Here's a thought i had while reading the article about the punter who cracked the gambler's code on Bloomberg. In it, he mentioned using the bookie odds as a starting point and refining them, which i suppose is to identify any mispricing. Screenshot of that portion of the article:
In my earlier article on tools to analyze matches, i mentioned that sofascore has a section that shows how frequently a team clears the current odds that the bookies have on them to win. If you haven't seen it, it looks something like this:
If these stats are real, then betting on Boston River seems like a good deal. The two caveats being:
These set of predictions have a few simple rules. If all are met, i'll add them to the list:
Here's the list for tonight till Saturday afternoon:
| Match | Prediction | Odds | Outcome |
|---|---|---|---|
| Boston River vs Jaguares | Boston River | 2.04 | W |
| Tijuana vs Toluca | Tijuana | 2.29 | W |
| Rudes vs Belupo | Rudes | 2.6 | W |
| JEF Chiba vs Albirex Niigata | JEF | 2.04 |
My first reaction when preparing the list was that the 25% filter seems to weed out a lot of matches. Thinking of relaxing that to 15%. Although, if we get a win rate of 75% or more from this sample, i'll continue monitoring with a strict criteria. After all, the first rule of making money is to not lose money!
Odds shift over time and as that happens, the probability of a team clearing those new odds shift as well. This snapshot is compiled with up to a 36 hour lead time, so it would be interesting to see if there's a major shift in the odds before kick-off.
Just an observation - 2 teams with probability to clear odds less than 30% tend to draw. Something to think about when building a more comprehensive model.
Time to tackle 
Update 1:
3 wins out of 3 so far. Seems like there is some value in keeping to a strict filter.
The odds for the last match have changed and they don't fit the criteria anymore. Maybe the call has to be made with less than 24 hours before kick off. Either way, let's see how it does.
