pat rogers backtesting journal

Discussion in 'Journals' started by progers82, Mar 29, 2006.

  1. The things that I have struggled with throughout the years has been backtesting. I could take my system and over optimize and make any report look good. I was continually chasing the best that performed best two months ago.
    So one long three day weekend i spent close to the entire 72 hours trying to prove or disapprove this new theory I had.

    This is what I did.
    I backtested my portfolio from june 30 2005 back the following days. 30, 60, 90 200, 400, 600, 800 and 900 days. then I moved forward from july 1, 2005 to march 1 2006 to see the results on the unseen data. For my portfolio, there was a very consistent result and that is the more days i tested, the better results for the future period. 800 and 900 days were very close to having the best results.
     
  2. Bob111

    Bob111

    now open IB paper account and trade for few months, to gain some confidence. when you feel ready-swith to real account. take it easy,start with small size.:D
     
  3. then i tried one other theory. What if i backtested back 600 or 800 or 900 days back from june 30 2005 and then recorded the results for the future two months. then i would move up my back testing period to august 31 2005 and re test 600 800 and 900 days and again i recorded profit for two future months of unseen data.

    The results really surprized me. 600 days back of backtesting was giving my portfolio the best results of future trading.

    so on march 1, 2006 I did a through backtesting, 600 days back and those inputs are the inputs I am using for the months of march and april. the next time I will back test will be on april 30 and use those inputs for may and june.

    actual trading results using TS8.1 automated trading produced the following results for march. these are live results. I will keep you updated on both my backtesting results and my live trading.
     
  4. im trading one emini es with a $6,000 account. forgot to mention that. the system takes between 3 and 6 trades a day and does not hold overnight.
     
  5. todays trade
     
  6. trades through today on my $6,000 system that was backtested 600 days from march 1, 2006 and will be re backtested on may1, 2006. i will then use those variables for the months of may and june only. these are live trades and past results cannot be guaranteed into the future.
    pat
     
  7. the first 10 days of April have been flat with no real gains. But, no losses. So we will keep trading the same inputs that were optimized on March 1, 2006. I do plan on optimizing all of my systems on April 30 again. There are 11 systems and close to 45 inputs in all 11 systems so the process of optimizing will take a couple days. If my theory proves true, I should be able to show that live results from may 1 2006 to june 30 2006 will be better if i use the new inputs as opposed to the old inputs. as a reminder, i backtest 600 days and will optimize every two months.
    the attachment are live trading results but i can't guarantee that i can do that again in the future.
     
  8. NOTHING new. good day today. will re backtest on april 30, 2006. will also post the two month actual report then.
    pat rogers
     
  9. LOOKS like we are going to end April with a small profit. I will present a financial report of live trades in the attached file.


    to summarize, on march 31, 2006, i optimized the xxxx out of my sytem to get the best inputs to move forward. I have recorded the values and will use as my benchmark. Since this weedend is the first of may, im going to again optimize the xxxx out of my system going 600 days back. in the next two months i will report what the benchmark profits/losses for may and june against the profit/losses for the newly optimized system. and i will keep doing that. again im trading this live with a $6,000 account at TS8 and RJ Obrien.

    any constructive comments, feel free to speak up.
     
  10. a couple things.

    1)I have accepted a position as system designer with Apphia Capital LLC and

    2)I have renewed my membership with the NFA/CFTC

    And because of that, I think it best that I discontinue discussing my theories on backtesting and hypothetical results.
    I didn't have enough time to prove my theories in real time.

    Apphia Capital started trading my stuff on March 1, 2006, and My NFA membership was effective last part of April, 2006.

    thank you for allowing me to share.
     
    #10     May 3, 2006