Having a custom backtesting system built

Discussion in 'Automated Trading' started by hilojack, Dec 17, 2010.

  1. What reports do I need this system to generate to properly show strategy results?

    Does anyone know of any examples out there I could model after?

    I haven't found anything that is "standard".

  2. Check out the links below. Granted they're not "standard" but they're a good starting point:


  3. There are alot of statistics that people use for analyzing systems and portfolios.

    I agree about Tradestation being a good starting point.
    http://www.striker.com/ has an impressive display.
    www.FuturesTruth.com has a credible selection.
    NinjaTrader has slightly different, and smaller available stats.

    Depending upon what you want to do with the system, you may well want to code it for one of the many retail backtest systems out there. Then you will have their statistics without having to debug your own interpretation. (And have people worry about the difference.)
  4. Thank you all for replies and suggestions. I have the code written and it black-boxed so that I am broker/platform agnostic.

    I don't need to optimize or anything like that, just pump in my historical data and run the metrics. I've seen what Ninja has and its a bit remedial, but I don't want to go overkill either.

    This is being presented to sophisticated investors, so I want to have all of the bases covered.
  5. dloyer


    Some of the basics:

    win %
    profit factor
    payoff ratio
  6. -risk in each trade (would be nice to measure the risk of not taking a trade too)

    -correlation of trades to each other and to an index

    -total capital at risk during the backtest, not just measured by the drawdown but by a series of measures like the size of the stop loss on the underlying, and some other portfolio liquidation rules to avoid crowded trades

    -how many signals taken versus available

    -profitability of all the signals triggered, but not taken by the portfolio(to avoid luck or reliance on overoptimization )

    -measure the liquidity at the time of the trading signal to determine strategy scalability or if the strategy has a capped sized.

    -develop techniques which measure or value trading signals and see if they are degrading or becoming more volatile and risky over time.

    -be able to combine multiple strategies and have a systematic way of sizing the bets and measuring the combined risk to the portfolio of these strategies on the PNL and risk to the portfolio.