Developing trading systems

Discussion in 'Strategy Development' started by Murray Ruggiero, Aug 10, 2011.

  1. Murray Ruggiero

    Murray Ruggiero ET Sponsor

    Most people address system development wrong. They simply try classic ideas and try to force them to work.
    When developing systems you should first develop a premise and then test your premise. You are looking for your idea to perform like expected.

    In this thread we will walk though how to develop reliable trading systems.
  2. Most people (about 98% ) expect 100% winning trades.
    If your system will have two losing trades in a row, you'll be percieved as very bad system developer.

    Are you sure you want to discuss such matters with the public?
  3. From what i've learned over 10 years, the key to build successful trading systems is simplicity. Its precisely what people will avoid at all cost. Se Jack Hershey threads. The more complex method is, the more popular it is. And the more people lose trading it.

    Are your systems simple? Can be coded within 20 lines of code?
  4. Murray Ruggiero

    Murray Ruggiero ET Sponsor

    I totally agree with you , most good systems are about a half of a page of code. With that said sometimes concepts take more code, for example if you where using tradestation a concept like skipping a trade after a large winner could take a page of code all by itself , but the concept is simple.

    In TradersStudio we have functions for many concepts like disabling a signal after it was stopped out until a given condition is met. Another set of functions we have allow you to filter a system based on performance, like skipping trades after a large winner or filtering signals based on equity curve performance.
  5. Murray - developing trading systems manually nowadays is like writing email on a piece of paper and sending it via post office. Even if you have what appears to be a good idea today it doesn't worth the effort because it may turn out to be a bad idea in the near future. You have to shoot for the extraordinary system and this requires considering many different markets and using loads of data in mechanical ways.

    Google for trading system lab, price action lab, safir-xp. In the time you can think, code and test one system, these program can come up with systems 10x better with the parameters you want.
  6. Jack Hershey's method is simple. It's built on SPX fidelity data in wealth lab pro, and is available hard coded. He blathers on about stuff that has absolutely nothing to do with the actual system, and you can test it yourself by looking up my posts on the subject.

    20 lines of code is a joke system.
  7. This is done by curve fitting. Not the same as having a working method for every symbol. Optimization for methods that are consistent won't produce out of sample results different or significantly different than the results you backtested on.

    10x better? Hardly, maybe if you start with a 0.5% apr and make it a 5% APR. Even then, still with 20-50% drawdowns, so those programs don't work. I've seen the neural networks look at candlestick bars and all they are doing is curve fitting.

    If you have a strategy that can be applied the same way every time, your optimization is not a curve fit, but simply by optimizing does not make it a curve fit if you have enough evidence that it has statistically significant results and produces robust results.

    A 1 year backtest that makes 80% and drawsdown 10% is a curve fit due to its short term data. Adding 1 year will likely negate whatever edge it might have found, but working with at least 2 years should solve that problem. I'm not opposed to using all of the data. I think that just limits how robust the system is, and if you find your system needs to be optimized regularly then optimize regularly. There's nothing wrong with that. You always want to have the best edge possible, but certainly if you take it too far you'll get results on a walk forward that aren't what you'd expected.

    People with no experience writing their first system have to sort a lot of ideas and once you've been able to classify system types eventually you come to a realization about what works and what doesn't, but it's a lot of coding and backtesting.

    20 lines? Lol. Every good system I've seen with 12 point font is at least 20 pages long. That's around 500 lines of code.
  8. Murray Ruggiero

    Murray Ruggiero ET Sponsor

    See, the one thing about writing a system manually or semi manually is that you create a valid premise, your testing a premise. Automated system develop can lead to beyond curve fitting.

    I remember a trader who made a lot of money from 1998 to 2002 trading a system for the Nasdaq ,developed using models which were not filtered using human expertise. He lost 40% in 2002-2003.
    Why , because the core logic of his system was based on the 5 day price change, it accounted for 80% correlation to the main indicator which was created for him.

    He was trading the Nasdaq with this system and did not know that this was the underlying logic because it was developed using a neural network and about 1 dozen inputs.

    See, domain expertise is king and will alway be if you want to build reliable system.

    Advance methods like Genetic algorithms and programming, neural networks,SVM are useful but must be guided by domain expertise and used to prove and fine tune a given premise.
    They also can be used to search for relationships but then a human expert should use these and formulate valid trading premises.
  9. BoWo,

    For christ sake, give it up already! You're so full of useless knowledge its getting sad.

    Do you want me to link to all your losing systems on C2, as in every system you've posted has lost money for years on end? All of 'em?

    After this many posts and after so many have tried to steer you in the right direction, you're still stuck. Unbelievable.

    Your post is 95% BS. You haven't learned anything after all these years.

    Let the pros here discuss this material and STFU. I sincerely hope people reading do their homework about you and realize how misguided you still are.


    P.S. My best system is 18 lines of code! Not including variable declaration etc... with all that its about 30.
  10. a5519


    This is really funny to read such writings.

    Look at this classical example, it is a good illustration where can your approach lead to.

    Similar "models" can be seen everyday also in trading.

    Best of all read carefully the post of Murray. This time he has disclosed a secret how to bild a working tradin system.
    #10     Aug 10, 2011