We’ve had our statistics expert (Ted from the mail-room) working day and night on our algorithm for predicting the results of any fight between any two MMA fighters. He’s finished the beta version and we’ve got it running on our liquid-cooled Titan8000 super computer for the first time. As a test we’ve analyzed the pending BJ Penn vs Rory MacDonald fight (UFC on Fox 5).
Ok, well the Tital8000 isn’t fully functional, and it might not be liquid-cooled so much as sitting in our air conditioned (slightly leaky) basement…but you get the idea.
Click here to see the results.

Ha…Nice. We’ll see if they get it right. Seems to me that if BJ loses, it’ll likely be by decision (that seems to be how it happens with him historically). I haven’t seen Rory fight though, so who knows!
I don’t get it. How was the result determined? I’d be shocked if things went down that way.
This is all that is posted:
The FightersTM Prediction Engine examines career fighting statistics for each fighter to determine the most likely winner, the type of win (Decision, KO/TKO, Submission) and the winning round. Striking, takedowns and submission attempt/success probabilities are calculated using attacker statistics modified by defense capabilities (of opposing fighter). Each round is simulated using this data to calculate the probability of all fight eventualities.
Don’t know much about how odds are are calculated. Maybe they’re doing a dumbed down version of that…?
Sweet feature!