fractional spreads

Discussion in 'Options' started by Baozi, Mar 15, 2019.

  1. Baozi


    Hello everybody,

    I would like to write about a strategy that recently got my interest.

    Basically it's just a bull/bear front spread, but instead of having the classic 2short:1long ratio it is more like 9:7 or 5:3, basically something less risky than 2:1 but also less conservative than 1:1. I'm not aiming at getting it for a credit, a small debit is fine.

    I created a tool that scans all the possible combinations, and for each one spits out the risk/reward ratio and calculates the expected delta once the underlying price touches the short strike. (I assume delta of the short by then will be 0.5, and the long is derived by arbitrarily plugging in the moving average of the spot IV and half of the available days to expiration in the formula).

    I will be basically picking the best trade-off between risk/reward and high delta.
    Ideally, if the trade goes against me I will be losing a limited amount, if it goes in my direction instead, by the time it touches the short strike it will still have some steam(deltas) left. I will let it go just a bit beyond the point when delta reaches zero and starts inverting.

    Now, my questions for the veterans of this forum are the following:

    1-I will be combining this tool with entries from a trading system. I was thinking about using a mean reversion /failed breakout approach. Do you think this makes sense? Or should I use a trend following system instead?

    2-What could be the worst case scenario in this trade? I am thinking about a spike in IV, where the stock suddenly blows past the short strike and the short losses are much higher than the long gains. What else could go wrong?

    3-Which could be the best conditions to put on this kind of trade, in terms of IV rank and days to expiration?

    4-What could be a reasonable profit target?
    tommcginnis likes this.
  2. TommyR


    if i close a trade and want to recalculate the next LOWEST in an unordered list without breaking and making the whole array from scratch what is the best way:if( arraysize==1){} ,if( arraysize>1){i_0 = arraymin(X), j=i_0, arraysize(X)=N, while(j<N){X[j]=X[j+1];j+=1} arrayresize(X,N-1, 10000), i_1=arraymin(X). seems complicated. the max is easy. maybe do the maximum of a function of the array but need no zeroes/poles so 1/(x+C) or something? does that answer your question?
  3. Baozi


    There are 3 possibilities:
    1-you are trolling
    2-you are a spambot
    3-you are so much above my level that I didn't understand a single word.

    In any case however, no, that did not answer my questions..
  4. tommcginnis


    Great looking test regime you're describing here.
    I wish I had the time to follow it better, but I'm burning the candle at both ends now.
    But (FWIW) I like the questions you're posing -- in my *cursory* understanding, they sound like you're shopping for advantage, wary-but-accepting of [any] loss, and aware of a possible cascade loss. (Which is the way to rules-based harvests, right? :D )
    Hope to see more. :thumbsup::thumbsup:
  5. sle


    Just to make it clear, you are trying to find 1xN spreads that you'd like to trade? I am not certain what your evaluation criteria is (you do mention risk reward to delta, but it's unclear what your definition of "risk" is here).

    Well, think from the first principles. Your trade expresses the view that the realized distribution will be narrower than the implied distribution (that's why you are selling the N wings to buy the ats). What market environment does that imply?

    Your scariest scenario is that the stock gaps way past your breakeven. Chances are you not going to have any Vega exposure at those levels so IV is not going to matter that much. A jump in implied is going to hurt, but nowhere near as much as the stock move.

    It depends on the absolute level of vol and many other things (also, depends on the answer to my first question).
    tommcginnis likes this.
  6. Baozi


    Hi, thanks for taking the time and effort to reply.

    For the risk/reward part, I cannot evaluate the risk if the stock goes past the short strike as it is theoretically infinite, however I can take the max potential profit and divide by the cost of putting on the spread. That would be the first metric.
    The second metric is that the delta of the longs must be able to "overpower" the delta of the shorts as much as possible. I expect that the peak of the profit curve will be at zero delta, so more deltas will likely extend the breakeven point.

    These two metrics conflict with each other (if you add more shorts you reduce your cost basis, if you add more longs you add more delta), so in the end it will be a matter of choosing the best balance.

    as for the system, if I'm trading mean reversion I know that my profit potential is limited, and on the other side I want to be protected in that 25% cases when there is no reversion..
    to me it makes sense.. am I missing any elephant in the room?
    tommcginnis likes this.
  7. Baozi


    Thanks for the support sir, appreciated.
  8. sle


    Overall, you might have too many variables to solve for. My suggestion would be to fix one and then use some sort of underlying-specific metric to understand the risk. For example, make them all costless by solving for the ratio at each different strike and then try to understand your max loss by looking at the stock history.

    Well, it depends - for example, if you are planning to hold them to expiration, you max profit is pinning the short strike and that's roughly 50 delta?

    If you are trading a 1xN spread (selling the N-leg) you are assuming that the actual distribution is gonna be tighter then the implied one - that means that you are expecting mean reversion in the underlying. My point was that you want to be consistent.
    tommcginnis likes this.
  9. If you have a decently profitable system why not try it with just long calls or puts? Simplify simplify, then semper fi! Stick with it a few trades. Some will be losers of course, but some will make a decent profit, and a few will make several times the losers.
    tommcginnis likes this.