Total QBR in American Football, Explained

What ESPN’s Total QBR is trying to do that American football’s passer ratings cannot, what it needs to do it, and why this page computes no QBR figure. The model is unpublished and its inputs are not on a box score, so any number produced here would be a fourth quantity wearing a familiar name.

Three quantities quoted as a quarterback rating

What Total QBR is built from, and where each piece comes from

What this page does with QBR, and what it refuses

What QBR is trying to do, and why this page will not compute one

American football’s two passer ratings share a blind spot. Both take five counts off a passing line — attempts, completions, yards, touchdown passes and interceptions — and neither knows anything about the situation any of those plays happened in. A twelve-yard completion on third and fifteen counts exactly as much as one on third and three. A touchdown pass from the one-yard line counts the same as one from forty. Down, distance, field position, the clock and the score are all absent, and so is any account of whether the quarterback or the receiver did the work.

Total QBR is an attempt to fix exactly that, and the attempt is serious. It starts from expected points added per play: the change, caused by a play, in the points its team can expect from that position. It weights plays by how much they mattered to the outcome, through what ESPN calls a clutch index and a win probability model. It divides the credit for a play between the quarterback and the people around him — after the 2015 revision, credit for yards after the catch is “based on what is expected, given the throw”. And it adjusts for pressure, by having people watch the play. The result is scaled from nought to a hundred. That is a better question than either passer rating asks.

This page computes none of it, and there are two separate reasons, either of which would be enough.

First, the model is not published. ESPN’s own two explainers could not be read at all: both were fetched, on three different hosts between them, and every attempt returned a document with no body text. So what follows comes from three sources that did read ESPN’s material. A Montana State graduate writing project from 2017 records that “Total quarterback rating has to be taken from ESPN as it is proprietary”, that the calculation reportedly takes “thousands of lines of complicated code”, and that “ESPN has never been specific about what is going on in these thousands of lines of code”. The nflWAR paper, which builds an open alternative, states that “Total QBR is not directly reproducible, since it relies on human judgment when evaluating plays”, and separately that “ESPN has a proprietary expected points metric, but does not detail the specifics of how it is calculated”. Football Perspective’s account of the 2015 revision refers to “the proprietary nature of the statistic” and notes that “such a small percentage of the formula is actually made public”.

Second, the inputs are not available even if the model were. This is the harder obstacle and it is the one that settles it. The nflWAR paper lists what an expected points model needs, in full: the current down, the seconds remaining in the half, the yard line, the yards to go for a first down, whether the situation is goal-to-go and whether under two minutes remain, with three interactions between them. That is a row per play, not a season line. On top of it QBR needs a division of credit between players, which requires somebody to have watched the play, and a pressure judgement made by a charter. Nothing on any box score supports any of it.

Be precise about what that does and does not establish. This page has read no licence, no fee schedule and no terms of use for Total QBR, so it does not claim the metric is encumbered. The finding is the weaker and more accurate one: the model could not be established as published, three independent sources that read ESPN’s own material describe it as proprietary and partly judgement-based, and the inputs are not on a box score. Any of those alone would stop this page computing a figure. Together they make a computed QBR a fourth quantity that would be read as the third.

The same test applied to two neighbours, with different answers. DVOA — Football Outsiders’ Defense-adjusted Value Over Average — is described by the account read for this page as one of that site’s “proprietary formulas”, with the metric explained at a high level and the formula not published; the statistics have since “been moved to FTN Fantasy, which is owned by FTN Network” and the original site was no longer accessible from 1 September 2023. No licence or fee schedule was read, so the finding is again “could not establish that it is free to reproduce”, and this site builds nothing. Expected points added is different and the distinction matters: EPA is a family of models rather than one, and while ESPN’s is proprietary, open implementations exist with every covariate published — the nflWAR authors’ own model is “built entirely on publicly available data”. So EPA is not refused on licensing grounds at all. It is not built here because it needs play-by-play input this site does not take, and because “EPA” without the model behind it names nothing in particular.

What this page does instead. It tells you which of the three figures you are looking at, which is the question most readers actually arrive with, and it points at the two that can be computed exactly: the NFL passer rating calculator for the four-component formula and its caps, and the NCAA passer efficiency rating calculator for the college one, which gives a different answer on the same line. Underneath both are the passing rates themselves, every one of which a box score supports exactly. None of those is QBR, and all of them are checkable.

Every figure here is arithmetic on counts a league or a statistics provider publishes. This page produces no commercial prediction of any kind, and no page in this section does.

Frequently asked questions

Related calculators

References

1. Yurko R, Ventura S, Horowitz M. nflWAR: A Reproducible Method for Offensive Player Evaluation in Football, arXiv:1802.00998 — “Total QBR is not directly reproducible, since it relies on human judgment when evaluating plays”; “ESPN has a proprietary expected points metric, but does not detail the specifics of how it is calculated”; Pro Football Focus’s grades are “solely based on human judgment and proprietary to PFF”. The authors’ own expected-points model is “built entirely on publicly available data”, with its covariates listed in full — current down, seconds remaining in the half, yards from the end zone, log yards to go, a goal-to-go indicator and an under-two-minutes indicator, plus three interactions — the list that shows the inputs are play-by-play and not box-score. arxiv.org (read 15 Oct 2026).
2. Gomez N. A statistical analysis of quarterback performance, Montana State University graduate writing project, 2017, 56 pages — “Total quarterback rating has to be taken from ESPN as it is proprietary” and “as much of that data is not made publicly available”. On the structure: “their model has four major components: win probability, expected points, division of credit”, plus “the ‘clutch’ index, analyzes each play and determines how important that play was to the outcome”, all “scaled to be between zero and one hundred”. On the opacity: the calculation reportedly takes “thousands of lines of complicated code”, attributed to ESPN’s Dean Oliver, and “ESPN has never been specific about what is going on” in it. math.montana.edu (read 15 Oct 2026).
3. Stuart C. “ESPN’s Total QBR, updated for 2015.” Football Perspective — “The base statistic used throughout QBR is EPA, which stands for Expected Points Added per play”; the 2015 revision made credit for yards after the catch “based on what is expected, given the throw”, and the pressure adjustment relies “on game charters to determine the level of pressure”. On disclosure: “The exact changes aren’t clear”, “the proprietary nature of the statistic”, and “such a small percentage of the formula is actually made public”. footballperspective.com (read 15 Oct 2026).
4. ESPN’s own two explainers — Guide to the Total Quarterback Rating (story id 6833215) and How is Total QBR calculated? (Stats and Info blog, post id 123701) — were fetched three times between them, on the .com, .co.uk and africa.espn.com hosts, and every attempt returned a document with no body text. So this site does not quote ESPN on ESPN’s own metric, and it has read no licence, fee schedule or terms of use for Total QBR. Its position is the weaker and more accurate one: the model could not be established as published (attempted 15 Oct 2026).
5. Wikipedia. Football Outsiders — DVOA is “Defense-adjusted Value Over Average”, described as one of the site’s “proprietary formulas”, explained at a high level with the formula not published. The DVOA statistics “have been moved to FTN Fantasy, which is owned by FTN Network”, and the original site “seems to have been abandoned by its owners”, no longer accessible as of 1 September 2023. No licence or fee schedule was read, so the finding recorded on this site is “could not establish that it is free to reproduce”, not “it is encumbered” (read 15 Oct 2026).
6. Pro Football Hall of Fame. NFL’s Passer Rating — the four-category system in full: completion percentage scored as “(Comp. Pct. − 30) × .05”, with “If the results are greater than 2.375 (Comp. Pct. greater than 77.5), award 2.375”; yards per attempt as (yards ÷ attempts − 3) × .25; touchdown percentage × .2; and interception percentage as 2.375 − (Int. Pct. × .25), floored at nought past 9.5 per cent. “The sum is then divided by six … and multiplied by 100.” The scale runs from “.000” to “158.3 being the highest rating a passer can achieve.” It works a published example: Peyton Manning’s 2004, 336 of 497 for 4,557 yards, 49 touchdowns and 10 interceptions, rated 121.1. profootballhof.com (read 15 Oct 2026).
7. Sports-Reference. College Football Stats Glossary — “Passing Efficiency Rating; statistic used by the NCAA to evaluate passing efficiency. The formula is (8.4 * Yds + 330 * TD − 200 * Int + 100 * Cmp) / Att.” Also “Adjusted Passing Yards per Attempt; the formula is (Yds + 20 * TD − 45 * Int) / Att” and yards per attempt as “Yds / Att”. A short page, about 3,000 characters, with the formulas in body text rather than images. sports-reference.com/cfb (read 15 Oct 2026).