Why Probability Based Trading Is Not Working...

Part I: Credit Spreads

We’ve all heard: “If you set Probability and Expected Profit/Loss high in your trading, you will profit.” You were also likely told that this type of trading will mirror the casino business strategy, and, as everybody knows, this is a profitable model…


So then why are most traders that use probability based trading losing money? Does it mean that statistics are not working? Absolutely not! Statistics work just as well as a casino business model. What is not working is how you are applying the math to your trading.

Casinos generate numbers for their slot machines based on a normal distribution. So when Probability and Expected Profit/Loss are set up in the house’s favor, they can lose once or twice, but overall the casino will be profitable. All because the casino’s math calculations reflect the set distribution.


What are most traders using when trying to calculate probability of success? They are using a one-size-fit-all formula of Theoretical Probability that assumes Normal distribution and volatility equate to implied volatility.  


Are these assumptions right?  In most cases they are not. Individual underlying assets could have a historical distribution that does not resemble the normal distribution, and an implied volatility that is reflecting the market view could also be inaccurate.

To overcome the common challenges, consider the new methodology of probability calculation and also an introduction of two new Probabilities:

1. Historical Probability

Takes into consideration the specificity of each underlying historical behavior


2. Stress Test Probability

Uses magnitude of the historical moves and represents the capability of the underlying to perform certain moves without taking into consideration direction of that move


This methodology allows using specificity of each underlying asset. Like one medical doctor said: “You have to treat patients based on their personal medical history and not on an average temperature in the hospital…”


To prove this point we used Ez Trade Builder – a product that allows you to select a day in a history and then search through different option strategies, using Probability as filtering criteria.


To find out which probability provides the best results, it’s best to compare the Stress Test Probability to the One-Size-Fits-All formula (Theoretical Probability). In our first search we will set Theoretical Probability to be high and at the same time greater than Stress Test Probability.

Image A: Report courtesy of EzTrade Builder - Product of EzTrade, Inc.


As can be seen from the table (Image A) Theoretical probability of profit (“Prob Profit”) for these trades is greater than 77%. This is not a bad probability to consider for a trade… But the results are disastrous! Eight out of ten trades are at significant losses.


Could these results seed doubts of the validity of using Normal Distribution and Implied Volatility in Probability calculations? Absolutely.


The previous search was based on an analysis of data from one date... What if we want to test these criteria for a period of time? We can do this using Track Record report. It allows viewing trade signals, generated by filtering criteria, during a certain period of time.


Image B: Report courtesy of EzTrade Builder - Product of EzTrade, Inc. View full report here


Take a look at the two columns on this report (Image B):

1.      Best Profit/Loss – Results depict the best exit. These results have a low likelihood of being achievable… But provide a benchmark for understanding trade management.

2.      Last Day Profit/Loss – Results depict the easy exit. These results are the ones that traders can achieve by setting the trade and “do nothing” till expiration.


The yellow row represents summary results for the analyzing period. In our example, if you wait till expiration, results would be a loss of $40,189. Not good… It is a loss of almost your whole initial account of $50,000.


Why did this happen? Scroll to row 8 using the full report. The full report will allow you to see much more information, plus you can repeat the analysis for other trades on this report.


You can see that this trade is a loss of $3,300. Let’s investigate this trade further. Click on symbol FCX and you will see the Detail Report.


Image C: Report courtesy of EzTrade Builder - Product of EzTrade, Inc. 


As can be seen from the above table (Image C), theoretical probability is 81.89%. I think if someone offered you to get into the trade that has almost 82% of success; most traders would take it… and lose $3,300.




Take a look at Stress Test (ST) probabilities, calculated based on FCX 1 Year – 5 Years history. They are around 40%, which is even less than the probability of guessing heads or tails. Now, with this new information about probability of success for this trade, would you still consider taking it?


Let’s drill further and try to understand what has happened. Scroll down on the Detail Report and take a look at Volatility table.

Image D: Report courtesy of EzTrade Builder - Product of EzTrade, Inc.


Implied Volatility percentile is very low - only 12%. This means the market anticipates this stock will not move a lot in a near future. This also reflects on the superficially high theoretical probability.


Results – market “wisdom” is wrong, the stock moved 15 points and the trade lost over $3000. If you do a similar analysis of the losing trades on this report, you will see how important is to watch the Stress Test probability.


When you are analyzing the trade - you want to “Stress Test” it. In other words, find out what is the probability of success, if the stock moves against you.


This approach of calculating the Stress Test probability gives you the opportunity to avoid costly mistakes.


Finally I want to look at the Put Credit Spread Track Record Report where filtering parameters are set the way that allows you to take advantage of Stress Test probability over Theoretical probability. This report lists the trades for the period from 2008 till September expiration of 2013. Now take a look at the results for 2011, you will find a profit of almost $28,000. Compare this with the loss of $40,000 in our previous Track Record Report and you will see the power of using Stress Test probability.


Continue reading Part II - Short Puts.