Quantcast
  1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

OptionVue new ver. 7.71 Modelling Problems

Discussion in 'OptionVue Forum' started by Jay Winger, Oct 30, 2015.

  1. Jay Winger

    Jay Winger Active Member

    On 10/29/15, OptionVue came out with a new Mandatory Upgrade ver. 7.71, which among other changes (changes to OTM call skew modelling, expanded CEV time period reporting) appears to have tweaked the CEV model, reminiscent of ver. 7.65, which was pulled for having some issues. Well the issues are back, and then some. After going back through two years of backtested trades in Backtrader, I found instance after instance of T+0 line wonkiness, and thus greeks instability, effectively rendering my backtested adjustments worthless, and bringing into question the modelling in general, for me.

    In today's live trading I saw the same issues on live data, and numerous other traders in our groups reported and showed the same erroneous modelling behavior. I would have positions go from -60 delta with a convex T+0 'crick' near current price (collapsing gamma) to a convex crick below the money, sending delta to +30. These mutations would hold for some time.

    Below is a quick example of the type of T+0s seen on two expirations of backtrader data:

    http://www.evernote.com/l/AAKojlFYdqFK2JtmjmfTMYwg0coFGUsRAIs/

    If you haven't yet seen a clear manifestation of these issues yet, I urge you to go back through several months of backtrader data and see if it crops up - beyond the T+0 shape, look to see if there is unusual instability in greeks.

    Consider reporting this to OptionVue support at support@optionvue.com . Part of the problem is this is a Mandatory Upgrade, meaning we can't opt out of it, can't compare previous modelling to the new one, and can't complete due diligence on the core modelling changes they've made. I have emailed OV pleading with them to at least take off the mandatory feature of this upgrade.

    Jay
     
  2. Capt Hobbes

    Capt Hobbes Well-Known Member

    Indeed. Here is a very simple position (Rhino "simple" trade example entry from Brian's presentation), just a BWB.
    upload_2015-10-30_23-8-39.png
     

    Attached Files:

  3. ACS

    ACS Well-Known Member

    I noticed a significant change in the T+0 line of my live M3 where it slopes down on the right.
     
  4. MattW

    MattW Moderator Staff Member

    I got a response from Ken and sent over screenshots. No word as to what the solution is yet, but he did say they had a "hectic" day over there on Friday.

    Kind of reminds me of the Seinfeld episode in which the building super comes around and installs "low-flow" shower heads in all the apartments and there's a bit of an uproar.
     
  5. Jay Hattler

    Jay Hattler Well-Known Member

    2015-10-31_16-38-07.png Yes, sometihng has gone seriously wrong. Here is an old iron condor, modeled in Backtrader using version 7.71
     
  6. MattW

    MattW Moderator Staff Member

    Send it in if you haven't already.
     
  7. Capt Hobbes

    Capt Hobbes Well-Known Member

    No kidding. They really ought to have some sort of a release process instead of just throwing some bits over the wall every once in a while.
     
  8. Jeff

    Jeff New Member

    Hi everyone, my name is Jeff Plimpton, and I'm a Product Consultant for OptionVue. Len is working out the modeling
    bugs and will have a new release at the end of today. Please keep in mind that when there are issues with a release,
    we have to recreate the problem in order to come up with a solution. If any of you would like to send over an asset
    file to help with the process, we would appreciate the support. A picture from the Graphic Analysis won't solve the
    problem if the same trade looks fine when we enter the position on our software. Thanks...
     
  9. tom

    tom Administrator Staff Member

    Thanks Jeff. Where should people send the asset files to with a description of the problem they are having?
     
  10. uwe

    uwe Well-Known Member

    Off topic:
    Every time OV is talking about software development (be it new features or bugs), it is always Len who is working on it. Is your founder your only software developer?
     
  11. uwe

    uwe Well-Known Member

    I updated to 7.72 - seems to be even worse:
    screenshot.png
    A bearish butterfly with that many DTE never has the T+0 peak right below its center. In the past (before 7.71) this was better.

    I deleted my asset file and started fresh - no change.
     
  12. Jeff

    Jeff New Member

    You can send asset files to jim@optionvue.com...We really appreciate everyone's patience and help with this release...

    Jeff
     
  13. Jay Hattler

    Jay Hattler Well-Known Member

    Here is the exact same screen shot of my IC as above, but this time with version 7.72. Certainly looks much more like it used to before 7.71. (I don't have a screenshot of the IC prior to 7.71.) I have sent the current SPX.INX flile to OptionVue as requested.
    2015-11-02_19-16-01.png
     
  14. Jay Winger

    Jay Winger Active Member

    Hi Jeff,
    I appreciate you taking the time to respond on here and making the customer outreach, I must say it has been sorely lacking on the individual level on this issue (I started emailing Ken, James and the general support site with specific examples and screenshots starting Thursday evening with nary a substantive response).

    I am not entirely happy to hear that this doesn't sound like a true rollback to the previous model as much as an educated guess to unwind a problem introduced. Back on the ver 7.65 release, when there were also modelling issues (subsequently rolled back), I stressed with Ken the importance of allowing us access to previous versions whenever modelling changes are made, in order to allow us time to investigate what impact the changes will have on the strategies we have developed, sometimes after hundreds of hours of backtesting. Modelling changes attached to a Mandatory Upgrade effectively handcuff us to the new modelling, flawed or not, and cut us off from all our past research.....and it sounds like they've handcuffed Len/Jim as well, from being able to simply roll back to the prior long-standing model.

    I was pretty much trading blind today managing my many positions, and now there is a seed of doubt that the model may not be the same. I don't feel we should have to play detective to find out where the cracks are, when a working model is out there that we have built everything upon, and we should have an avenue to simply go back to it. This is our livelihood, and we depend on some consistency and reliability in our main trading tool while these tweaks are done to the core of the tool.
     
    Last edited: Nov 3, 2015
  15. MattW

    MattW Moderator Staff Member

    Based on a few backtests, it's not the same model. My T+0 line isn't predicting P&L at all. My black dot spent 85% of the trades under the T+0 by a large margin (not the case prior to the update) and only caught up to the line on large vol days. EIOIO is more accurate than whatever they have going on now. I seem to be profitable even using EIOIO numbers, so I guess I shouldn't mind all that much, but there is definitely a sense of "trading blind". I'd like to trust whichever model I'm using. On my three lot live trade, OV is giving me -15 delta, IB is giving my -7 and ONE is giving me 0. Who to trust?:(
     
  16. GreenZone

    GreenZone Well-Known Member

    Matt, it's perfectly normal for the black dot to not match up with the variable variance T+0 line.
    Yes, when using EIOIO, the black dot will match up, but that T+0 won't give you accurate projections of your greeks compared to variable variance.
     
  17. DavidF

    DavidF Well-Known Member

    One thing I've noticed is that the variable model is over-predicting upside on a down move. For example in a butterfly/M3 set-up, I'm short 100 SPX deltas and a 2% pullback predicts a profit of around $4k. However, I'm also short >1000 vega and the prediction would only be correct if I was short /ES futures, not options. The EIOIO model is flat and seems more accurate, i.e., that the vega would offset the deltas. Hoping it's just a temporary blip as I understand the variable model should be more accurate (and is relied on by many who know their stuff).
     
  18. Kevin Lee

    Kevin Lee Well-Known Member

    Jeff, there are many advance OV users in this community. Our trades are highly sensitive to IV modeling. Why not suggest to Len to use this community to test out any changes to IV modeling before rolling it out ? I'm sure there'll be many who will be willing to sign up as beta testers.
     
  19. MattW

    MattW Moderator Staff Member

    Yeah, I'm seeing the over-prediction of profit as well. When I initiate a trade, the T+0 line is telling me I should already be about a quarter of the way toward my profit target, which is absurd of course. There's also way more gamma/curvature than I would usually see with as many DTE and IV as I'm working with. It's as if the T+0 line has become a T+20 line without being labeled as such.
     
  20. Jay Winger

    Jay Winger Active Member

    OK, I was forced to play detective and go back to screenshots I happened to still have of positions from ver 7.64 to find out if there truly were differences, because things just didn't look right after the 7.72 update, and sure enough there is still something horribly wrong:

    http://www.evernote.com/l/AAKcymLojiRGY4q5j_GCavHzqtkv6f0UsG4/

    Another trader I'm in contact with found that the 'G/L to include previously realized G/L's' function was not working properly for him, and Len was able to reproduce it (it works for me so far), and Len is currently trying to play detective to fix THAT part of the code as well.

    Jeff, it's time to truly roll back to ver 7.67 or earlier, enough is enough. I have tens of thousands of dollars in positions that I simply can't manage now.
     
    Last edited: Nov 3, 2015
    Scott Slivnik, Kevin Lee and uwe like this.

Share This Page