OptionVue new ver. 7.71 Modelling Problems

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

  1. tom

    tom Administrator Staff Member

    We talked about this in our trading group meeting today. Try going to the quotes display, right clicking on the symbol and click the DA button. This will delete the asset file, but it leaves your trades untouched. Then double click on the symbol and OptionVue will rebuild the asset file. This seemed to do the trick for one our members and the Greeks seemed to be what he expected after doing this procedure.

    HTH
     
    GreenZone and Chuck like this.
  2. Jay Winger

    Jay Winger Active Member

    Thanks Tom,
    It does get your delta closer, but I just want to make it clear that we are still not back to the old model - the T+0 shape and projections (and thus the gamma in some of my positions) are pretty far off, especially on the upside.
     
  3. DavidF

    DavidF Well-Known Member

    Jay, not sure what positions you have on but (as I mentioned in another thread on variable vs EIOIO) my deltas are completely thrown off unless I have "small" on the calls as strike preference (on variable model). My current SPX position has +17 on small and minus 58 on moderate/large.
     
  4. Jay Hattler

    Jay Hattler Well-Known Member

    I have started deleting my asset files on a regular basis (every day or two) and letting OV rebuild them. I also consistently use temporary files when using Backtrader. My subjective experience is that I have had fewer problems, fewer of the Access Violations, etc. in working this way.
     
  5. Jay Winger

    Jay Winger Active Member

    Jay, what do you mean by using temporary files when using Backtrader?
     
  6. Jay Winger

    Jay Winger Active Member

    I found that using the shortcut key to reload prices, Ctrl-T, will often vastly improve the T+0 line, but it will often flip back to a mutant T+0 when I move a time interval in Backtrader, requiring me to reload prices to get it back to proper.

    Before hitting Ctrl-T:
    http://screencast.com/t/tTri8OHqa

    After hitting Ctrl-T, much closer to previous models:
    http://screencast.com/t/vXmKYYTlz
     
    GreenZone and Kevin Lee like this.
  7. Kevin Lee

    Kevin Lee Well-Known Member

    My goodness.... how many little tricks we have to learn to get OV to work ? :( Thanx Jay.
     
    Andrei likes this.
  8. Georges

    Georges Well-Known Member

    When you have a backup from an old version. You can restore that version and don't upgrade as long as problems are not resolved.
    I don't know there could be another modelling problem when using an older version. Every day when start OV the first time, it download the daily files.
    Possible in these daily files OV download CEV data and is based on the modelling version 7.72 and not modelled on an older version 7.67.
    Perhaps Jef can give the right answer how this works out?
     
  9. Georges

    Georges Well-Known Member

    About Implied volatility calculation in the Matrix when using back trader and go back in time.
    See, View, System Models, Interest Rates.
    When going back in time with backtrader Interest Rates remain the same values as today.
    To have the right Implied Volatilities in the matrix,
    adjust these values manually to the interest rates for the dates you test in back trader.

    Example for the month that you test, enter the values of this period.

    It would be great backtrader doing this automatically in a next upgrade.
     
  10. Jay Hattler

    Jay Hattler Well-Known Member

    When you start Backtrader, you are given the option to store the historical data in a temporary asset file. See the attached screenshot.

    2015-11-04_10-27-32.png
     
  11. ACS

    ACS Well-Known Member

    Here are the screen shots from the earlier and current versions.
     

    Attached Files:

  12. Capt Hobbes

    Capt Hobbes Well-Known Member

    Did you run the older version and it just worked? I found two backups of 7.64, but in both of them after I delete the asset file for RUT, OV fails trying to recreate it, giving a message "Unable to find background data for RUT".
     
  13. ACS

    ACS Well-Known Member

    I still have 7.66 on an old backup laptop and I just recreated the T Log for the M3. Quite a difference in Greeks and T+0 line for a couple of minutes apart.
     
  14. Jay Hattler

    Jay Hattler Well-Known Member

    I have just received an email from OptionVue with a discussion of what has happened over the last few days. (It looks like the email has gone out to all current OV customers.) I found it very helpful in explaining what has happened, as well as containing easily understandable info about how the Vertical Skew works. The tone was also quite good and customer-focused. I would be interested to hear how the more advanced members of this forum perceive the explanations given, especially in light of the claims being made in this thread about differences between 7.72 and the 7.6x versions.

    Jay Hattler
     
  15. Jay Hattler

    Jay Hattler Well-Known Member

    Attached is a PDF of the email mentioned above. (I hope I got all the content into the PDF.) I tried to copy and paste the email directly into this forum, but the images don't show up and they are important to understanding the emaill, at least for me. I hope this is helpful to everyone.

    Jay Hattler
     

    Attached Files:

  16. Capt Hobbes

    Capt Hobbes Well-Known Member

    Thank you for sharing this, Jay. The explanation is good, but I wish someone from OV was more active here on the forum and looked into the reports about the differences between 7.72 and 7.66, like the ones posted by ACS. I wasn't able to get an older version running, but today in 7.72 my November M3 was wobbling between -59 and -100 delta without any appreciable price movement within just a few minutes. At the same time TOS with volatility smile approximation was showing a pretty stable -11 delta.

    What concerns me in Len's letter, and I'm going to send the following to OV, is it appears that he doesn't realize the real significance of the skew model to someone trading a trading plan in which precise values of the greeks (a 10-20 change in delta for a $50K PC position is pretty precise) trigger specific adjustments. If a model is changed so that in the same situation it produces different greeks, the adjustments will be triggered at different times, and possibly the adjustments too will be different. The plan will produce different results, and the results may be better or worse, up to potentially turning the plan into a negative expectancy game. The only way to know for sure is to backtest with the new model, comparing the results with the old backtests.

    An improved model is nice--presumably it would produce the t+0 line even more predictive of the reality--but it doesn't mean it will necessarily improve the performance of a specific trading plan built for the old model. It's as if a stock charting package suddenly replaced the formula for, say, the RSI indicator with something that's supposedly better, but completely replaces the original RSI.

    Which all means, the core issue here is not more extensive beta testing to ensure the software is "stable and bug free". It's that the current model behavior--albeit not perfect--is depended upon and must not be changed freely and without notice. It doesn't matter that the change is an improvement. Ideally, the older model(s) should still be available as an option. If not, the main purpose of beta testing with customers it to let them verify that the new model doesn't break their strategies and give a green light for setting the changes in stone.
     
  17. Kevin Lee

    Kevin Lee Well-Known Member

    Thanx Jay. I wonder why I did not receive this email from Optionvue. I seriously think they should be much more proactive in making use of community like this one to gather feedback as well as address issues. Nowadays companies spend millions of marketing dollars just to get access to community of customers and opinion leaders. Access to this community is free. Not sure why OV is not more proactive in leveraging this channel. Someone should send this message to Len.
     
  18. DavidF

    DavidF Well-Known Member

    Kevin, I never recieved the mail either, have written to OV about this so I assume they'll be updating their mailing list.
    The mail confirms what I'd observed, the call settings on autostrike "small" vs "moderate/large" are a major issue in the large delta variations. My all-put positions are ok (rhino) and only display marked position delta variations in 2 settings:- EIOIO vs variable model (e.g., minus 50 vs minus 100 repsecitvley). Where there are calls (M3, Kevlar) the large variations in positon delta are dependent upon one additional setting, the autostrike input.

    Am new to OV and still not sure if the advantage of the variable model vs. EIOIO is more in relation to dynamic greeks, i.e., where they'll go with underlying price changes, or if it's also more accurate with static greeks, i.e, where they are currently.
     
    Last edited: Nov 5, 2015
  19. Jay Winger

    Jay Winger Active Member

    OV IS NOT FIXED
    Contrary to the letter from Len, OV is NOT fixed. Jim and Len appear to have an entrenched position they are defending for some reason instead of deferring to users, who are the ones on the front lines and are the ones who have intimate knowledge of the previous model in relation to their strategies (sometimes built from hundreds of hours of backtesting and live trading).

    Capt Hobbes is right, it's not about adding refinements on top of the model to make it 'better'. As I told Ken Dole on the phone after the ver 7.65 modelling problem a few months ago, if any new model is introduced, we NEED time verify the changes to the model and analyze how they affect our finely tuned strategies. At the very least this means allowing us to run previous versions with the previous model, so we can run them side-by-side for awhile. So what did they do? They attached a change to the core modeling function to a Mandatory Upgrade (because of a server side change due to another change they made in the release) - NOT the time to do that.

    Getting back to the continuing modelling differences, I won't run on in this post, I'll just share an Evernote note:

    http://www.evernote.com/l/AAJDRbVuZzREp4cFfk-lq9WQl6tRptn_lZc/
     
    Andrei likes this.
  20. Jay Hattler

    Jay Hattler Well-Known Member

    Looks like the first public beta for CEV modeling in the ONE software has been made available in the last couple of days. Perhaps this will eventually offer more alternatives and also prompt a healthy competition and attention to customer concerns across vendors. I believe some of the more advanced users in this forum have been involved in this (Kevin? Ron? Sorry, can't remember who). Perhaps Tom could have Andy come on a session to discuss.
    2015-11-05_10-00-05.png
     

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