Data Scientist Previews 50/50 Fights
Ulberg vs. Nchukwi
This is the closest matchup of the weekend with Ulberg at -120 favorite and Nchukwi at a +100 underdog. Overall, not including the 4.348% juice that the sportsbook makes on the fight, the probability is 52% to 48% respectively.
Tafon Nchukwi is 1-1 against fighters with 23 minutes or less for Total Time in Octagon. He is 0-2 against fighters with 95 or less for Opp Sig. Str. Per KnockDown.
I found these stats to be intriguing, but I was unable to really make a conclusion yet. A fighter who struggles against opponents with little experience in the UFC and fighters with no chin is odd. Is Nchukwi over swinging in the fight? Is he overconfident and then loses the fight by not being patient?
Using Bayesian Simulation and not machine learning models, BCDC is 51% sure that Carlos Ulberg is the better fighter with a risk of being wrong at 24%. This comes out to around-120 odds, which is about the same price as the sportsbooks. This gives no real edge to the simulation tactics.
However, our machine learning models are projecting this fight to be a striking fight that will go the distance. This kind of fight is Ulberg’s realm as he enjoys striking fights. He used a couple takedowns in his last fight, but Ulberg trains with Adesanya for a reason. In this case, Carlos Ulberg is the higher volume striker, but Carlos is the more accurate fighter in all categories.
Carlos Ulberg will win with a 29 to 28 decision. I see enough edge in the stats to place a bet for this fight, but I will be curious how the machine learning models predict this fight due to the profitability and accuracy.
Buys vs. Durden
This is the other closest matchup of the week right now. It is the same odds with the favorite being in the red corner. The only difference is it is scrappy flyweights who both need a win to stay with the UFC or keep their division afloat. JP Buys has not won a fight in the UFC and seems to have more drama outside of the cage. This is not typically the look the UFC prefers.
Durden is not much better in his UFC career with a 1-2-1 record. Both fighters are on the possible cut list with a loss. Even if they get a W, it still might mean the end of their UFC career.
Our model projects this fight to be a grappling fight that will not go the distance. Durden tends to get about 4 takedowns in a 15-minute fight with a takedown success rate of 38%. This means he should get about 1.546 takedowns in the fight. If Durden can get a takedown, then it is wonderful news for him as he averages a littel over a minute of control time for his takedowns. This control time could be enough to win a fight if nothing crazy happens. JP Buys is 0-2 against fighters with 13 or more for Control Time.
JP Buys has not had any good stats so far in the UFC. It is not a knock, but, overall, with 2 fights under his belt, I’m not sure what the sportsbooks see in him. Cody has won as a favorite before, and I don’t think the bookies know how to rate JP Buys.
Look at the below graph to see the huge discrepancy with odds for JP Buys’ past fights. He was listed at +435 for one and -165 for the other. These are massively different odds for a fighter… and even at those different odds, he lost both times. It seems like the sportsbooks just can’t pin him down yet. Not to add insult to injury, but, since 2011, fighters who are in the red corner with odds of -120 are 23-36 (38% win rate). I am going to roll with the data, unless our models predict otherwise. I think our models are going to see what we see.
I am riding with the underdog in this fight and see Durden getting his 2 takedowns and winning the fight 30-27 on a scorecard or two. Cody Durden will initiate the fight and prove the models accurate by making this a grappling fight.