Line Lab Research

The A/B/C Model: Full Daily Archive, and Why We Retired It

Retired 2026-08-07

For four and a half months this site's daily journal was a sharp-money model: three rules that fired on underdogs when the betting splits showed money arriving against the public. Every slate it ever graded is still linked below, wins and losses alike. It is retired, and this page explains why in the same detail we used when it was working.

What it did

The model read the gap between the share of bets on a side and the share of money, and backed underdogs where that gap said the sharper money disagreed with the crowd. Rule C — the strongest of the three — required the gap to appear in the morning and then cool by first pitch, which is the signature of a line that has already absorbed the sharp side.

Why it's retired

It stopped clearing its own costs. Our expected-value gate, which measures each rule's trailing 30-day return at the prices actually available, closed on Rule C at the end of July and never reopened — the moneyline leg had decayed from roughly +16% to negative. That gate is mechanical and it had been placing essentially zero bets a day for a week before we called it. The live record over the final review window was 13-18.

The honest summary is that the edge was real in backtest, marginal out of sample, and gone at the prices we could get. Publishing that is the whole point of keeping this page.

Every day it ever graded

← All research