Other: visual basic programs in the following applications: MarketScanner (in CSI Data's UA), Excel, Trading Blox Builder. Also use TS2000i to do quick prototype programs of ideas, and to cross-check my visual basic code. -edb
You need to be fluent in some .Net language for you to use it.... It interfaces with FIX or TT or IB and some others, so downloading data is not a problem
I am not trying to wind anyone up here but it occurs to me that backtesting is more of a sales tool for overpriced retail brokerages than a real trading tool. To me markets are far to dynamic to think that you can just write code and watch it make money. I would think that time spent on backtesting would be better spent watching the market in real time and learning how buy and sell factor interact with each other. That said I send a great deal of time looking at old charts trying to pick out patterns that happen over and over again. It is also important to look for reasons why the pattern fails to play out. As an obvious example tomorrows noon to close chart will look a lot different then the average one.
IMO, what you say is not really accurate. Backtesting does the following: 1. Eliminates obviously inferior strategies. 2. Identifies strategies with a high likelihood of statistical success. 3. Provides the ability to auto trade and take the human factor out (good and bad). Especially useful for portfolio type strategies. Where backtesting doesn't work is in situations where the data is inaccurate or unavailable -- can't handle fundamental data for example -- or the backtesting tool doesn't resolve to a sufficient granularity (tick based) to accurately mimic the market and test the strategy. The changing nature of the markets doesn't really play into the situation that much in my experience. There are systems that trade trends and systems that trade ranges. There really aren't other possibilities. If you have a portfolio of systems that handles both types of markets you've pretty much got it covered. Just my $.02
Backtesting does not guarantee a strategy will work ,but a strategy failing a backtesting test guarantee it will fail.
I can program for patterns or technical and price levels on 1,000 stocks for the past 15 years, and run them in about 60 seconds. How long would it take you to accomplish the same thing by manually looking through charts? There's no comparison.
What if you happen to backtest on a crappy system? That's why I wrote my own. __________________ Contributing Poster Elite Trader and Vice President for www.fight-nononsense.com
Hi, Can AIQsystems users please give feedback on strength and weakness of the software. I'm contemplating buying/leasing it. Here are features that I would ike to have: 1) Backtest portfolio and apply variable parameters. 2) Backtest using user defined symbols. say a pair from a basket. 3) Create custom indexes (on broad range of symbols) and apply realtime statistical indicators such as % up/down etc.. Thank you for your attention.
All of the above, pros and cons, may be true. But one additional thing that the sheer 'speed and volume' of backtesting can do is teach 'newer' traders how to focus on their EXITS, as well as their selections. The rapid repetition and ability to change the exits in the tests can show delta in results quickly. FINALLY, many traders begin to get the message that cannot be given, verbally! EXITS rule! imo!