Four Players the Talk Shows Over-Covered, and What It Cost Them

One thing this database is good for: finding the players whose names got said the most and checking whether the saying was worth anything.

A note on the calendar first. The most recent fully graded season in our data is the 2025 season; those grades landed in batches through August 2026. There is no graded 2026 play yet, so everything below is the season the industry spent last summer, fall and winter talking about.

What "over-covered" means here, and what we threw out

Two filters. First, volume: a player has to have drawn a meaningful pile of graded calls in the 2025 cycle. Second, the player has to have been available. A pundit who says a receiver hits 1,000 yards and watches him get hurt in October made a reasonable call that reality vetoed. That is not what we are measuring.

So two of the biggest accuracy sinkholes in the database are excluded here:

• Joe Burrow: 17 graded takes, 11.3 average accuracy. He finished 30th among quarterbacks in passing yards and second on his own team in passing yards, one of Rich Eisen's own graded articles is about the Bengals surviving without him. Injury, not misjudgment.
• Lamar Jackson: 27 graded takes, 9.9 average accuracy, a 23rd-place finish in passing yards and 20th in fantasy points. But the takes themselves keep hedging on health: Joe Pisapia ranked "a healthy Lamar Jackson" second, and Pat Fitzmaurice wrote that he was ranking him "as if he's going to play." We're leaving him out rather than pretending we can separate the two.

What's left is four players who were on the field, performed, and still burned everyone who talked about them.

The four

Forty-six graded calls from 20 different pundits, averaging 25.4 accuracy and −12.7 skill. Across all 6,249 graded takes in the database, by our arithmetic on the per-type table, the average take runs about +23 skill. These four names ran 35 points of skill below the house average.

The shapes differ. Allen's ten calls are five MVP re-picks and five variations on "best quarterback in the league"; he finished 11th in passing yards, which is what the site grades that claim against. And yes, this means we're also asking ourselves if he truly earned his MVP status. Hurts drew four separate "he's the fifth-best quarterback in the NFL" calls in June and July, each graded at 8.33 accuracy and −116.67 skill; he finished 16th. Mayfield's file is the October MVP surge: eight straight award calls scored zero, offset by four preseason "top ten quarterback" calls that landed, because he finished 10th in passing yards and 12th in fantasy points. His accuracy is wrecked; his skill isn't.

Taylor is the subtlest case and maybe the most instructive. He was good: third in rushing yards. But eight of his 13 graded calls were MVP or Offensive Player of the Year picks that went to Matthew Stafford and Jaxon Smith-Njigba, and the other five. All versions of "best running back in the league" scored 83.33 accuracy with negative skill, because the naive guess already had him fourth and predicting first was a reach past what he delivered. Coverage that heavy needs him to be the story of the season. He was the third-best rusher.

This isn't just "award picks are hard"

Fair objection: Award Picks league-wide average 21.7 accuracy across 351 graded calls, because one man wins and everyone else is wrong. Of the 99 logged MVP picks for 2025, 26 named Stafford; Mayfield got 8, Allen 5, Taylor 4, Hurts 1.

But award picks still average +26.07 skill site-wide, because the scoring gives credit for beating the obvious guess. Our four averaged −12.7. And the damage isn't confined to awards: Positional Finishes average 34.6 accuracy and +10.6 skill across 271 graded calls, while the Allen "best quarterback" cluster averaged 18.3 accuracy and −78.3 skill and the Hurts top-five cluster averaged 22.2 accuracy and −88.9. Worth flagging honestly: the site settles "best quarterback" claims by passing-yard finish, which is a proxy and not a verdict on quality. It is, however, the same proxy every one of these calls was graded against.

The mirror image

Over-coverage isn't the only way to be wrong. Drake Maye has 38 graded takes at 30.9 average accuracy, and his problem is the opposite. Preseason he was ranked QB20, QB12 and QB10; he finished second among quarterbacks in fantasy points. Then the industry over-corrected: MVP picks in January that lost to Stafford, plus Pro Bowl and All-Pro calls that graded zero. Ignoring a player costs you too. The lesson isn't "talk less," it's "talk in proportion."

Would abstaining have helped?

Mostly? No. We recomputed each pundit's mean accuracy over their graded takes with their flagged calls removed. (Raw arithmetic on the graded takes; the site's published score weights types slightly differently, so treat these as approximate.)

Shannon Sharpe is the real case. Three of his 21 graded takes are on these players: one Hurts top-five call, two "Mayfield has played better than everybody" calls. And they average 19.4 accuracy and −58 skill against his overall 59.49 accuracy and 11.44 MissedTakes Score. Drop them and his mean accuracy goes to roughly 66 and his mean skill roughly doubles. Three takes.

For the high-volume names it barely registers. Eisen's 11 flagged calls (eight on Taylor, two on Allen, one on Mayfield) averaged 22.7 accuracy; removing them from 332 graded takes moves him about a point and a half. Chris Brockman's eight flagged calls averaged 25 accuracy and −15.5 skill; removing them from 190 graded takes is worth about the same.

Which is the actual conclusion. Over-coverage doesn't sink one pundit's record. It spreads a thin, uniform layer of wrong across twenty of them. The cost isn't visible on the leaderboard; it's visible in the fact that 46 graded calls on four healthy, productive players returned less than nothing against a naive guess.