$40 Slippage sending good systems to the trash can!
$40 Slippage sending good systems to the trash can!
Can anyone throw some light onto a problem i am having with slippage settings?.....Im testing a medium term trend following system, and it produces good results. It even tests well when i introduce $10 commission and 50% slippage and 50% slippage on contract roll. HOWEVER, as soon as i add a fixed dollar slippage figure of over $40 it goes completely down hill. The performance becomes dire!. So i leave the $40 fixed dollar slippage in and set slippage% and slippage on roll set to zero and test again......still the same scenario. So then im thinking the average trade must be too low, but it doesnt make sense. So i tested a long term turtle model of 100 in 30 out on a portfolio i knew it does well on, and the same thing happens again. It tests well even at slippage% set to 50% and contract roll slippage% set to 50%, but as soon as i set those two figures to zero and the fixed dollar slippage to $40 or above, it goes down the drain. Can anyone throw some light on this please. I know $40 slippage is a low amount to be testing with, i know $10 commision is high incomparison to electronic trading costs, i know 50% slippage on entry, roll and exit is way to high and it should be around 10%, i know about different results on different portfolios. Ive tested this deliberately on systems that have a high average winning trade, and do particularly well on curve fit portfolios to find the root of the problem.......but im hitting a wall. Thanks....
Perhaps you and TBB have different opinions about how the slippage settings will operate in a simulation. One way to study this is to calculate how many contracts were actually traded in a simulation run, then read the Total Slippage number from the simulation output, then divide the two numbers. Presto: slippage dollars per contract. Maybe this old post might give you some ideas: viewtopic.php?p=16092&highlight=slippage#16092
- Attachments
-
- perfor.png (24.33 KiB) Viewed 5768 times
Thanks a lot sluggo.....you taught me something valuable there. You seem the type of person that knows TB inside out, but i`ll tell you my findings anyway. I kinda went around it in an odd fashion but it taught me what was going on. If i set all slippage settings in global parameters to 0 AND set account for contract rolls to true, and then set commissions at $75 per contract, then the results are good! That is the equivalent of saying $75 for slippage and commissions per contract on a trade to trade basis. However.....if i set commissions to 0, and set minimum dollar slippage to $75 then cometh the trash can!!!....TB must be applying $75 slippage to entry, exit and roll, and the exit of the rolled contract. Ergo the real slippage figure on a trade that doesnt roll when using the minimum dollar slippage at a setting of $75 is $150, and the slippage on a contract that rolls is $300. Would you agree? And would you also agree that this is very excessive.....
And thanks again......
And thanks again......
Further to my last post which i forget to mention was that using the calculation you provided in your post, that for me to get figures of around the $75 slippage mark, using just the slippage% and roll slippage(atr), and no minimum dollar slippage requires a figure of around 50%!......i remember reading a post by c.f. saying that realistically he expects around 8%.......im confused! When i set slippage% to 10% AND account for contract rolls to true AND set roll slippage to 10%, and then applying the calculation you gave above, i get a figure around the $27-29 mark!.....any thoughts on that one?
Ok now im totally baffled!!!.....as im trying to problem solve with these problems im encountering, it appears my account for contract roll setting does not work. Im running 2.0.11 beta. I get the same results for when its set to true or false. Ive now run it on an ultra long term turtle system to ensure i am actually running into contract rolls.....has anyone else encountered this problem?