What do you know about if I can backtest or not? Before YOU open your mouth about that - I made a profession out of that and have forgotten more than that you'll ever know. Now, if you want to know why backtesting is useless then you have obviously not even realised the automated order flow analysis systems will completely throw havoc with your results. And then I am not even talking about the autospreaders and spoofers / flippers and the games that the hedgefunds play with the spoofers / flippers. The markets are an interactive system that morphs the moment you put a trade of decent size on. Or you think the market is not affected when GS puts an order of 10000 cars in the Bunds? Keep on dreaming with your one lot trades! I run circles around guys like yourself but you'll never know. Holmes (and this time definitely over and out)
I think there needs to be a distinction made between backtesting and being a 100% pure mechanical trader... First and foremost.Everyone, in every profession performs some sort of backtest if they have a modicum of intelligence.To say backtesting is unneccesary or useless is naive at a minimum,and pretty dam close to deusional.. I cant think of a professional sports team that doesnt employ a statistician is some manner to "scout" the opponent... Eyeballing charts is a crude form of backtesting.. Trying to discern patterns or trading off candlesticks should involve a form of backtesting.... I am not even sure as to what the debate is about
Curtis, this describes my experience exactly. I guess it's because I'm a very analytical person and the computer lets you try a large number of variations in a short period of time, which I became fascinated with. The problem I ran into was that after optimizing several variables, I lost a bunch of money paper trading. So I took a look at the trades the optimized system executed and they were dumb, but lucky, if you know what I mean. I would triple my money in six months, but only if the conditions were repeated exactly as in the past, which is pretty unlikely. As a result, this weekend I went back to the drawing board, threw out the multiple indicators and started using a very simple system that seems to be working so far paper trading. It back tests ok, but more importantly I understand it and I can see relatively soon when it's not working. This is another thing Van Tharpe suggested--keep it simple and don't use multiple indicators that just present the same data in different ways. Chuck
When it comes to short-term trading backtesting has value but I dont recommend going too far back. Once I anticipate an edge then I forward test for at least 6 months to check the validity. Worse case scenarios (slippage/commisions) are also priced into the tests. I am also a firm believer in hand testing. It really beefs up your chart reading skills. People have been trading the same patterns for years (Dips, Breakouts, Narrow Range Bars, etc....) The key is knowing when and where to find the action and trading multiple vehicles as well. This is what Traderdragon is referring to when he mentions proper backtesting. Maybe he does it a little differently then I do but whatever it is we do seperates us from the backtesting masses. Backtesting is an art and takes years to learn. One thing I stay away from is backtesting Long Term Trend following systems since it takes years to forward test them. These systems are usually pretty simple since all you are using is technical data (MA's, RSI, MACD, ADX, Charts, Candles,etc...) and probably explains why most fail and have unbearable drawdowns. Most of the backtesting public want to sit on their hands and make money but unfortunately there is no free ride in the markets. A good tester/trader can make the same returns (or more) in the short-term as someone trading longer term systems but has to really put the time in and will also benefit from a very low drawdown. He also has the luxury of knowing early when to pull the plug and has a back-up system/plan ready to go. This is what they call shifting your style. TraderDragon All I can say to you is that most of these guys will never get it. Dont bother trying to explain it to them. No offense to anyone here. Different strokes for everyone I guess. I calculate risk and testing is the best way for me to do this. It has made me a better trader and also keeps my stress levels down
Not backtesting is silly. Remember, even something as basic as "buy and hold" is based on a backtest. Fletch
I think you folks have to understand. Backtesting is problematic, because the majority of ET members lack the skills to properly execute tests, and evaluate the result. In the hands of an incompetent amateur, woodworking tools will produce a piece of wooden crap. In the hands of a skilled cabinet maker, the result is a piece of fine art. I could say more, but it is about that simple. Steve
Everyone keeps going on about backtesting but no-one has mentioned the benefit of forward testing. Forward testing to me is the most important element in developing a profitable strategy. At any one time, I am forward testing at least 7 or 8 strategies. If a system is profitable in real time with market entries, then I will look at refining it further through methods such as integrating multiple parcel entries back from the current bid/ask in an attempt to get a better average entry price on the total position. Obviously, the longer time frame strategies are better for this but I'm mostly looking to take the major daily swings. Runningbear
Like it or not Back testing and Forward testing are no different and both activities fall under the umbrella of so called historical data fitting. My advice to you is, be very very careful with the machinery, it may eat into your pockets rather quickly, much quicker then you can think!!!
Unfortunately for some amateurs computer backtesting is all they think they have to get a handle on the market. They believe in it regardless. The point is that basic OHLC data is skeletal and primitive as an input. But that is what is basically being relied on. Incidentally few or none are suggesting that historical market data is not important. Even if you only look at back charts that is what, underlyingly, you are looking at.