Do Offensive Line Injuries Matter for Your Survivor Pick? We Tested It

September 11, 2026

A quarterback gets ruled out on Friday and the whole board reacts. The number moves a couple of points, your pool’s group chat lights up, and by Sunday everybody has adjusted. A starting tackle gets ruled out and the line barely twitches.

That asymmetry is the seed of a genuinely appealing idea. Every offensive snap runs through five linemen, and almost nobody in a survivor pool is checking who is starting at right guard. If line play is underweighted by the market, that is the kind of gap a survivor player would want to be standing in.

We liked the idea enough to test it properly. It did not survive.

What we tested

Everything below is our own calculation from public game data and Pro Football Reference snap counts, not a published study. Definitions first, because this argument is normally had without any.

Established offensive line starter. Entering a team’s game in week W of season S, a player counts as an established starter if his snap-count position is tackle, guard, or center and he played at least 80 percent of that team’s offensive snaps in at least two of the team’s three most recent prior regular season games of that season. Requiring three prior games drops weeks 1 through 3 of every season.

Out. An established starter is out for a game if he recorded zero offensive snaps in it.

The sample is every regular season team-game from 2013 through 2025, weeks 4 through 18, with a moneyline, a non-tie result, and at least four established starters identified, so that “nobody out” means an intact line rather than a team whose baseline we could not establish. Final count, 4,997 team-games: 1,474 missing one established starter, 277 missing two or more. For each we converted both closing moneylines into implied probabilities, stripped the sportsbook’s margin out proportionally, and compared the market to what happened.

The answer is no

Teams missing at least one established starting lineman won 47.80 percent of the time against a market-implied 48.84 percent. A residual of 1.0 percentage point, in the direction the theory wants, with a standard error of 1.1 points across 1,751 team-games (p = 0.35). Nothing.

Then the cut a survivor player actually lives in. Restrict to teams the market made at least a 65 percent favorite and you get 408 team-games with at least one established starter out, winning 74.5 percent against 74.9 percent implied: a residual of 0.3 points with a standard error of 2.1.

That is as flat as anything in the study, and it is the population a survivor pick actually comes from.

The one cut that looked like something

Reporting the inconvenient result is what makes the rest worth reading, so: teams missing exactly two established starters won 38.8 percent against 45.0 percent implied, landing at p = 0.042. Pool that with the three-or-more group and it comes to a residual of 6.2 points with a standard error of 2.8 across 277 team-games.

We do not believe it, for three reasons.

We ran a lot of cuts. 124 of them with at least 20 games, of which 10 came in under p = 0.05 against 6.2 expected by chance alone, and zero under p = 0.01 against 1.2 expected. That distribution is the signature of a null. To put a number on it we re-ran the entire family of 32 offensive line cuts on 5,000 random shuffles of the absence labels, recording how extreme the best result got each time. Ours was a z of 2.23. A search that wide produces something that big about a third of the time when nothing is there, for a search-adjusted p of 0.34.

It does not replicate across eras. 2013 through 2019 gives a residual of 1.5 points with a standard error of 4.1 (n = 129); 2020 through 2025 gives 10.3 points with a standard error of 3.8 (n = 148). Essentially all of it lives in the recent half, and no coherent story is available, because the headline one-or-more-out result is flat in both halves.

It is definition-sensitive. Swap in a stricter starter definition (80 percent of snaps in at least three of the previous six games) and the two-out residual shrinks from 6.2 points to 2.1 with a standard error of 2.2, while the one-or-more-out residual flips sign to positive 0.5 with a standard error of 1.0. A finding that depends on a three-game versus six-game lookback is not a finding.

There is also a practical problem. Among survivor-grade favorites, two or more established starters are out about 2.7 times per season league wide. You pick one team a week. And that combination, a strong favorite with two or more starters out, is 35 team-games in 13 seasons, too few to test: at 80 percent power it would only have surfaced an effect of about 20 points. So we retire that cell as unknown, not as a null.

Why this null means something

A null is only worth publishing if the method can find things. Ours can, and the cleanest demonstration is what it sees in the market. Regressing the posted spread on team-season and opponent-season fixed effects plus absence flags, a team missing its quarterback is priced about 2 points worse, an estimate 12.5 standard errors from zero, and a team missing two or more established linemen about 1.3 points worse. The pipeline picks out the absences the market reacts to, plainly and with room to spare. It also tells us the lines in this data already know who is inactive, which makes the question narrower and more useful: given that the market has already discounted the absence, is the discount wrong? It is not.

The same machinery sets the scale of what we were hunting for. Run it on quarterbacks: 529 team-games with an established starting quarterback missing, winning 35.5 percent against 39.0 percent implied, a residual of 3.5 points with a standard error of 2.0. Read that carefully. Even for the most valuable player on the field, the measured gap is only about three and a half points, and even that does not clear the usual significance bar. Any offensive line residual should be a fraction of it.

One more number, because “we found nothing” and “we could not have found anything” are different claims. At 80 percent power the one-or-more-out sample of 1,751 team-games would have detected 3.1 percentage points, so anything large enough to move a pick would very likely have shown up. The survivor-grade favorite cut is smaller, 408 team-games, and rules out roughly 6 points rather than 3, so the guarantee on the cut that matters most is the looser one. Informative null, not an empty one.

The useful thing we found instead

Here is the part that changes what you do on a Sunday morning, and it has nothing to do with linemen specifically. Building this test meant deciding who was actually absent, so we cross-checked our snap-derived absences against the official weekly injury report across 24,883 established-lineman game observations. Two findings came out, pointing opposite ways.

The first is that the report’s strong designations are essentially perfect. Of the 826 times an established lineman was listed Out or Doubtful on the final weekly report, the player went on to record zero offensive snaps every time. Not most. All 826 cases.

The second is that only 38.1 percent of actual absences get flagged that way. There were 2,166 games in which an established starter took no offensive snaps, and just 826 of them carried an Out or Doubtful designation. The most likely reason is mundane rather than sinister: players on injured reserve drop off the weekly report entirely, so a lineman hurt a month ago simply stops appearing. We did not measure how large that share is, and some of these absences are benchings, trades or suspensions rather than injuries at all. Of the 1,340 unflagged absences, only 39.3 percent show up anywhere on the report in any capacity.

So a survivor player who opens the injury report, sees a clean sheet, and concludes everybody is available has learned much less than they think: roughly three out of five absences are never flagged, and most of those never appear on the report at all. The report is a strong positive signal and a weak negative one.

So: trust an Out or Doubtful designation completely, do not trust silence, and check the inactives list and roster status rather than the weekly report alone.

What we could not test

Two more limits, where the honest answer is “unknown” rather than “no.”

The blindside version is untested. Public snap data records a lineman’s position as tackle without distinguishing left from right, so the most plausible form of this thesis, that losing the blindside protector specifically is underpriced, cannot be isolated with this data at all. The closest we get is tackles against interior linemen, and those are indistinguishable from each other and from zero: at least one tackle out gives a residual of 2.0 points with a standard error of 1.6, at least one interior lineman out 1.2 with a standard error of 1.4. A weaker test of a different question. Untested is not disproven, and settling it needs alignment data we do not have.

The early-lock version is untested, and it is the better argument. Our moneylines are closing lines, and they already know who is inactive. So the study answers whether the market’s discount is wrong, not whether a pick locked on Wednesday gets punished by news that lands on Saturday. That would take timestamped line history that no free public source publishes and this data set does not contain. Of everything here it is the strongest surviving reason to care when your pool locks, and it is the structural problem that makes a quarterback ruled out after your deadline so dangerous.

What to do instead

Four things, none of which require a spreadsheet.

Start from the price. The closing line has already absorbed the line injuries, about right as far as we can tell, so re-discounting a favorite charges you twice.

Check availability, not the injury report. Inactives and roster status catch the absences the weekly report structurally cannot.

Notice when a storyline moves your pool without moving the number. If people around you are talking themselves off a strong favorite over an injury the market shrugged at, that favorite likely got cheaper in pick-equity terms while nothing we can measure changed its chance of winning. That first half is reasoning rather than measurement, since nobody publishes pool-wide ownership data clean enough to prove it. Same mechanic as trap games.

Then spend the attention where it pays. What decides survivor seasons is not whether this team wins Sunday, it is what the pick costs you in November, the subject of the complete strategy guide. New to the format? How survivor pools work covers why a pick is a one-time expense.

We went looking for an edge in the trenches and did not find one. Worth knowing, because the hours you were going to spend on depth charts are better spent on the two things that do move a pick: how likely your team is to win, and what using them now costs you later. Survivor Caddy runs both, with win probabilities from our own Elo-based power ratings. Join the waitlist and we will email you once at launch.