Hi all,
I haven't finished ploughing through all the past threads yet, so excuse me if this question has been answered elsewhere.
When backtesting a system, I usually assume a fixed number or fixed ratio slippage. However, the slippage occured in real trading are fluctuant. Sometimes in a drastically volatile market the slippage can even beyond one's worst expectation.
So, I suppose that adding volatility factor to simulation will bring more emulational effect. That is, increase the amount of slippage in a trending market and decrease it in the trading range. But I worry whether it would bring some side effects in simulation process. Do you think it's a better way or is there other ways to calculate slippage?
Any ideas
Better ways to estimate slippage?
When I am backtesting I always just use the absolute worst price the stock traded at that day. So if I'm buying, I always assume that I'm buying at the high of the day, and selling at the low of the day. That way, when it's real money you're not going to get worse than the backtested results due to underestimating slippage. I took William Eckhardt's comments in one of his interviews literally and take a very hard nosed approach to the testing thing. You should be trying to break your sytem in every way, and then if it is still profitable, you just might have something worthwhile.
Bottom line: It is far cheaper to find out if you are just kidding yourself with your system during the testing phase, than when money is actively traded. I hope this helps.
Bottom line: It is far cheaper to find out if you are just kidding yourself with your system during the testing phase, than when money is actively traded. I hope this helps.