In my opinion, it makes perfect sense. Isn’t the act of handicapping an analysis of the figures (stats) of athlete or team or race horse or player’s, actions in whichever games, race, event, etc? I was a harness horse groom, then trainer in the 1980s and the most reliable way to handicap a race is/was to compute in our heads/on paper/early computers the most likely winner of a race, the horse most likely to come in 2nd, 3rd, the horses to eliminate for consideration, all based on past performances. Yes, I know that past performances are no guarantee of future results, but you have to begin somewhere when handicapping.
Wouldn’t AI be the most reliable means upon which to rely when the much of the act of handicapping is analyzing the player’s stats?