Statistics CAN Beat the Vegas Market, But How?

Why it can be done?

The Vegas spread market is indeed just a market like the stock market. The “price” of the game is its spread line, but their returns are far better when played intelligently. In this market, the efficient market hypothesis is also assumed; this implies that the spread of the game is already affected by all known information. In other words, according to this hypothesis, it would be impossible to consistently profit using any information that the people already know (including stats, injuries, and trends among many).

Supposedly, this market is indeed efficient. One scientist “proved” it by testing that the average difference between the Vegas spread and the actual spread on ALL games in a time frame of 10 years is NOT statistically different from zero. But what if you can systematically and consistently pick games whose difference is significantly bigger (or lesser) than zero?

If people like Warren Buffet were able to find stocks that were undervalued, why can’t someone find games that are under/over-valued? Buffet studied thousands of stock reports and financial statements to find companies he considered solid and undervalued. If one studies and analyzes thousands of games (years and years), wouldn’t one find situations in which games are undervalued? But how can one go back and study these many games? Luckily, computers can aid us in this research.

How it can be done?

There are 3 aspects of any strategy one must keep in mind in order to be successful. Buffet and his teacher Benjamin Graham followed them, the MIT guys from the book "Bringing Down the House" (Now a movie titled “21”) followed it to profit from Blackjack, and many financial managers keep them in mind when tough decisions need to be made:

1. Create a system – Create a system that exploits the disadvantages of the market, test its success, and stick to it. There are an infinite number of ways to look at the overwhelming available information. Find one that has consistently performed well over the years, keep it a secret, and profit from it.
2. Analyze – Remove the guesswork and do your homework. Analyzing every game of the week takes a long time, but if it was easy to do, efficient market theory would have caught up with you. Use as much information available, filter out the noise, and stick to the facts. Statistics do not lie, but they can be easily misinterpreted.
3. Be psychologically composed – Creating a system not only involves how you will pick teams, but also how you will develop an investing plan. If you’re system contradicts everybody else’s opinion, why should you care? Stick to your plan. If your bankroll starts to diminish, do not worry, luck is not always on one’s side. But in the long run, you trust your tested system that will give you consistent profits. I go into more detail on this previous post.

How I do it?

I just told you, but I will elaborate (just a little).
1. Create a system – My system is based on 2 things:

a. A statistical model - I created a statistical model that predicts spread outcomes and finds under/overvalued games based on team trends, offensive, and defensive stats. Then, I back-tested predictions to estimate how they would have fared in 10-years. The models were tweaked until back-testing returned above 60% accuracy on average for 10 years.
Three important qualities of the model:
a) Dynamic – Model adjusts weekly to new game results
b) Weighted – Model weighs recent games more heavily
c) Robust – Model is not sensitive to outliers

b. A confidence measure – Each game is evaluated through a prediction confidence measure. This percentage confidence measure is highly correlated with prediction accuracy. When a game surpasses a specified threshold, only then is the game considered for a pick. The game is picked only if no key players have been recently replaced (injured).

2. Analyze – I have done my homework. It has taken me more than two years to build this. From gathering the data, researching all available literature, and finally putting it all together to build this decision tool that has so far given me 59% and 61% ATS in the past two years.

3. Psychological composure – This is the hardest part, but I am learning more and more as I am put into difficult situations. Every time that I have failed to follow the system, I lose. Whether it is by not placing a bet I should have or betting on a game I shouldn’t have. Live and learn, composure is the key to my success.

Is it Sunday yet?


rmp78701 said…
Hi Jaime,

I would like to create a betting system, but I cannot seem to find a source for historical data. Could you share your data or let me know where I might obtain it?