I was wondering how people model slippage for back testing and paper trading? It occurred to me that it could be modeled as a random picks from distribution around the entry price, another simpler method would be to assume that it always you always get a certain amount of negative slip. Can anybody offer any advise? Thanks in advance.
At Ed Seykota's website, he uses a very brutal slippage calculation. If you get a buy signal, your entry price is half-way between the open and the high of the following bar. If you get a sell signal, your entry price is half-way between the open and the low of the following bar.
I use two methods: 1) Suppose my system trades at the opening price of the next interval. I have a slippage parameter, say it is 0.5, my system calculates the difference between the high price of the interval and the opening price of the interval, multiplies by 0.5 and adds the opening price. So if the opening price is 20 and the high price is 22 the difference is 22-20 = 2. 2 x 0.5 = 1. Add opening price 20 + 1 = 21. I use 21 as my transaction price. This is Ed Seykota's way. 2) I just add or subtract a percentage from the historic data price. This method can show transactions at prices that are not within the high and low prices of the interval. I trade in real time, track slippage in real time and compare observed slippage with simulation results. I find for long term trend following of stocks about 1 % slippage is a little more than I actually experience. My simulations show for a buy low sell high system using limit orders slippage might work in my favor. I don't trade this way though.