You know what is humbling for a prediction model. Being wrong. You know what is more humbling. Being wrong while a coach is reading the Slate on the bus, his team down ten in the bottom of the fifth, asking the kid in seat fourteen why anyone trusted GameIQ in the first place. That happened to GameIQ at least once this week. Probably twice. We have receipts.
This is Maine in May. The dirt is dry, the lacrosse goals roll on carts the JV crew swears they were not pushing, and every prediction model in the state is one Saturday away from a haircut. We told the paying audience we would post a weekly performance review. What did GameIQ get right, what did it whiff, and is the math actually sharpening or just catching the calendar at the right angle.
Here is the unspun version. Baseball, softball, and lacrosse, this week versus last.
Baseball: from seventy four to eighty two. A week ago GameIQ called one hundred and eight of one hundred and forty six baseball games correctly. This week it called one hundred and fourteen of one hundred and thirty nine. That is an eight point bump. Eight points is not luck on a sample that big. The model has more games to chew on now, the early season chaos has tapered, and Class A chalk is starting to behave like Class A chalk. Coaches who screenshot the Slate on Mondays just got a quietly better Monday.
Softball: eighty eight, eighty eight, eighty eight. Last week eighty eight point four percent. This week eighty eight point seven. The needle did not move because there was nothing to move it. Maine softball at this stage of the season is what statisticians call a stable distribution and what coaches call we know who the good teams are. Two schools are running the table, the bottom of Class A is winless, and the chalk holds. GameIQ is not getting credit for predicting weather in May. It is just reading the room.
Lacrosse Boys: from eighty eight to seventy six. A twelve and a half point drop in one week. That is the worst single week swing across all four splits and it is not close. Thirty four right, eleven wrong, and at least three of the wrong ones were favorites who got cooked in regulation. Either two top five programs are quietly less top five than the PIR believes, or this was the random week the dice rolled the other way. The next week tells us which.
Lacrosse Girls: from eighty eight to eighty three. A softer drop, mostly because two coin flip games went to the dog. Forty three right, nine wrong. The misses were the games GameIQ already had at fifty two percent home win probability. That is not a prediction. That is a shrug with a graph attached.
GameIQ is running calibration on every game it reads, every night, learning from what went wrong yesterday so it whiffs less tomorrow. Eight points up in baseball is the receipt. Twelve and a half down in lacrosse boys is also the receipt. Both of them get rechecked next Sunday morning, and both of them get a column on the report card that does not lie about either one.
If you are a coach, this is your week to read the Big Board with one extra eyebrow raised. If you are a parent, this is your week to remember that going seventy six percent on lacrosse boys is still going seventy six percent. If you are GameIQ, this is your week to take the haircut, eat the loss, and run the numbers again.
It is May in Maine. Predictions are humbling. Pour the coffee. The next slate goes up at midnight, and the GameIQ is already studying for it.