Are there pullback systems that backtest well?

Discussion in 'Strategy Building' started by jimcrist, Feb 24, 2004.

  1. jimcrist


    I'm a swing trader that trades breakouts using a system I developed and backtested. I tried to develop and backtest a pullback system, but I was never able to find a system that backtested well. I'm curious as to why this is. Is it 'cuz breakout systems are just better than pullback systems or is it 'cuz I never hit on a good system or what? Does anyone have a pullback system that backtests well? Any insights on this conundrum are appreciated.
  2. lindq


    What parameters were you testing for a pullback system that wasn't giving you good results? My experience has been that a good pullback system is easier to trade and generally more profitable than breakouts for swing trades.
  3. nitro



  4. Perhaps try this book (p.92):

    The Acme F System is based on the work of W.D. Gann and a recent book by Steve Woods called The Precision Profit Float Indicator. The float of a stock is simply the number of shares that are available for trading. By analyzing the supply and demand patterns of a certain stock based on its float, the F System pinpoints breakout points and turning points by combining float turnover with chart patterns such as triple bottoms and pullbacks.

  5. works for me so far .... but you do have to pick your market and your spots very very carefully.

    Pullbacks are harder to trade too because you can't really scale in - there isn't enough space in a pullback - it's time & length constrained. So you have to bet big going in on the first pass - which can result in crippling losses if you on the wrong side facing a runaway market.

    Breakouts are easier to trade in the sense that you can put small bets on and scale in as you go along - trends are longer and take more time. Each loss in the chops is small.

    Finally, you can trade massive size in trends, but not in pullbacks, which is probably why more of the big players seem to trade trends than pullbacks (although i don't know this for certain!)
  6. NO.

    The simulations are excellent on a number of programs and packages.

    When I designed and presided over design of multi million dollar private client trading systems development, we would spend countless hours on complex mathematical smoothing processes to verify not just projections but backtestors. These paled in comparison to sinking someone's boat through trading losses and were simply unacceptable.

    We opted for real world analysis that humans do on an ongoing basis, and paid the cost for hiring those with the most expensive degrees from some of the most expensive snob schools in the country instead.

    eye to eye.....

    that's the answer, sorry.
  7. :confused: :cool:
  8. LOL :D



  9. jimcrist


    Thanks for the great responses. To lindq and nitro, I'd like to verify that the systems backtested well and I'm curious as to what programs they used. To OddTrader, the book Professional Stock Trading by Mark Conway and Aaron Behle looks great except I'm not a TradeStation person. I assume they lay out all the rules for their systems in english? Or do I have to dig 'em outta the code.

    That book led me to Trading Systems that Work, by Stridsman and Trading Systems and Methods by Kaufman. Guess I should get busy...

    To limitdown's assertion that backtesting doesn't work, let's agree to disagree, since my system has been performing like a champ since I started trading it 7/25/02. That's my biggest problem. It's hard to psych up to expend alot of effort researching systems and reading books, since I already have a good system.
  10. nitro


    I use TradeStation, Mathematica, Matlab, and C# for my system testing.

    FWI, I am very close to agreeing with limitdown. It is INCREDIBLY hard to backtest _correctly_. If you use _only_ the tools as they are sold that are on the market to come up with a winning system, I would say that is nearly impossible to do so.

    I use TS to protoptype an idea. But once I think I have something, I go to one of the other programs, eventually landing in C# for the production system, often in conjuction with Excel.

    It is an incredibly hard road...The ideas that I have used to come up with winning futures systems, I have never seen before...

    #10     Feb 25, 2004