If you had to choose between VRP and TRP?

Discussion in 'Options' started by .sigma, Jan 30, 2020.

  1. .sigma

    .sigma

    If you had to choose which short vol premium to collect, which would you choose:

    Variance Risk Premium

    or

    Term Risk Premium
     
  2. .sigma

    .sigma

    To elucidate:

    VRP captures the difference between the near IV and realized IV (theta/gamma)

    TRP captures the difference between far IV and near IV (vol-rolldown)

    Answer as if they are equally easy to "harvest"

    Which do you prefer?
     
  3. samuel11

    samuel11

    Are you the source of that question?

     
    .sigma likes this.
  4. .sigma

    .sigma

    I'm in that thread yes and I speak with Squeeze about these topics. He's the one who introduced me to dark pools and gamma risk.
     
  5. TheBigShort

    TheBigShort

    "collect" is the not the right term to use.

    Plot the IVOL-RVOL spread (lagged RVOL 20 days) over the last 10 years. The mean is less than 1 point. After commissions, taxes and sub par hedging you would be lucky to come out with a profit. Data is provided by ORATS. RVOL is 20 bus day tick data. Iv is 20 bus day IVOL. Mean spread is 0.80 points over the last decade

    The right phrase should be - "which risk premium do you like to work your ass off for...".


    Screen Shot 2020-01-30 at 9.53.11 PM.png

    Screen Shot 2020-01-30 at 9.53.23 PM.png

    I saw a recent video that stated the cost of gamma is more expensive than vega. But I cant seem to remember what it was called.
     
    .sigma and TooEffingOld like this.
  6. traider

    traider

    Does this imply that all the VRP studies are outdated? There is no longer much VRP
     
  7. TheBigShort

    TheBigShort

    There is still a risk premium, .80 is not nothing. Just don't think you can start selling 30 day straddles, have a mediocre hedging strategy, pay your taxes and walk away doing better than buy and hold the S&P.

    If you were selling a variance swap you would do much better because its priced closer to a 90% moneyness strike. The problem with options is that you have gamma. So if you sell the 90% moneyness options you will realize most of your PnL (gamma gain/loss) when realized vol is greater than what was implied.
     
    tommcginnis and .sigma like this.
  8. .sigma

    .sigma

    Sinclair said something,

    "It seems sensible that spreads will be larger in situations where option sellers are more risk averse: They will demand more of a premium to compensate for their fear. It also seems sensible that spreads will widen with higher moments: All options are exposed to all moments of the physical distribution.
    However, the model doesn't seem to fit with the fact that, at least for the SPX, spreads (in percentage terms) actually narrow as the Volatility Index (VIX) increases."

    Is this because as spot drops, vol increase, thus the spread between implied vol and physical vol tightens because you're late, the hurricane has hit, the car has crash, the deductible is being processed. But when markets are quieter and ATR is smaller, this risk premium is there, even if it's not much.
     
    tommcginnis likes this.
  9. .sigma

    .sigma

    The Vol premium as a function of VIX levels

    VIX < 20
    Days: 3,102
    Avg spread: 27.75%

    20 < VIX < 30
    Days: 1,925
    Avg spread: 19.78%

    30 < VIX < 40
    Days: 407
    Avg spread: 13.73%

    40 < VIX < 50
    Days: 111
    Avg spread: 9.26%

    50 < VIX
    Days: 56
    Avg spread: -11.52%
     
  10. TheBigShort

    TheBigShort

    There are quite a few things wrong with using these numbers.

    VIX like the Varswap is closer to the 90% moneyness strike than the ATM strike so those numbers are already biased high. As a retail trader it's best to compare the ATM strike as you are trading options and can not trade the varswaps or the vix cash. You can try and replicate a varswap by selling a handful of strikes on the S&P but I think you give up a lot of alpha moving in and out.

    The second problem is the data points you have for each case are drastically different. You cant compare selling VIX at 50 when you only have 50 data points vs under 20 when you have over 3k.

    Lastly, its better to use vol points rather than % here. The vix is mean reverting and stays in a range. Since it does not trend it's better to use a fixed number (points rather than %).

    In conclusion there is a variance risk premium. Its not easy to extract and its not going to make you 20% CAGR after taxes and transaction costs. If you couple it with the S&P drift ie selling puts and not delta hedging. You might be able to do better but still not amazing.

    With that said you can create timing models. When is vol rich vs cheap. When do the tails get fatter etc.. But now you're not simply trying to collect the VRP.
     
    #10     Jan 31, 2020
    .sigma and tommcginnis like this.