Could it be this easy?

Discussion in 'Trading' started by farmerjohn1324, Dec 31, 2022.

  1. I recently ran a backtest on NinjaTrader on 1minute charts over the past 12 months.

    I'm not even going to go into the details of the script because it was one of the preloaded MA scripts that comes included as a sample.

    The results were a profit of $30,950 on 1 ES contract ($13,000 maintenance margin, I believe). This is a return of 138%.

    Now the obvious answer of "if it was that easy, everyone would do it," I think applies here so I must be missing something.

    But I'm not really talking about the details of this particular algorithm. I've seen countless algorithms that backtest to similar or better results.

    Can anyone give me specific reason why this wouldn't work this way if traded live? Or specific reasons that it would?
     
    HeSaidSheSaid and Darc like this.
  2. wrbtrader

    wrbtrader

    Just try it with the minimum position/real money trading...you will not need answers from an anonymous trading forum.

    wrbtrader
     
    spy, zghorner, Rams Fan and 1 other person like this.
  3. Sekiyo

    Sekiyo

    Don’t forget to include fees.
    Commission and slippage.

    Past 12 months on 1 minute is plenty of data.

    Also make sure there is no error in logic.
    I mean stuff that aren’t realistic.

    Check if there is no margin call in between.

    Not sure how NT backtesting engine works,
    But I am suspicious that a built in strategy is returning 100% P.A

    I don’t know how many parameters there are,
    But don’t cherry pick based on performance.

    If MA(50) returns 22k
    If MA(49) returns 122k
    If MA(48) returns 18K

    Then MA(49) is simply over fitting,
    You won’t be able to reproduce the same.
     
    Last edited: Dec 31, 2022
    oshjdf, farmerjohn1324 and Darc like this.
  4. schizo

    schizo

    As @Sekiyo already pointed out, commish and slippage alone will take a big chunk out of your profit if you're using market orders. Also, when backtesting, there's the issue with overfitting. To avoid this, it's better to forward-test or run the algo in a sim for an extended period. You will still need to take fees into consideration. In NT, you can configure commission and slippage.

    upload_2022-12-31_12-56-38.png

    upload_2022-12-31_12-58-26.png

    It's also worth watchin this video




    Finally, there's an inherent flaw with a lot of backtest engines on the market. What do I mean? Take a look at this video. The video is about TradingView, but the same limitation he talks about applies to other platforms IMO.

     
  5. Yea I'll have my answer in 12 months.
     
  6. You seem knowledgeable and experienced...

    Have you ever personally back tested and decided to run an algorithm that has generated consistent profit? Have you considered it and then decided against it?
     
  7. schizo

    schizo

    As they say, "When it's too good to be true, it's usually a fake." I personally heed that advice for everything in life, even with backtests.

    Hope that helps.
     
    MarkBrown and Darc like this.
  8. Snuskpelle

    Snuskpelle

    On a positive note: Current market conditions are volatile and unusually trendy (compared to past decades). Certain simple algos can work (for now).

    On a negative note: Poorly made backtests lie (and unfortunately quite a few vendors do it poorly on purpose) w.r.t. to transaction overhead. A lot of algos look great if you get to buy instantly on the midpoint price in sim but rapidly lose money IRL...
     
    Darc likes this.
  9. MarkBrown

    MarkBrown

    this will not end well....
     
    Bad_Badness likes this.
  10. Darc

    Darc

    I think we've all seen Systems advertised for sale, that boast great results from backtesting, but the Owner fails to come close to in reality.
     
    #10     Dec 31, 2022