Terrance McKinney is the pick at 72.0% against Jim Miller's 28.0%, and the single factor behind it dwarfs the rest: Miller is 43 on fight night and McKinney 32, and that age gap is the largest listed effect on this card. Everything else is closer or points the other way — Miller is rated higher, 1922 to 1817, and the model lists both the rating difference and Miller's output-against-absorbed profile as factors in his favour. Miller is the more efficient fighter on the tape: 45.9 strikes landed and 57.3 absorbed per 15 minutes against McKinney's 56.9 and 62.0, in a division averaging 47.5 and 58.4. McKinney's volume figure carries an uncertainty of 65.3, larger than the number itself, because his fights end so early. A decision is likeliest at 52.0%, which is the least likely outcome of this pairing on the evidence of either man's record.

Miller is a measured, economical veteran with one standout weapon. His 0.94 submission attempts per 15 minutes is the highest figure in this bout against a lightweight average of 0.64, and his submission-finish share of 0.113 against the division's 0.04 is why three of his last five wins came by submission, over Jared Gordon, Damon Jackson and Gabriel Benitez. His takedown accuracy of 0.435 beats the division's 0.37. The rest is unremarkable and slightly below par in volume: 45.9 landed against 47.5, 35.0 at distance against 36.1, 7.8 clinch and 7.9 ground strikes against 11.2 and 10.9, with 192 seconds of control against 194 and 240 conceded against 228. He is 43 with 5.8 months off and a 0.53 takedown defence against the division's 0.54. McKinney is pure variance and the profile barely survives contact with a decimal point. Nine of his recent fights are first-round affairs in one direction or the other: knockouts of Kyle Nelson and Damir Hadzovic, a submission of Viacheslav Borshchev, a knockout loss to King Green and a submission loss to Chris Duncan. His measured output of 56.9 strikes per 15 minutes carries 65.3 of uncertainty, his ground striking of 15.1 carries 29.7, his submission attempts of 0.84 carry 1.42 — in every case the uncertainty exceeds the value, which is what happens when the denominator is minutes nobody fought. What is better established: he attempts 4.8 takedowns per 15 minutes against the division's 4.7, controls 219 seconds against 194, and absorbs 62.0 strikes against 58.4 while conceding 0.86 knockdowns against 0.62. He is 32, out 3.7 months, and has lost two of his last five by stoppage in round one. Two fighters who end fights early, one by submission and one by either method, and a model that calls a decision the likeliest result at 52.0%. That tension is the most interesting thing here and I think the method split is the part to distrust rather than the pick: Miller's 0.94 submission attempts meeting McKinney's 0.552 takedown defence, or McKinney's knockdown output meeting a 43-year-old's 0.533 striking defence, are both far more likely to produce a finish than a 52% decision rate implies. On the pick itself the model is making a narrow, defensible claim: Miller is the better fighter per minute and McKinney is eleven years younger, and across 4,314 walk-forward fights the age penalty has been measured as slightly too weak rather than too strong. If you want to disagree with 71.6%, disagree on the ground that McKinney's own chin has failed twice in five fights against far less experienced submission threats than Miller.