'On what principle is it that with nothing but improvement behind us, we are to expect nothing but deterioration before us?'
Lord Thomas Babington Macaulay
In sum, we assign each fight a score - the higher the score, the more likely the fight will finish before the allotted time (a finish). Various fighter metrics and variables combine to calculate the score. Using an increasingly large dataset of fighter histories, we rank every fight with our calculated score. A bell curve of data identifies the mean - fights on the higher side of the mean have increasingly higher chances of finishing before the allotted time. Each fight is given a corresponding implied probability of outcome. We compare that probability with the odds on offer. Through trial and error, and large-data analysis, we have discovered the types of fights and corresponding odds ranges that return a profit in the long run.
Here are some examples of how it works.
Deiveson Figueiredo vs Brandon Moreno, UFC 256
This was a main event with 5 rounds between two hard-hitting opponents. Bookies measured the likelihood of the fight not going to a decision as very high. The crowd agreed and odds for the fight going to a decision closed at 3.25 (+225), which translates to 30% implied probability. But our model calculated the chance of a decision at 56%, which is a significant margin over the bookie. We took the bet with a 9 unit stake. The fight was a decision draw and we profited 20 units.
Rachael Ostovich vs Gina Mazany, UFC Vegas 15
Female fighters are more likely to go to decision. In this case however our calculated outcome was 68% for a finish. Bookies disagreed and offered odds 2.60 (+160), with a 38% likelihood. The fight ended in the 3rd round. Our 8 unit wager earnt us nearly 13 units profit.
On so many occasions our model has put our money in difficult places. We've lost plenty, but not more than we've won. Counter-intuitive wagering, while using a strict proportional staking method, is the kind of discipline in gambling that yields long-term profit.