Hi everybody, I joined a while ago, but this is my first thread. The probability of profit numbers shown on tastytrade look nice, but here is how I look a typical trade. The numbers are real, from a real option chain. SPY options for coming January: 252 / 250 puts have delta = 0.16/0.14 Sell 252 put for $2.31 Buy 250 put for $2.06 Credit $0.25 Max profit = $25 per contract Max loss = $200 per contract Based on delta, probability of 252 put expiring ITM is 0.16, therefore probability of making max profit is 1-0.16 = 0.84 Then probability of 250 put expiring ITM = 0.14, therefore probability of max loss is 0.14 In the long run average profit looks like: 25*0.84 – 200*0.14 = 21 – 28 = -7 I’m obviously doing something wrong here, but how can you profit in the long run when selling put spreads like this? Any comment is appreciated. Thanks a lot! Corto
1. you automatically have a negative edge crossing bid-ask spread. 2. Why do you expect to make money? You have not give any reason why on average you should be profiting. (a.k.a. +ev trade) 3. Do not trust any "probability" number the program tell you.
Hey Corto, thats awesome you are looking at trades from an expected value point of view. You are way ahead of most of the tasty traders. First thing I would say is stop watching them. Tasty trade does a great job of getting beginners excited about the options market BUT thats about as far as your journey with them should go. Something you should know is that your delta changes with implied vol. So your first assumption should be, is vol over or under priced. You also need to look at the skew because you are trading a long and a short option. So is skew over or under priced? You are getting a negative expected value, because the deeper otm options you are buying, have a higher implied vol than the closer to the money.
Option delta is an estimate based on a pricing model. There are a few different pricing models that will give you different values. What this means is you can't calculate probability/profitability from the numbers that is based an opinion subjective model. You're looking at it the wrong way. You can't calculate your way to profits on paper. The market is not dumb. Like horse punting, the odds are totalled to negative and most people lose money while a small minority of pro punters make money. Like picking the right horse, you have to pick the right analysis on the underlying. And as mentioned above, do not use delta as a measure of probability of profit. And probably stop watching Tastytrade beyond the basics.
Corto: your question was on your method, and secondarily, on 'how can [you] make money. Your method is off because expectancy is based on covering an entire event space -- i.e., all probabilities sum to 1.00 -- but you've got a spread in there, with a P(notITM) = 0.84, and a P(ITM) of 0.14 -- which don't sum to 1.0. (Thus, a 2% range of Great Joy to Maximal Suffering. Egads. ) So, your (trivial) error may easily *understate* loss potential. (As far as not using deltas as a measure of probability -- while algebraically correct to note that there is no direct connection, with 2 rather insignificant assumptions, the equivalence proof is trivial. There is at least one YouTube video on it (MIT? Stanford?, which is actually fun to watch. Seriously. ) Now, methods-wise, we can argue all day about the particulars of making the expectancy calculations operational, but in selling credit spreads, facing $1 worth of gain against $10 worth of loss is an everyday occurrence. (And so, it is MUCH MORE important to have Take Profit and Stop Loss rules in place before entry is made, versus a long-short underlying trade.) So, in taking a look at the regular ol' Expectancy Calculations, would you EVER let a loss go to 2x-3x-4x-10x of anticipated gains, for a long/short trade? EVER? Your expectancy calculation assumes the spread is held to expiration: but with appropriate attention, you would never hold it that long -- so when do you exit an option spread? What gain-%? What loss-%? And now you're off to the land of P(Touch) == which is roughly twice that of P(ITM). Uh-ohhhh.
Tommcginnis, you're right, I try to close the trade when the loss gets too high, but Tom Sosnoff keeps saying that you don't need to do that with limited risk strategies. In his own words "you need to let the probabilities to play out". Sometimes I did nothing and waited and it worked, sometimes I closed the trade at a loss to only see the stock recover few days later. I'm sure I don't tell you news here. Anyway my thought was that the strategy works and maybe my probability of profit calculation is not right.
That's the part I liked.. Y'know, we have all these fancy pants formulas, got everything just right except one thing, how much fear and how much greed tomorrow. We call it the implied volatility, we know it's value yesterday, we don't know it's value tomorrow. As long as they let us trade yesterdays options we'll be just fine.
I'm reminded of that aphorism about 'when the tide goes out, we find out who's got a bathing suit and who doesn't.....' In any event, those who price options perforce reveal their estimates of adverse events. We pick those probabilities up by backing out the volatility implied by those prices. It doesn't mean those things will *come*off* as the pricing predicted, but only what the a'priori fears were. So, although we make claims about be "directionally-neutral" and all that, there are still competing theses of Trend-Following versus Mean-Reversion in trade after trade. Interesting juxtaposition.....
Go to this show as one of many examples of Tastytrade's "go to strategies": https://www.tastytrade.com/tt/shows/market-measures/episodes/the-power-of-the-put-06-27-2018. Its a short SPY only study. First, Google what an equity curve is, then look at the equity curve in the slide titled ""We observe that performance of an ATM short put was nearly identical to that of a long stock..." That picture says it all. Their "go to" strategy is almost identical to simply long the SPY. By the way, the strangle version is too. They advocate levering this up by the way so that the 'equity draw-down" (another critically important term to look up) goes beyond the -50%!!!! you would have experienced in the non-levered, base case version. AWFUL strategy. Professionals would gag. With a 30% use of buying power reduction (which they recommend). This would have a 75% max equity drawdown during the 2008-2009 crisis.
Using statistics in data analysis can be very tricky. This is why a good idea is to spend some time to fully understand the model and especially its outputs. Then it is easier to make logical conclusion if its outpus can be used for the purpose of profit caluclation.