there is something about Thursdays after all!
sidinuk, Could you speculate with a fundemental reason/thought as to what is unique about Thursdays? Are the big players using this day for repositioing? Also, Have you tested Fading Thursdays? One more thing. Have you tested "Long only" with your system on the ES? Michael B.
Any system using opening or closing prices is by definition using future data and is not viable unless it can absorb multiple tick slippage on each and every such entry/exit.
I have several EOD ATS for ES and NQ. My program checks for signals at 16:14:53 and enters trades at 16:14:55. I always double check to make sure I didn't miss a signal or enter a trade I shouldn't have once the settlement price is posted. so far I average about 1 tic slippage per trade (entry and exit). fan27