How Banged Up Bills started
BangedUpBills started the way a lot of things do in Buffalo: staring at the Bills injury report every week and wondering what it really meant. I’m Kyle Trimble, a Doctor of Physical Therapy, and I knew the report leaves out the part fans need most. “Questionable” can mean a player who never misses a snap, or one who doesn’t make the trip.
So in 2017, I started writing about it, talking about it, and eventually tracking it: every injury report, every practice line, and what actually happened on Sunday. What began as a Bills project grew into a league-wide database covering every team, going back years, and from there into forecasts.
Banged Up Bills started because injuries require more context than just a body part designation and whether they’ll play or not. Understanding the why behind the injuries and how they impact performance every Sunday makes you a more knowledgeable football fan.
What we’re building, and why it’s on Spartera
The goal is to turn the injury report into numbers you can act on. Not “he’s questionable,” but “he’s 72% to play, and here’s what he’s likely to score if he does.”
The products live on Spartera, a marketplace for data products. Instead of one big subscription, each BangedUpBills product answers one question, and you open the one you need. Several are free. Every product refreshes as new injury reports and game data come in.
Selling forecasts comes with a duty to prove they work. That’s what the rest of this page is for.
Grading ourselves
Every week we put a percentage on whether each injured NFL player will suit up. Then, after the games, we check every one of those calls against what actually happened, and we publish the results. Misses included.
Anyone can say their injury calls are “80 to 90% accurate.” Very few show you the calls behind that number, or explain how they counted. This page does both. It explains what our percentages mean, exactly which calls we grade, and how to read the scorecard below.
What a play probability means
A play probability is our estimate of the chance a player is active on game day. It is not a pick. When we say a player is 70% to play, we are saying that out of 10 players in his situation, about 7 will play and about 3 will sit.
That means a 70% player who sits is not a miss. It is one of the 3 we expected. You can’t judge a percentage by one player, the same way you can’t judge a 70% rain forecast by one dry day.
The right test is the group. Take every player we gave 70 to 79%, and check how many of them actually played. If the number lands in the 70s, the percentage meant what it said. If 95% of them played, or only 40%, it didn’t. That test is what the scorecard runs, for every range of percentages.
Which calls we grade
We only grade the hard calls: players listed Questionable or Doubtful on the final injury report. These are the players nobody can be sure about, so they are where a forecast earns its keep.
The rules, so anyone can check our math:
- Left out: players ruled Out, and players with no injury designation. Almost all of them go the way the report says, so counting them would make anyone look accurate.
- Locked before kickoff: a live forecast is the last one we saved before noon Eastern on game day. Nothing is changed after the games.
- What “played” means: the player was active on game day, whether or not he got a snap.
- What a “right” call is: a forecast of 50% or higher counts as a “plays” call, and under 50% counts as a “sits” call. The call is right if it matches what happened.
How to read the scorecard
The scorecard below updates every Tuesday after the week’s games. Each column answers one question:
| Column | What it answers | How to read it |
|---|---|---|
| Confident calls right | When we were sure (80% or higher, or under 30%), how often were we right? | This is the number to trust most. Higher is better. |
| Toss-ups right | When we were in between (30 to 79%), how often did the call land? | Lower by nature. These players are close calls for everyone. |
| All calls right | Across every Questionable and Doubtful player, how often were we right? | The overall number, with the count of calls beside it. |
| Sharper than a league-average guess | How much better are our percentages than just using the league-wide rate for that designation? | Measured with a Brier score, the standard grade for probability forecasts. Higher is better; 0% would mean we add nothing. |
The count under each percentage (“49 of 58”) tells you how many calls stand behind it. A percentage built on a handful of calls can swing a lot from week to week; one built on thousands won’t.
Live and backtest, kept separate
The scorecard has two kinds of results, and we always label which is which.
- Live means forecasts we published and locked before kickoff, then graded after the game. Live grading started in Week 8 of the 2026 season. This is the strictest test, because the forecast existed before anyone knew the answer.
- Backtest means we re-ran the same model as of each past week, from 2022 on, using only what was known at the time, and graded those forecasts. It gives thousands of calls to measure against, so the long-run numbers are steadier.
The point of showing both: if the live results hold up close to the backtest, the model works on real weeks, not just on paper.
Where we’re less sure
Some players are a true coin flip. When our number lands between 40 and 59%, the honest answer is that it could go either way, and the results show it: those calls land close to half the time.
We don’t hide that band. We flag it every week in our Toss-Ups list, so you know which players to wait on. For those players, check the inactive list about 90 minutes before kickoff and have a backup ready.
How fantasy players and bettors use it
Injury news moves lineups and lines. Each product is built for one of the decisions that comes with it.
Fantasy
- Start or sit a questionable player. Play Probability, Current Week gives his chance to be active. Expected Fantasy Points If an Injured Player Plays tells you what he’s likely to score if he does, so you can weigh the risk against your healthy option.
- Look up one player fast. Individual Player Play Probability answers the one name you’re stuck on.
- Avoid the late-scratch trap. This Week’s Toss-Ups lists the coin-flip players. In DFS, don’t lock them in early; keep a pivot ready for the late games.
- Find the replacement. Next Man Up shows who gains snaps and targets when a starter sits, for waivers and pivots.
- Plan for players coming back. Return Odds for Injured Starters gives the chance he’s back next game or within 3 games. Return Projection: Fantasy PPR projects his first 4 games back against his normal output, since most players don’t return at 100%.
- IDP leagues. Return Projection: Tackles (DB) and Return Projection: Pass Rush do the same for defenders, which almost no injury tool covers.
Betting
- Player props. Before you bet a questionable player’s over, check how likely he is to be active. For a player just back from injury, the Return Projection listings show how much his snaps and production usually dip in his first games back, including tackle and pressure numbers for defenders.
- Read the team, not just the tag. Team Tag Tendencies (free) shows which teams’ questionable players nearly always play and which rarely do.
- When the starting QB is out. QB Out: How Backups Do Against the Spread shows how backup quarterbacks have fared against the spread.
- Know before inactives. The percentages update as each practice report lands, so you have a read days before the 90-minute inactive list.
None of this guarantees a win. It’s better information about who plays and how much; the decisions are yours. If you bet, bet responsibly.
See every range, free
The scorecard above is the summary. The full graded breakdown is free on Spartera: every percentage range side by side with how often those players actually played, filterable by season, live or backtest, and Questionable or Doubtful.
See the full track record, free
Every product named on this page is on the BangedUpBills seller page. Sign up is free.
Media outlet, sportsbook, DFS, or fantasy tool? We share the full player-by-player graded log with qualified partners on request: kyle@bangedupbills.com.
Note: This article was written with the assistance of AI and edited by Kyle Trimble.
Discover more from Banged Up Bills
Subscribe to get the latest posts sent to your email.

