...thanks. I must confess that I am not real big on portfolio management (the math leaves me cold). I prefer instead to use multiple systems over different time frames on the same instrument. Have you used any particular method to determine the optimum portfolio mix, or is it simply empirical? BTW, thanks for the stimulating exchange. So rare here on ET. Typical of a serious thread, this one has attracted no flies. My recent favorite, which achieved all of 2 posts and only 213 views, was http://www.elitetrader.com/vb/showthread.php?s=&threadid=24127 Tells you how many people are really serious about backtesting.
DK, Im wondering what software you used to backtest? I use excel almost exclusively. Using excel enables me to get the information I want and nothing else. One problem I have found with it, and others, is that they do not take in to account the trades that would/could have happened while another trade is on the same security. Example: AMD BUY 1000 Shares Today.. tomorrow another signal is thrown to buy AMD. How I have excel set up it does not take into account the trades that would have been placed if I had not been in that position already. Which can drastically alter the results you are getting...
t0yland, I made the software myself.. the real key to portfolio backtesting is to give all your indications/recommendations a score based on the whatever formula you use. Then you optimize the system with a range of portfolio "slices" ie 1-10 and a minimum score required to enter a trade ie 1-6 or whatever. Your system must be setup so that any trade you're in preempts any other until that portfolio "slice" is free. When all slices are full a position must be closed prior to entering a new trade. When a slice is free then the best scoring rec gets taken. You'll find the settings with the minimum score too low enter too many poor trades and prevent good ones from being entered. The best settings will have an exposure of 85%-95% and a healthy combination of minimum score and portfolio slices. Anyways, do any experienced traders have any input regarding the slippage & commission %s of their backtest returns?
I'm confused by your question concerning commissions. Are you asking how one should take commission costs into consideration when backtesting? If so: your backtest should replicate as closely as possible your actual trading experience. So, as far as commissions are concerned, use the commission structure your brokerage actually charges. Or have I missed something? Richard
Commissions are fixed and perhaps I should have excluded them from my question, I assume 0.1% of my return will go to commissions. What about slippage? I have analyzed many 16s tick files and found that a similar 0.1% eachway slippage is accurate - but I would like to hear from experience as well as my analysis to make my backtests as accurate as possible.
I am working on doing the same. I use slices as well. My system cuts the portfolio into 8ths. It does a grand job at smoothing equity curve. I never thought about using a point system before though. You gave me an idea to try it with different entries. I have a standard exit system I use that will break even with random entries. I could use a score system on different entrees and use the best one that is currently available. Doing such would keep my money in the market and always at the best 'possible' price. Thanks for the idea. On to you question. Its really hard to determine what the average slippage will be. It all really depends on how liquid the equity is and what size you are trading. Even if we knew that none of us know how you backtested your system or what type of system it is. Questions like: What time of the day did you enter the market? What type of market did you enter; trending? Obviously in a fast moving market your going to get more slippage.
t0yland good luck with the scoring.. keep in touch ok? I'm in at the opens and out on the closes of stocks with minimum price*volume of 80 million. Looking to invest $100,000 each time.