Successful strategies, holy grail and backtesting

Discussion in 'Strategy Building' started by silent_tunes, Oct 25, 2010.

  1. I am not sure if this is the best forum but I will post anyway. It is okay if someone moves it to the appropriate forum.

    Let's say I have a few successful trading strategies. By successful, I mean the backtesting results have been good to great ! Now I know that a lot of real traders/old timers here frown upon newbies getting excited by backtesting results.

    Since nobody can predict the future successfully with any consistency, the past is all we have in expecting/predicting how the future will turn out.

    Also, my strategies require low commissions, better trading speed, good executions/fills (in other words, a retail account trading online is not optimal). The backtesting has been done to account for real world glitches (position sizing so as not to move markets, risk management, stop losses etc.)

    My question is whether there are firms out there that will let me trade my strategies real time (maybe with paper money to begin with) and if my trading results are profitable, they will let me come on board to trade real money.

    If you are asking why I don't use my own money to trade my strategies, read a couple of paragraphs above (I am retail and the results would be vastly better with better EXECUTION).

    I saw other posts with Patak, Pulsar etc. by Google searches ended up revealing that they are not legit.

    I am only 31 with full time trading experience of less than 2 years so walking into a prop firm and asking for trading previleges is out of the question.

    To phrase the question clearly, are there firms out there that will take me on board with good backtesting results instead of actual trading results or an MBA in finance?
     
  2. Curve fitting is always succesful...
     
  3. No sir, my backtesting is most definitely NOT of the type "If I had bought GOOG in 1998" or "If I had bought MSFT in 1982" etc...

    I can prove that you pick any random stock from the S&P 500 and my strategy is simple enough that it will work on that stock. In fact, the criteria for my strategy is any dumb investor (starting with me) can win using a simple strategy.

    I am not going to compete and win against the behemoths with their information networks, databases, number crunching supercomputers, tried & tested TA tools etc.

    My strategy is based on simple statistical probability.

    As I said before, I am sure most of you frown on backtesting results; but still it would be better if you can answer my original question.
     
  4. nLepwa

    nLepwa

    What's the trade frequency?
    What is your data quality?
    What is the amount of slippage/commissions that your strategy can tolerate before negative expectancy?

    Ninna
     
  5. I use back-testing to dis-qualify, not to qualify; back-testing is 10% or less reliable.
     
  6.  
  7. nLepwa

    nLepwa

    Alright, then start trading with a retail broker (IB is good).

    You need to get the ball rolling. Nobody (except friends&family) will fund you on backtesting results alone.

    Proove that your strategy works for a few months and then you stand a chance to get funding and better execution.

    Ninna
     
  8. Right. Backtesting not good enough for anything. You can backtest a system and put some extra variables and get anything you want. 10%/day if you like. Backtesting can be illusion.
     
  9. Some thoughts...

    I am suspicious of any system developed with backtesting that can only succeed with execution timing unavailable with retail platforms. Especially your weekly system.

    I'm a retired engineer so tweaking a system is in my DNA. None-the-less, I put every system through four stages. Bakctesting with out of sample historical data to avoid "curve fitting", paper trading, live trading with modest funds and finally full production. It's a lot of work, but I pay my current expenses with trading so I must be careful.

    Familiarize yourself with one or both of the following books.
    Design, Testing, and Optimization of Trading Systems by Robert Pardo

    Trading Systems and Methods by Perry J. Kaufman
     
  10. Cheese

    Cheese

    Amateur backtesting results are illusory, as here, but nevertheless they are a favorite diversion at ET. The data input, being skeletal, is of course insufficient and so correspondingly the results are always deficient or invalid. The usual fantasies of success or riches are entertained on such results.

    Instead it requires a problem solving aptitude to work out the puzzle of the markets and to construct or adopt a reliable methodology to take gains sequentially out of all the swings that make up the daily gyrations of a market, open to close (eg CL).
    :)
     
    #10     Oct 26, 2010