[Expectancy = (Probability of Win * Average Win) – (Probability of Loss * Average Loss) [/QUOTE] Everyone on this forum should have this tattooed on themselves somewhere visible. There seem to be so few who understand it.
Everyone on this forum should have this tattooed on themselves somewhere visible. There seem to be so few who understand it.[/QUOTE] Exactly. Which is why i started this thread. Yes credit spreads have a great looking win rate. But using the above equation their returns looks pretty bad in anything other than an unusually high vol environment
Exactly. Which is why i started this thread. Yes credit spreads have a great looking win rate. But using the above equation their returns looks pretty bad in anything other than an unusually high vol environment[/QUOTE] Since you can buy or sell them, knowing that one type has a poor return tells you the opposite will have a good return, no? (Trading costs excepted of course).
Since you can buy or sell them, knowing that one type has a poor return tells you the opposite will have a good return, no? (Trading costs excepted of course).[/QUOTE] I like your thinking. However in my experience markets are pretty efficiently priced and without putting a unique twist on things (having an edge) you end up around break even - costs
How do I do that? I don't mean the tattoo part, I mean the computation. My broker will provide me with a probability curve: probability that the stock price will be at a certain price at expiration. How do I calculate the probability of win, probability of loss and the average win and average loss? And how do I calculate those at any date prior to expiration? As an example, I sell a call credit spread: Stock at $100, sell March 17 call at $105 and buy March 17 call at $110. How do I calculate the expectancy? Thanks.
The risk-neutral implied probability provided by your broker is rather useless... You will have to figure out how to assign the probabilities yourself. As to the the "win" and "loss", you can assume that the "average win" for selling a spread will be the premium you get to keep. The "average loss", conservatively, is the distance between the strikes.
I noticed the probability curve from my broker is a lognormal curve with the default IV of ATM strikes. I can assign a different IV and it will output a different curve. If I trade OTM strikes which have different IV from theATM strike, should I input those? Or should I use HV for the duration, or a range... It is very complicated. So set that aside for the time being, In the example I wrote, I can use your definition of average win and average loss. Assuming I come up with a probability curve what probability values should I assign to the average win and average loss? Thank you very much. Your comments are very helpful.
In general, the question of how to assign your own probabilities to different outcomes is a difficult one. If this is your process, you need to come up with a way to do it without relying on mkt pricing. Alternatively, you can also go about it the other way round. Start with the mkt's probabilities and then figure out if those appear "wrong" to you. If and when this happens, you might have a trade.
ironchef, the calculations you are talking about would be based on implied volatility or historical volatility, neither of which represents the truth. The truth cannot be known, and there are many parameters involved in calculating the option price; this is why I'm encouraging you to use only 1 or maybe 2 parameters and think a bit about the UL. One of the ways to think about an edge is to decide when to trade and when not to trade. For example (this is only an example; I am making it up and not recommending this specifically), I might decide I am going to put on bullish spreads on GLD. However, I could decide to put a spread on only when GLD is above its xx-day MA, or maybe when it has a certain relative strength, and since I am talking about a bullish spread, I might also like to put the trade on when GLD is having a "down day" based on some multiple of its average true range or some oscillator or even its implied volatility. This method would likely produce a very different result from the practice of trading such spreads regularly. Whatever I decide to do would, of course, have to be back tested (making some assumptions about how option prices vary with volatility) or forward tested with paper trading over different types of conditions. In this way, I might be able to do better than chance by staying in cash at the most inopportune times. To have an expectancy better than zero, I only have to find an UL that I think I know something about, and find a signal to help me sidestep a few of the most unfavorable outcomes. If I use elements of technical analysis like this, I might confine it to only 1 or 2 simple parameters. I need to consider the time frame over which the TA works, and compare that with the number of days to expiration (or the maximum number of days I intend to hold the position). Naturally, to test this, I also need a rule about when to exit, based on profit, loss, time, or volatility. The problem is that options are not magical: they don't really buy me anything but leverage. And I don't want leverage until I can successfully predict the direction of the UL or the direction of its volatility. I find it easiest to set my rules using the UL, because there are not so many moving parts. Then, one can just use the options to light a fire under it.