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

Correlation between steepness of Volatility Skew and Butterfly prices

Discussion in 'Options' started by SVL, Apr 14, 2016.

  1. SVL

    SVL Well-Known Member

    Volatility skew is much steeper and the price of ATM 50 pts butterflies is much higher today than 30 days ago.
    Knowing this is great but I am still not 100 % sure how do you materially benefit from this knowledge when trading JL M3 strategy. I only sense it in a way that the risk has increased exponentially. Especially it is true for new trades for June expiration to be opened in this volatility environment. If RUT goes south, not only the value of your DIMC will drop but also the value of butterflies will drop due to an increased volatility and a flatter skew. Is there is a certain $RVX level where you should be more cautious and trade a smaller size with M3 strategy.

    Please see below the prices of butterflies with 50 pts wings for May 16 expiration:
    2016-04-14 RUT May 16 50 pts wings butterfly prices.png
    Please see below the volatility skew for RUT May 16 expiration for today and 2 weeks ago:
    2014-04-14 Volatility Skew RUT May 16.png
    Please see below the volatility skew for RUT May 16 expiration for today and 1 month ago:
    2014-04-14 Volatility Skew RUT May 16 today  and 30 days ago.png
     
    Murphy Tan, N N and Gail like this.
  2. DavidF

    DavidF Well-Known Member

    Hi SVL,
    In a very low vol environment, the best methods I found (backtesting and more recently live) the M3 to prevent being hit by a major spike in volatility whilst not sacrificing much sideways or upside are:-

    (a) pick a DITM call with close to zero intrinsic and buying a $5 put (same total extrinsic)
    (b) in the same way that some traders like to switch from an M3 to a BWB in high vol, I´ve found a reverse BWB, i.e, pulling the lowest leg up and compensating for the increase in neg. deltas with a call outperforms a standard M3 in a sharp down move. The spike in vega in the lowest leg seems to perform better than the increased vega in the call. It also outperforms to the upside, but gives up some if all stays flat (obviously paying more time premium).

    I find both methods provide even more resilience than a standard M3, and perform very well in vol. spikes.
     
    Capt Hobbes likes this.
  3. vega4mike

    vega4mike Well-Known Member

    For a comparison look at the RVX levels in 2013, trading in the low teens, as I recall, any fly strategy with RUT was very challenging. But we are not yet at those levels. My personal view is that it we probably wont get down to those levels (in the absence of a fed put, money printing presses), as the mkt would still like to keep some perspective on potential risks, hence, the lower ranges of the RVX and VIX will be higher, at least that's what I've observed so far this year. The RVX or VIX hits a certain low point and then even if the mkt continues to rally, the vol indexes stop falling.
    Also note how the VIX futures premium remains stubbornly high with a mkt rally...the futures crowd not necessarily buying into the mkt rally, of course this doesn't mean that they are right, but just another pointer to the level of uncertainty in the mkt place, which provides a floor to how low the vol indexes can fall.
     
    JohnP likes this.
  4. Paul Demers

    Paul Demers Well-Known Member

    Regarding the spreadsheet I am seeing two variables that are affecting price: IV and Decay. I think that to isolate the price difference based on IV you would need to compare butterflies at the same DTE with different IV levels to get a better picture. With trades I have done in the past in RUT I have a threshold of 35 RVX where I would not initiate a new trade.

    Hope this helps
    Paul
     
  5. SVL

    SVL Well-Known Member

    David,
    Thanks for the good suggestion with a protective put. I also found the excellent discussion on JL forum (NUANCES OF M3 #2 (ITM Call) where you explained in detail how it works. I like this idea and will back-test for Aug 11, Oct 14, Aug 15 cycles.
    Sergei.
     
  6. SVL

    SVL Well-Known Member

    Paul,
    Mar 16 RUT 37 DTE IV 28.3 ATM Butterfly price $11.25
    Apr 16 RUT 37 DTE IV 21.2 ATM Butterfly price $13.50
    May 16 RUT 37 DTE IV 15.7 ATM Butterfly price $17.50

    My question was related to a lower threshold of RVX where you simply are not paid enough for a risk taken.
    I think you meant 15 RVX, as 35 RVX is an excellent level for butterfly trades as your entry price is so low.

    Sergei.
     
    Last edited: Apr 15, 2016
  7. SVL

    SVL Well-Known Member

    Mike,
    When I back traded M3, I lost money in most of 2013 cycles . Now I understand exactly why.
    On Feb 6, 2013 RUT was 909.20 with ATM IV 15.4.
    The price of 860P/910P/960P butterfly for Mar 13 expiration was 21.80 with 37 DTE.
    You are doomed from the start to fail with M3 strategy when BF prices are so high.
    Sergei.
     
    Last edited: Apr 15, 2016
  8. DavidF

    DavidF Well-Known Member

    Hi Sergei,

    On a related note I listened to Jim Riggio talk about the institutional tools (mostly vol. skews) he uses to ascertain what trade and when based on IV skew. As you may know he favours a BWB set-up in higher IV conditions and a standard M3 set-up in lower IV conditions. He also mentioned the superiority of OV´s P/L curves versus ONE and TOS in predicting neg. effects of a downward move.

    Which gets me wondering:- if the OV P/L curve is relatively accurate, do we ever actually really need to look at IV numbers, skews etc., eventing is already built into OV´s model? In a high IV environment the BWB will provide a better P/L graph than an M3 and vice versa.
     
  9. Paul Demers

    Paul Demers Well-Known Member

    You are absolutely correct, my apologies for misunderstanding. I have a completely different stance on how I interpret the relevance of an elevated RVX.
     
  10. SVL

    SVL Well-Known Member

    Hi David,
    Can you please provide a link to the discussion where Jim Riggio talked about the institutional tools for volatility skews.
    I think it would be better for a more experienced traders like Kevin Lee to provide more insight about the accuracy of any Option Analytics software. From Kevin's interview with John Locke I remember that they discussed this topic in a great detail and said that no software available to a retail trader could accurately predict the volatility skew effect. The whole fiasco with OV last year was caused by their attempt to try to "accurately" model volatility changes along different strikes. I think interview with Kevin Lee will be posted next week on JL's free YouTube channel.
     
  11. SVL

    SVL Well-Known Member

    The below is the summary of prices of 50 wings ATM put butterflies with 37 DTE for last 39 expiration cycles.
    Average values are: RUT 1117.04, ATM IV 18.17, BF price 15.86.
    It gives a pretty good idea how butterfly prices fluctuate depending on volatility and how it compares with average values :
    2016-04-15  37DTE BF prices.png
     
    Last edited: Apr 15, 2016
    Rakesh, N N and DavidF like this.
  12. DavidF

    DavidF Well-Known Member

    Hi again Sergei,
    Jim Riggio talked about it in his Kevlar update presentation (under classes, very last presentation), you may not have access to it?

    I
     
  13. SVL

    SVL Well-Known Member

    David ,
    Thanks. Unfortunately, I am not a subscriber of Jim's alert service and I have no access to it.
     
  14. vega4mike

    vega4mike Well-Known Member

    its accessible from the members ->Library, if you are a member. Video was posted yesterday.
     
  15. SVL

    SVL Well-Known Member

    Mike, thanks. I found it and will watch it over the weekend.
     
  16. SVL

    SVL Well-Known Member

    I watched Jim Riggio's presentation and really like how he could explain very complex things with such a simplicity.
     
  17. vega4mike

    vega4mike Well-Known Member

    Hi Sergei,
    While 2013 may have been tough for RUT Flys trades, I think JL with his M3 & Rock trades still did ok - weekly updates. For the M3 he recorded 22.2% gain for the year & for the Rock a 177.7% gain for the yr. The worst return was for the Bearish fly. So you may want to go back over those updates to see how he coped with the very low IV. The figures taken from JL website.
     
  18. Paul Demers

    Paul Demers Well-Known Member

    Today the majority of hedge funds would kill for a 22% return
     
  19. tom

    tom Administrator Staff Member

    If you missed Jim Riggio's review of the Kevlar April 2016 trade, you can watch it here:

     
  20. Kevin Lee

    Kevin Lee Well-Known Member

    Hi SVL,

    Thanx for sharing the data. It's a lot of hard work. Let me comment on the following :

    1. How to benefit from the butterfly price table ? - I use it to help my entry and exit. After you have done this enough time, you will roughly know what butterfly should cost for similar IV and DTE. This info will help to avoid or to take advantage of temporary butterfly mis-pricing.

    2. How does IV skew affect butterflies ? IV skew affects both prices and greeks but delta is specifically important in trading M3. Firstly we must understand two aspects of the IV skew - (a) the slope and (a) the shape. Look at the IV curve - the left is roughly linear and then about 40 to 50 pts above ATM, it starts to curve upwards.

    upload_2016-4-18_2-55-59.png

    The slope of the linear part typically is steeper when : (a) IV is lower, (b) DTE is smaller.
    The lowest point on the curve gets closer to ATM as DTE becomes smaller.

    As market moves up, IV skew becomes steeper, strikes above the market will suffer a bigger IV crush than strikes below the market. Therefore, knowing where the right leg of the butterfly is on the IV curve is important.

    Typically after a market sell off, IV skew will be relatively flatter. For instance, after a sell off, if RUT is about 10 to 20 pts below short strike of the butterfly, then the right leg will likely be at the point where IV crush will be very severe when the market rebound. In this case, the T+0 line will be inaccurate. It'll give us a false sense of safety on the upside. That means even if you see a flat T+0 line to the upside, it's fake. Once the IV crush on the right leg, the T+0 line will sink all of a sudden. That's because when IV crush, delta becomes much more negative and the P&L suffers more than expected. To mitigate this effect, we need to hedge the upside much more aggressively or pull back the right leg closer, ie turning a regular butterfly to a BWB, or alternatively hedge with more positive that what the software shows.

    So we must not think that the lowering of IV is always good for butterfly. That is only true if the IVs of all the legs go down simultaneously. In the real world, that will never happen. As a result, we must not take at face value any of the greeks we see on an analyzer, although OV will be much closer to truth than TOS or ONE. Instead, we must be aware of where the legs of the butterfly are on the IV curve, understand how individual leg IV will behave and then mentally compensate the greeks.

    3. Why are the analysis software inaccurate ? The main reason is they don't model the changes in IV accurately. There are three dynamics that need to be taken into consideration : (1) horizontal shift, (2) vertical shift and (3) slope shift. Let me use an example to illustrate. When market moves up, firstly the entire IV curve is shifted horizontally. Example - if current RUT Is 1000, ATM IV is 20%. RUT moves to 1020. The IV curve will first shift horizontally such that IV @ 1020 is 20%. After this shift the original strike of 1000 actually shows an increase in IV ! See below L1 to L2.
    upload_2016-4-18_2-10-32.png

    Then step two - vertical shift down. We know that IV goes down as market moves up. Example below - from L2 to L3. So this effect lowers the IV for every strike, countering the increase in IV in step 1. Depends on which effect is bigger, individual IV might end up lower or higher.

    upload_2016-4-18_2-13-8.png

    Then step 3 - shift in IV skew. In the case of a reduction of over IV, the IV curve steepens. Example below from L3 to L4. After this clockwise shift in IV curve, some of the lower strike IVs might end up higher than original (ie L4 vs L1).

    upload_2016-4-18_2-14-34.png

    To accurately model option pricing, software needs to model all 3 shifts. OV takes into consideration step 1 and step 2 via its CEV. But it does not take into consideration step 3, ie the steepening or flattening of IV skew. TOS and ONE are worse. They do not incorporate any of the shifts. They assume a static IV curve.

    To forecast future IV accurately is not a trivial task. The horizontal, vertical and slope shifts are different for each underlying and these shifts also depend on other factors such as absolute IV level, DTE and market outlook. To model this accurately, the software must start with an accurate regression analysis of the past data - at a minimum incorporating underlying, DTE and IV level. Therefore, my opinion is that the CEV setting must never be a manually inputted number like in ONE.

    4. Why did OV's modeling create a problem few months ago ? It started with OV trying to do a better curve fit. The old projected IV model was based on linear regression. It did not take into consideration of the non-linear part of the IV skew (ie the curving up part). The new model tried to achieve a better curve fit by incorporating the non-linear part. The irony is OV actually succeeded in a closer curve fit, but because OV doesn't take into consideration of the IV slope change, the net result is the upper strikes IV became way over projected and the overall IV curve became too flat. That created all the problems in T+0 and deltas that we all experienced. Now, they reverted back to a "worse" curve fit but nevertheless a more accurate T+n lines.

    What OV needs is an accurate modeling of the shift in IV slope. If that can be achieved, then the better regression model will work. But I know this is really difficult. I hope OV succeeds so we can all benefit.
     

    Attached Files:

    Last edited: Apr 18, 2016
    ygu, Rakesh, Aditya and 16 others like this.

Share This Page