Our reports refuse to print a confidence number unless both players are confirmed active. We went back and measured that rule against two full seasons of real pre-kickoff injury reports — and we're publishing the result even though it's the less flattering kind: the rule helps, and it is not enough to brag about.
A number next to a player who turns out to be inactive isn't analysis, it's a guess wearing one. So the product's standing rule is: if either player in a call is out, doubtful, or unknown, the slot prints “no call” and says why. The question we owed ourselves — and you — is whether that rule actually improves the numbers that do get printed.
On the same frozen set of 2,056 real start-sit decisions our earlier league study graded — a different, retired measurement from the grading record the product publishes today — applying the pre-kickoff injury report keeps 1,597 calls (77.7%) and drops the rest: 173 where a player was ruled out, 286 where one carried a questionable tag. The filter is fair to condition on because the injury report is published before the games are played.
| When we said… | Calls, no filter | We hit | Calls, filtered | We hit | Verdict |
|---|---|---|---|---|---|
| 50–55% | 608 | 53.3% | 483 | 54.2% | calibrated |
| 55–60% | 525 | 53.0% | 404 | 53.2% | calibrated |
| 60–65% | 390 | 52.6% | 303 | 52.5% | off |
| 65–70% | 245 | 54.3% | 191 | 53.9% | off |
| 70–80% | 208 | 56.2% | 166 | 58.4% | off |
| 80–90% | 37 | 54.1% | 25 | 48.0% | too few to judge |
Because it's an honesty measure, not a marketing measure. It stops us printing a confident number about a player who is listed as doubtful — which is the minimum a paying reader should expect — and this season, for the first time, we're recording who was active every single week, so the rule's live performance gets measured on the public record rather than reconstructed. That's also why a slot in your report sometimes says “no call”: the honest version of this product declines more calls than it makes.