Thomas Petersen is the pick at 68.3% against Gabriel Lorenco's 31.7%, and the honest reason is that we have no record for Lorenco at all — no bouts, no style profile, no reach and no stance. The model is comparing Petersen's 1747 rating against the heavyweight prior, so the 68.3% is a statement about an unmeasured debutant and not a read on this fight. Petersen himself is a grappling heavyweight: 5.2 takedown attempts and 254 seconds of control per 15 minutes against division averages of 3.9 and 203. He has lost three of his last five, two by stoppage. Read the number as the division's base rate for a debut, and nothing more.

Gabriel Lorenco has no fight in this database. There is no style profile, no rating, no uncertainty, no reach and no stance, and nothing below is about him. He was born in September 1999, which makes him 27 on fight night. That is the whole of what we can say. Petersen is measured and the measurements point one way. He attempts 5.2 takedowns per 15 minutes against a heavyweight average of 3.9 and holds 254 seconds of control against the division's 203, while conceding only 148 against an average of 190 — so he gets on top and stays there. His striking is average in volume and better than average in accuracy, 49.0 landed at 49.9% against 49.4 at 47.0%, and he absorbs 50.5 per 15 minutes against the division's 57.0. He is 31, has been out 3.3 months, and his record is streaky rather than bad: submitted by Valter Walker in round one, a majority decision over Guilherme Pat, stopped by Vitor Petrino in round three, a decision over Don'Tale Mayes, stopped by Shamil Gaziev in round one. His pace ratio of 1.56 against an average of 1.06 carries plus or minus 1.35, so it says nothing. With one side of the fight unmeasured there is no style matchup to describe, only what Petersen will try to do: get a hold of his opponent early and win the fight on the floor. The 50.2% chance of a decision and the 33.3% chance of a knockout are the heavyweight division's shape given Petersen's profile, not a prediction about an exchange nobody has data on. Two things would make the model badly wrong here and neither is visible: Lorenco being a legitimate prospect rather than a short-notice replacement, and Petersen's own chin, which has failed twice in five fights. If you want a reason to doubt the 67.8%, it is that the only thing it knows about this fight is one man's name.