Unusual kurtosis

Discussion in 'Options' started by nitro, Mar 15, 2010.

  1. nitro


    I see lots of unusually high kurtosis in stocks that I trade, about 200 of them.

    If you recall, the center of a probability distribution is the most likely outcome. Kurtosis measures the height of this curve. The taller the curve, the more likely the mean (percent) changes are likely to occur.

    This is probably obvious to most that trade, since all we have seen is the market go higher by relatively small percent changes for what seems like an eternity, generally leading to more peaked curves.

    If the risk neutral distribution has a much higher kurtosis than the real
    distribution, one possible trade is to sell OTM options (call and put) and buy ATM or near-the-money (NTM) options. So a long butterfly centered around the ATM would be a kurtosis trade if you believe this thesis, or short a butterfly if you don't (if you believe the curve will flatten), or no kurtosis trade at all if the real and risk neutral coincide.

    Do you risk it with four days to expiration and an FOMC day in between? Either way, the shape of the real distribution is probably flatter. There is an enormous amount of complacency in the market here, more than I can remember in years.
  2. nitro


    The problem is the following. You go to the horse races. With 30 seconds to race start, the public is saying (through parimutuel betting) that a horse whose real probability is say 8:1, is paying 17:1. This is a great bet if this race were run 100 times. The problem is, you have to wait several times on average to see a winner, but when it pays, it pays for all your losers, and then some.

    But on any one run of this race, you are overwhelmingly a loser.
  3. It's called an overlay, and the only way I bet the horses. However, 8 to 17 is the clsoe to the same as 4 to 8, which happens quite often. If you start with a longshot to begin with, then as you've rightly pointed out you won't win with a high frequency.

    But if you start with a 2nd or 3rd place expectation, and then get additional odds, your win rate will improve.
  4. The classic "overlay" that you describe is an opportunity and a great bet if your determination that the real probability is 8:1 is an accurate assessment. And since that is the hypothesis I accept it as fact.

    The fact that you need to bet scores (if not hundreds) of overlays to be sure of showing a return is no different than many business propositions that require you to view the return from your advertising or whatever over a longer period than a single day.

    Nitro, as you know, it is risking too much on each of the bets that make up the series that knocks the typical plunger out of the box not the "problem" of needing the multiple plays. If you are able to handicap well enough and manage your capital rationally you win.

    I agree with you 100% about the level of complacency. It is spooky to see the world breathing so easy. It is almost as if the great bull run we have just experienced (and are still experiencing) is proof positive that all is well as opposed a great run that may or may not hold.

  5. Higher than normal kurtosis (leptokurtosis) means fat tails. You could take the Black Swan approach with this. Buy treasuries, and use the interest income to buy OTM calls and/or puts.
  6. this unusual kurtosis, do you see it only in march options? what about april?
  7. The resident expert on all things horse related is Mark Brown, who by his own admission has the winner in every race. Maybe he'll be kind enough to pass. I meant pass on his wisdom.

    One problem with high odds overlays are the long and tedious losing streaks. It requires a much smaller 'bet' to avoid the probability of ruin.
  8. "Never bet on a long shot."

    - Frank "Lefty" Rosenthal
  9. Where the “real odds of 8:1” are comes from? Unless one uses another [larger] poll’s odds like simulcast or exotic. Do you have something similar in option?
  10. "Kurtosis" would refer to the distribution of <i>realized</i> stock price-changes over some lookback. But OP might be confusing it with (options) skew. Hard to know, unless OP posts some data to clarify what exactly he's looking at.
    #10     Mar 16, 2010