Backtesting in Optionvue

Discussion in 'OptionVue Forum' started by Rtb, Sep 29, 2016.

  1. Rtb

    Rtb Well-Known Member

    I have just started backtesting in Optionvue. I was backtesting a basic idea based off the greeks. The trade was up $5k at exit. A day later I went through the trade again to check it. The greeks were different in Optionvue this time at the adjustment points. I had written down the greeks at the adjustment points so I was 100% sure what they were. Because the greeks were different the adjustments were different and the trade ended up being different and the P/L was a lot less. Any feedback from Optionvue users would be appreciated.
  2. Steve S

    Steve S Well-Known Member

    Welcome to the nightmare. You are using software that is a black box insofar as they do not disclose many critically important details of their models, and which is deliberately designed to give different, unpredictable prices and greeks EXACTLY when it should never do so. Caveat emptor.
    Chaitanya and Rtb like this.
  3. uwe

    uwe Well-Known Member

    Yes, if you reload back trader data the greek change. I also find this strange. The data is exactly the same, so I expect the greeks to be exactly the same too.
  4. AKJ

    AKJ Well-Known Member

    I agree with earlier comments on it being a blackbox.

    Thought I would add some of my own thoughts as to WHY the blackbox gives different results when you would expect the same output. I think what we have going on is that OptionVue looks at an extremely large quantity of current and historical data for predict how P&L and Greeks will change at different underlyings and at different times in the future. It doesn't have the computational power to do all the necessary computations quickly, so there is some algorithm that it uses to reduce the number of calculations OV needs to do. This method allows for OV to spit out an answer quickly and was most likely designed to find the right balance of speed and accuracy. The algorithm that determines how OV goes about doing all the calculations and determining which short-cuts to take probably has some random seed value that can change with each time-step, and this can lead to slight (and sometimes not so slight) variations in the results you see.

    Now I haven't tested this, but I would assume that if you use the Black-Scholes setting and EIOIO, and turn off the things such as "True Delta" and "True Theta" then the output would be consistent each time you ran them. These calculations are pretty straightforward and easy to do for a system like OV, and should NOT require any estimation techniques that improve speed and reduce accuracy. The output should just be the sum of the greeks of the individual positions. If people find that this is not the case, then I would have some serious questions about what OV is doing.
    Rtb and uwe like this.
  5. Rtb

    Rtb Well-Known Member

    Thanks for the replies guys. I am going to spend the weekend backtesting. Andrew I will try what you have recommended and report back.
  6. Rtb

    Rtb Well-Known Member

    I used the Black-Scholes setting and EIOIO, and turned off "True Delta" and "True Theta" (used Standard models). I recorded the greeks for 20 trading sessions on a buttefly spread. When i replayed the trade in Backtrader what I noticed was, the greeks were exactly the same throughout the 20 trading sessions if they were exactly the same on the 1st trading day. If the greeks were different on the 1st trading day, they were different for all of the 20 trading sessions. This varied by only being fractional different to being different.
  7. Rtb

    Rtb Well-Known Member

    I just recorded the greeks again for 20 trading sessions on a butterfly trade like I did on Friday. When i replayed the trade in Backtrader this time the greeks were exactly the same each time and I replayed the trade multiple times. This wasnt the case last week. I upgraded to 8.02 today. Not sure if this has helped resolve the issue.

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice