Deep Dive - IV Skew Impact on Butterflies

Discussion in 'Round Table Presentations' started by Kevin Lee, Sep 11, 2016.

  1. Kevin Lee

    Kevin Lee Well-Known Member

    I created a video to answer some questions about IV skews and butterflies. Let me know your comments or if you have further questions.

  2. LuisG

    LuisG New Member

    Thank you very much.
    Your explanation of IV skew makes it so easy to understand the concept as well as its impact on the option pricing.
    You have developed such an elegant tool and more importantly how to use it in option trading.
    I enjoyed every minute of the presentation,
    Kevin Lee likes this.
  3. GreenZone

    GreenZone Well-Known Member

    Hi Kevin
    Good skew video.

    I have two extra bits of feedback:

    1) I fully agree that we need to look at the extrinsic of each strike, in order to then work out the "volatility impact" of the skew changes. Even though the name "weighted vega" still makes perfect sense in terms of describing this relationship, I think it may end up confusing people a bit, since we normally use the phrase "weighted vega" when describing the effects of lessening volatility sensitivity the further you go out in time.
    You may therefore wish to use a slightly different name to refer to this relationship you mention in the video.

    2) Issue of "skew cross-over"
    I noticed that same cross-over effect some time ago and my mind started buzzing with possibilities of ways we could potentially start taking advantage of it.
    The skew cross-over suggests that even if IV increases for ATM strikes, that very far OTM strikes would actually have IV levels go down compared to where they were trading the prior day.
    I tested this out by simply buying a far OTM put at a location where the skew curve showed that IV at that strike would actually go down.
    I ended up finding that the IV for this far OTM strike still ended going up rather than down.
    I repeated this test a few times and it seemed to be consistent.
    It therefore made me think that this skew cross-over effect may just be a modelling thing, which doesn't seem to actually play out in reality.
    I still need to investigate this further, but I thought I would share this with you and others so that more people can get involved in the testing.
    Kevin Lee likes this.
  4. DGH

    DGH Administrator

    Excellent presentation, Kevin. Thanks for sharing with the CD community.
    Kevin Lee likes this.
  5. DavidF

    DavidF Well-Known Member

    Fantastic overview Kevin, also liked the rationale at end regarding why it´s important to understand IV skew with respect to adapting to changing markets.

    I´d be very interested in your thoughts on skew on SPX vs. RUT.

    I´ve written about this a fair bit before but based on backtesting and current observations I don´t think RUT pricing provides the same risk-reward as SPX. I´m unable to make the same returrns with RUT M3s in backtesting versus SPX. I also can´t get the t=0 line anywhere nr. as good on RUT as I can on SPX.

    If you buy a 50 point wing BF it´s more expensive on RUT vs. SPX. The obvious explanation is that SPX has a higher expected move and you´re thus being compensated for the fact that the price will end up outside your 50 point wing butterfly.

    But the modelling/backtesting doesn´t support this. I can place a 100 delta 50 point wing BF on SPX with a DITM call that has essentially no risk to the upside. I can´t do that with the RUT, i.e., the fly is costing more in relation to the implied move versus SPX. The downside is also better on the SPX, as is gamma. I´ve included an example below of 50 point wings M3s placed on Friday close. This isn´t cherry picking, I´ve found this at multiple historical dates (2015/2016) at various levels of IV.

    My amateur interpretation is that these changes are related to IV skew and that RUT is distorted as it´s less favoured by institutional hedgers and more favoured by income traders. If you look at when and where the volume comes into RUT it almost looks like it´s a vehicle for income traders. The risk premium and IV skew is thus unfavourable. To be more concrete there could be a flood of sellers of the 20 point under the money strike at 50-60 DTEs on the RUT that lessens premium (and IV) of that strike and distorts skew. Likewise for the long strikes. The OI can almost double on these strikes in this time period.

    Are these ideas way off? Know a lot of top traders use RUT so would imagine they´d have sniffed this out a long time ago.

    Screen Shot 2016-09-12 at 10.07.22.png Screen Shot 2016-09-12 at 10.05.51.png
    Last edited: Sep 12, 2016
    Challenger, Rakesh and Kevin Lee like this.
  6. Kevin Lee

    Kevin Lee Well-Known Member


    Interesting question. Unfortunately I don't have a ready answer. Skew could be a reasonable answer but I shouldn't speculate. It's worth doing some research.
  7. Kevin Lee

    Kevin Lee Well-Known Member


    Good points.

    On the weighted vega... I used that term for a lack of creativity. I was thinking this is a weighted vega for vertical skew whereas what you're referring to is for horizontal skew.

    I'm interested to hear what you find out about the skew cross over. That's quite common in my research. I'm planning to do a more thorough characterization of IV skew shapes and under what circumstances do they appear. Instead of just picking some samples to form an intuition, I'm thinking of doing a comprehensive quantitative analysis on the past few years of data. Only problem is the amount of data is so huge, Excel isn't able to handle it very well. I'm dumping them into mySQL and probably have to use something else to analyze. Haven't found time to really do it yet but will eventually get to it.
  8. jim

    jim Well-Known Member

    You did a great job with the basics of PUT Skew. As you mentioned in your video, PUT Skew in not linear. That is KEY!!!
    One thing that that the Institutional Options / Risk / Skew tools offer, is the ability to model the change the non-linear shape of the PUT Skew. Unfortunately, Put Skew modeling is the single most important aspect of an option modeling tools. About four years ago, I had the great opportunity of having one of the proprietary options trading systems that I developed to go through a Quality Review with the former Head of Quantitive Trading for JP Morgan. Everything went absolutely great until the Q & A section when he asked me "What type of PUT Skew modeling are you using in your assumptions." That question ended our meeting and sent be back to the design room.

    As for how PUT Skew changes with market declines, the answer is "it depends." It depends on the what people (mostly institutional) are setting up the hedges for further declines. What are they trying to protect? Down to what level? For example, earlier this year in JAN and FEB, the market dropped all IV went up, but the closer to the money IV went up more that the IV's far below money. No Strike's IV went down.

    However, in Brexit inspired market decline in late June, the IV's of all Strikes went up, and the PUT Skew got steeper. See chart of 2-month PUT skew on the S&P 500.

  9. Rakesh

    Rakesh Active Member

    Hi Jim, What do you mean by "2m PUT skew"? More specifically, can you give more details on how to read this chart?
  10. jim

    jim Well-Known Member

    Yes, but only a little bit. The "2m PUT skew" is a 2 month, or to be more specific a normalized 60-day, interpolated moving average, not this similar to the way that the VIX uses a normalized 30 day interpolated moving average. However, it uses the difference between the IV of different Strikes. It is a formula based on research by HULL and Goldman Sachs and then updated and modified by 25 years of real-world trading experience of professional and institutional options traders.

    The exact formula is the proprietary intellectual property of EAB Investment Group. EAB is the ex-Goldman guy that I worked with to take two mutual funds public: namely the James Alpha Managed Risk Domestic Equity (symbol: JDIEX) and James Alpha Managed Risk Emerging Markets (symbol: JEIMX)

    Sorry, but that is all the information that I am allowed to share.
  11. Steve S

    Steve S Well-Known Member

    Thanks Kevin for another great contribution ... thanks Jim for your posts on this thread. I see a teachable moment after Jim's skew chart post:

    First, I want to state strenuously that I have great respect for intellectual property rights and great personal respect for Jim so in no way do I wish to step on the proprietary nature of Jim's data ... but let's face it, Jim's chart is basically just a vol ratio or beta chart like a lot of us stare at every day (for confirmation, see the image I'm posting of my own version of the 20-delta-put versus 25-delta-call ratio). Jim, if that makes you sore then please accept my humble apologies and send someone around to break my legs.

    Ok, with that over, the point I want to make is that the standard plain vanilla put-minus-call vol ratio time series is almost a tragedy because it starts with separate put-side and call-side data and then throws away the two distinct data sets by mashing them together into one. The tragedy is that the put-side and call-side skews have different dynamics and different behaviors, so why not do a tiny extra bit of work and show the whole picture?

    The chart for 2016 is a great example because the call side played such an unusually significant role during Brexit (and arguably a less-significant role than normal in February). See the image I hopefully succeeded in posting: At the February lows, the skew behavior was solidly dominated by the put-side behavior ... but during the Brexit drama the call-side became unusually volatile and displayed its own interesting dynamics for a time before becoming again highly correlated with the put-side.

    This is just one interesting and timely example. Once you get your data together to the point where you are creating these charts yourself, you are also at the point where you can tackle the whole skew with a manageable set of metrics. For example, the framework I have been testing recently attempts to describe the entire skew in terms of 5 zones: skew low to 25 delta call, 25 delta call to forward atm, forward atm to 35 delta put, 35 delta put to 20 delta put, and 20 delta put to 5 delta put. The 3 middle zones comprise the "butterfly zone" that CD members are generally most interested in. The purpose of inserting the 35 delta put is to capture the concavity/convexity behavior in the butterfly zone, which I'm convinced is very important.

    My 2 cents for today ...


    Rakesh and AKJ like this.
  12. jim

    jim Well-Known Member


    As you pointed out, understanding the entire Skew curve (i.e., strikes above and below the money) is critical. So is understanding the full term structure between the different expirations. It is only after you understand the complete volatility surface map and build accurate probability models that your hedge can be computed.

    EAB's IP is not mine to share. EAB Investment Group is an institutional risk management group that hedges hundreds of Billions of dollars of notional value for other large pension plans, hedge funds, family offices, RIA's, etc. The two-month skew graph that I had shown was the only one that they would allow me to share. They only allow me to show it because it is so basic. The quants at EAB come from Hull Derivatives and Goldman Sachs. I have a huge amount of respect for the institutional work that they have done.

    If you have something to share, I would be glad to review it.

    As for making jokes about having your legs broken... you do realize that I was born and raised in Queens and Brooklyn... and that my last name ends not in a single vowel, but in a double vowel... and that my middle name is Carmine... Right? So shouldn't make jokes about these things.

    Jim Riggio
  13. Steve S

    Steve S Well-Known Member

    Sounds like you might be a little bit irked Jim, but I hope not. I figured you would be glad that somebody popped this topic out of the IP envelope you've been properly respecting, so that now the whole community can discuss freely in the future without fear of stepping on toes.
  14. Rakesh

    Rakesh Active Member

    Steve, what are the units of the left axis in your graph?
  15. Steve S

    Steve S Well-Known Member

    Those might be called "vol ratio percents", which is how some folks including myself prefer to express vol ratios. It's just the pure vol ratio times 100, so the units would be "percent of atm vol change in vols". For example, the 20-delta put ratio of 30 would be "vol change of 30% of atm vol between strike = SPX forward and strike = 20-delta put strike."
  16. Rakesh

    Rakesh Active Member

    So if ATM strike has IV=20 and 20 delta put has IV=25, the vol ratio percent would be (25-20)/20*100 = 25% -- is that correct?
  17. Steve S

    Steve S Well-Known Member

    Correct ... it would also be common, for a single strike, to leave out the vol diff, so someone might prefer to say the 20-delta put ratio is 25/20 = 1.25 ... only difference between that approach and the diff approach is adding or subtracting 1. Used consistently, the diff approach makes more sense in a broader context because then all your ratios are closely related to actual mathematical slopes of the vol curve ... I'm supposed to post a really long boring thing next week that discusses some of this stuff, so look for that if you like geeky skew talk.
    Mark Joseph and Rakesh like this.
  18. Rakesh

    Rakesh Active Member

    Thanks Steve. Looking forward to your skew talk post!
  19. Mark Mosley

    Mark Mosley Member

    If anybody is still paying attention to this topic...this is all very interesting, especially since many options educators teach that a butterfly trade benefits from falling volatility as a rule of thumb. It seems like the opposite is the case in many instances. I realize rules of thumb may be hard to apply, but given that a butterfly may benefit from increasing volatility... can we apply a rule of thumb, like entering under low volatility might be preferable to a high volatility entry?
  20. GreenZone

    GreenZone Well-Known Member

    Mark, it isn't just about seeing the current IV or skew levels.
    It's also about your expectation for what will happen next.
    Once you can put those two together, then you can start to add an adjustment (or time entry into a new trade) which can potentially take advantage of the IV/skew changes you expect.

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