SignalTuned SignalTuned

Rankings tuned to the signal, not the noise.

Need help?Sign in
TE2026-07-24
VerdictAlready priced in

The 'safe TE floor' premium is real. Your TE tier already has it.

Consensus on trialtested 27held 4busted 11no edge 12

The cliffhanger I left

Nine days ago I went looking for why our TE board loves the top of the position and shrugs at the middle. I found something real: at the same target share, TEs who stay on the field (14+ games) and catch what's thrown their way (68 percent or better) beat their same-volume peers by 2.35 points a game the following year. Not a fluke either, the gap ran positive in all five target-share bands I checked and cleared the bar in three of them, including the low-volume TE2 range. I called that a real signal the engine wasn't yet using, staged it at zero weight, and said the next step was checking whether the edge holds up against what the engine already sees coming from these players' own track record.

I also owe a correction. That post went out with a table of current board ranks, and within 48 hours the ranks had drifted enough that the buy and fade advice was backwards on more than one name. It got pulled two days later. The cohort finding was never the problem, the board snapshot was stale before the ink dried. This post uses fresh numbers off the live board (engine v1.6.132, generated 2026-07-23) and the promised second gate. Both are done now.

The test I owed

Same move I used to close out a rushing-QB premium and a route-runner premium earlier this summer: don't just ask whether the possession-floor TEs beat the field, ask whether they beat their OWN projection. Take each TE's incoming production and age, run it through our base TE aging curve (no archetype bonus stacked on, that would just tilt the test), and get a projection built entirely from information the engine already has. Then check: do the sure-handed, always-active TEs land above that number more than everyone else does?

If yes, that's a real gap the engine should close. If the gap disappears, the 2.35 was never missing, it was already sitting inside these players' own prior production, the exact thing the projection is built from.

I ran it on 421 TE year-over-year pairs, 2015 through 2024, 149 of them clearing the possession-floor bar.

What the data actually said

The possession-floor cohort landed 0.13 points a game above its own projection. The field landed 0.23 points above its own projection. Cohort minus field: -0.09, with a confidence range of -0.64 to +0.45. That interval sits on both sides of zero. There is no gap here worth acting on.

I also tried forcing the issue: what if I gave the model a small nudge upward for possession-floor TEs anyway, would it help? Every nudge from 4 percent to 20 percent made the model's predictions WORSE, with confidence intervals that stayed entirely on the wrong side of zero. A 2 percent nudge came out flat. Nothing helped, and nothing came close.

What the engine already figured out

Our TE archetype logic reads target share, nothing else, it has no idea whether a given TE is a sure-handed possession type or a boom-or-bust one. So there was a real question here: is that a blind spot? The answer is no, not because the engine secretly tracks catch rate, but because it doesn't need to. A TE who plays 17 games and catches three of every four targets thrown his way already RACKED UP the points that make him valuable. That production is the input the projection is built from. Paying him again for being reliable would be paying for the same season twice.

The 2026 names

Here's who this actually touches, using this year's real numbers, not the stale table from two weeks ago.

Trey McBride, TE1 on our board, ARI. In 2025 he played all 17 games and caught 74.6 percent of his targets, the exact possession-floor shape this test defines. His TE1 slot already reflects that. Hold. There's no hidden markup coming.

Kyle Pitts, TE4 on our board, ATL. This is the name I have to be honest about. Pitts is the TE our own known-error data flags as the worst single miss we've ever made, ranked 36th in 2021 when he finished 204th. That was a real bust, and it was a non-floor season, low games, low catch rate for his volume. It is also four years old. In 2025 Pitts played all 17 games and caught 74.6 percent of his targets, same shape as McBride, same shape this test rewards. Our board has him at TE4 today with high confidence. I am not fading him. Hold him at TE4. The 2021 bust is history, not a live signal, and this test gives no reason to second-guess where he sits now.

George Kittle, TE6, SF, age 32.9. Missed six games in 2025, so he doesn't clear this test's 14-game bar this cycle even though his per-target hands are still elite. That's not a knock on him, it just means this particular signal has nothing new to say about a TE6 vet coming off a shortened season. Hold.

Brock Bowers, TE2, LV. Also missed time in 2025 (12 games), same story as Kittle: outside this test's cohort for now, still ranked on pure talent and per-game rate, which this test doesn't touch. Hold.

Harold Fannin Jr. (TE3, CLE) and Tyler Warren (TE5, IND). Both sit right on the catch-rate doorstep this test uses, 67.3 and 67.9 percent against a 68 percent cutoff, with a full healthy season each. Too close to call floor-or-not off one year in the league. No verdict from this test either way. Hold and watch the catch rate trend.

What to do about it

If you were holding a possession-floor TE hoping the board would eventually catch up and rank him higher for being reliable, stop waiting. It already has, for both McBride and Pitts, exactly the way the section above lays out. There's no coming correction. What DOES move a tight end's rank is target share and production, same as ever, not a bonus for showing up healthy and catching what's thrown. One honest caveat: this test only covers TEs who've logged real target volume in a full-ish season, so it says nothing about a rookie top tight end who hasn't built a track record yet, and it isn't a verdict on Fannin or Warren specifically.

Receipts

← Back to all posts