Back in January, I tweeted out this poll:
just finished running reliability analyses for avg depth of target. for which position do you think aDOT stabilizes in the fewest games?
— Danny Tuccitto (@IR_DannyT) January 29, 2017
Admittedly, I was quite surprised that RBs won. Given that quarterbacks and wide receivers (and tight ends to a certain extent) are far more involved in the passing game on average, I actually thought RBs would garner the fewest votes, not a winning plurality. If I were to attempt a journey into the minds of voters, a more statistics-savvy group than the general population it’s worth noting, I’m guessing a vote for RBs was based on the idea that there’s less variation in distance from one target to the next for this position, and less variation (generally) equals more consistency. If you voted for RB via different logic, let me know in the comments or on Twitter.
Whatever the reasoning (and the validity thereof), findings I’ve presented so far have already rendered the conclusion incorrect. Namely, whereas it takes 14 games (or 30 targets) for a RB’s average depth of target (aDOT) to stabilize, it takes 10 games (or 286 aimed throws) for QBs and only 4 games (or 24 targets or 117 routes run) for WRs. However, although we now know that RB aDOT doesn’t stabilize fastest, the question at the heart of the poll remains outstanding because I have yet to present the result for TEs. Today’s post achieves that end.
Methods
As always, below is the standard procedure I use to determine a stat’s stabilization point; this time applied to TE aDOT:
- I collected aDOT data for all TEs that played at least 8 games for the same team from 2006 to 2016.
- Starting with TEs that played 8+ games for the same team, I randomly selected two sets of 4 games for each TE and calculated their aDOT in both sets.
- I calculated the split-half correlation (r) between the two randomly-selected sets of games.
- I performed 25 iterations of Steps 2 and 3 so that r converged.
- I repeated Steps 2-4 in 8-game intervals, from 16+ games all the way to 72+ games.
- For each “games played” interval, I calculated
- the number of games at which the aDOT variance explained, R2, would mathematically equal 0.5. ((The formula is (Games/2)*[(1-r)/r].))
- the “true” aDOT for a hypothetical TE that’s had an observed performance of 8.50 aDOT through that number of games. ((The formula is [(Observed Performance * Games) + (League-Average Performance * Stabilization Point)] / (Observations + Stabilization Point) ))
- I calculated a weighted average of the results from Step 6. ((Weighted by group size.))
Results and Discussion
Below is the familiar stability table, this time for TE aDOT:
| Games | n | r | R2 = 0.50 | Avg aDOT | Obs 8.50 aDOT |
|---|---|---|---|---|---|
| 4 | 320 | 0.31 | 9 | 7.87 | 8.18 |
| 8 | 197 | 0.53 | 7 | 8.07 | 8.29 |
| 12 | 142 | 0.60 | 8 | 8.15 | 8.32 |
| 16 | 106 | 0.63 | 9 | 8.25 | 8.37 |
| 20 | 80 | 0.66 | 10 | 8.45 | 8.48 |
| 24 | 54 | 0.71 | 10 | 8.63 | 8.57 |
| 28 | 38 | 0.74 | 10 | 8.54 | 8.52 |
| 32 | 25 | 0.80 | 8 | 8.48 | 8.49 |
| 36 | 20 | 0.80 | 9 | 8.52 | 8.51 |
| Wtd Average | 9 | 8.14 | 8.32 |
Focusing on the bottom “Wtd Average” row, we see that TE aDOT takes 9 games to stabilize. Given an average of 3.8 targets and 18.2 routes run per game in my sample, 9 games is equivalent to 32 targets and about 157 routes. With this information, we now have our answer to the poll question: At 4 games (or 27 targets or 117 routes), WR aDOT stabilizes the fastest among offensive skill positions. Furthermore, we also now know that the position picked by poll voters to finish first actually finished last.
Although TEs didn’t win, adding their results to those of QBs, RBs, and WRs reveals two points to ponder. First, the fact that WRs and TEs exhibit more stable aDOTs from game to game than RBs suggests that the depth of “standard” targets within the context of an offensive scheme (i.e., primary or secondary reads) are more consistent than the depth of “non-standard” targets (i.e., tertiary reads aka dumpoffs). Conceptually, this makes sense, as the vast majority of WR and TE routes are designed, taught, and practiced with a specific depth in mind, whereas the vast majority of RB routes are of a nebulous “find open space in the short zone, wherever it may be” variety. ((Fifteen yards is the NFL official scoring threshold for “short” targets, and less than one percent of games in my RB sample (104 of 15,411) had an aDOT of 15 yards or more. )) Incidentally, this hypothesis would also explain why WRs and TEs have a lower aDOT stabilization point than QBs: Unlike the crystal clear route-design/target-depth aims of the former, the latter has their target depth waters muddied by the cloudiness of RB routes.
The second TE result worth pondering is the inverse of what I discussed in my RB aDOT post. To refresh memories, there was a trend in the RB results table that suggested aDOT decreases as tenure on the same team increases. In today’s table, we see the exact opposite trend for TEs: From 8 to 48 games, the “Avg aDOT” column steadily increases. The point to ponder here is the possibility of an emergent survivor effect. It may be the case that teams cut bait on pass-oriented RBs before they cut bait on pass-oriented TEs. Or maybe a RB’s receiving skill decreases with age, whereas a TE’s increases. Or maybe I just have too much Tony Gonzalez, Antonio Gates, and Jason Witten in my data set.
My waxing theoretical/hypothetical done, here’s True aDOT for TEs that played at least 25 percent of snaps in 2016 per Pro Football Focus:
| Player | Tm | G | Targ | aDOT | Rk | True aDOT | Rk |
|---|---|---|---|---|---|---|---|
| Rob Gronkowski | NE | 8 | 36 | 15.1 | 1 | 11.6 | 1 |
| Greg Olsen | CAR | 16 | 122 | 11.8 | 3 | 10.6 | 2 |
| Erik Swoope | IND | 9 | 21 | 11.9 | 2 | 10.2 | 3 |
| Delanie Walker | TEN | 15 | 96 | 10.6 | 4 | 9.8 | 4 |
| Jimmy Graham | SEA | 16 | 93 | 10.2 | 7 | 9.6 | 5 |
| Coby Fleener | NO | 16 | 80 | 10.1 | 9 | 9.5 | 6 |
| Cameron Brate | TB | 15 | 78 | 10.1 | 10 | 9.5 | 7 |
| Ed Dickson | CAR | 10 | 17 | 10.4 | 5 | 9.5 | 8 |
| Vance McDonald | SF | 10 | 43 | 10.3 | 6 | 9.4 | 9 |
| Jared Cook | GB | 10 | 50 | 10.2 | 8 | 9.3 | 10 |
| Hunter Henry | SD | 13 | 47 | 10 | 11 | 9.3 | 11 |
| Garrett Celek | SF | 13 | 47 | 9.9 | 12 | 9.3 | 12 |
| Vernon Davis | WAS | 15 | 59 | 9.8 | 14 | 9.3 | 13 |
| Austin Hooper | ATL | 11 | 25 | 9.9 | 13 | 9.2 | 14 |
| Tyler Eifert | CIN | 8 | 39 | 9.8 | 15 | 9.1 | 15 |
| Tyler Kroft | CIN | 6 | 12 | 9.8 | 16 | 8.9 | 16 |
| Levine Toilolo | ATL | 13 | 18 | 9.2 | 17 | 8.9 | 17 |
| Clive Walford | OAK | 15 | 49 | 9 | 18 | 8.8 | 18 |
| Zach Ertz | PHI | 14 | 101 | 9 | 19 | 8.8 | 19 |
| Dwayne Allen | IND | 14 | 48 | 9 | 20 | 8.8 | 20 |
| Mychal Rivera | OAK | 10 | 24 | 9 | 21 | 8.7 | 21 |
| Nick O'Leary | BUF | 10 | 13 | 9 | 22 | 8.7 | 22 |
| Antonio Gates | SD | 13 | 85 | 8.8 | 24 | 8.6 | 23 |
| Darren Waller | BLT | 7 | 15 | 8.9 | 23 | 8.6 | 24 |
| Richard Rodgers | GB | 15 | 43 | 8.7 | 25 | 8.6 | 25 |
| Charles Clay | BUF | 15 | 81 | 8.6 | 26 | 8.5 | 26 |
| Brandon Myers | TB | 7 | 14 | 8.6 | 27 | 8.5 | 27 |
| Eric Ebron | DET | 13 | 85 | 8.4 | 28 | 8.4 | 28 |
| Jordan Reed | WAS | 12 | 85 | 8.3 | 29 | 8.3 | 29 |
| Trey Burton | PHI | 14 | 58 | 8.1 | 30 | 8.2 | 30 |
| Jermaine Gresham | ARZ | 16 | 58 | 8.1 | 31 | 8.2 | 31 |
| Julius Thomas | JAX | 9 | 47 | 8 | 32 | 8.2 | 32 |
| Anthony Fasano | TEN | 9 | 12 | 8 | 33 | 8.2 | 33 |
| Daniel Brown | CHI | 6 | 20 | 7.9 | 35 | 8.2 | 34 |
| Luke Willson | SEA | 10 | 19 | 8 | 34 | 8.2 | 35 |
| Virgil Green | DEN | 10 | 34 | 7.7 | 37 | 8.0 | 36 |
| C.J. Fiedorowicz | HST | 15 | 82 | 7.8 | 36 | 8.0 | 37 |
| Gary Barnidge | CLV | 16 | 77 | 7.7 | 38 | 7.9 | 38 |
| C.J. Uzomah | CIN | 8 | 37 | 7.4 | 42 | 7.9 | 39 |
| Ryan Griffin | HST | 16 | 72 | 7.6 | 39 | 7.9 | 40 |
| Demetrius Harris | KC | 12 | 30 | 7.5 | 40 | 7.9 | 41 |
| Jesse James | PIT | 16 | 57 | 7.5 | 41 | 7.8 | 42 |
| Zach Miller | CHI | 10 | 62 | 7 | 44 | 7.6 | 43 |
| Kyle Rudolph | MIN | 16 | 120 | 7.2 | 43 | 7.6 | 44 |
| Jacob Tamme | ATL | 8 | 30 | 6.7 | 49 | 7.6 | 45 |
| Josh Hill | NO | 8 | 19 | 6.7 | 50 | 7.6 | 46 |
| Dion Sims | MIA | 13 | 34 | 7 | 45 | 7.5 | 47 |
| Tyler Higbee | LA | 11 | 26 | 6.9 | 46 | 7.5 | 48 |
| Travis Kelce | KC | 16 | 114 | 6.9 | 47 | 7.4 | 49 |
| Jason Witten | DAL | 15 | 91 | 6.8 | 48 | 7.4 | 50 |
| Ben Koyack | JAX | 9 | 24 | 6.3 | 54 | 7.3 | 51 |
| Jack Doyle | IND | 16 | 72 | 6.6 | 51 | 7.2 | 52 |
| Will Tye | NYG | 16 | 68 | 6.6 | 52 | 7.2 | 53 |
| Dennis Pitta | BLT | 16 | 116 | 6.5 | 53 | 7.1 | 54 |
| MarQueis Gray | MIA | 8 | 16 | 5.8 | 57 | 7.1 | 55 |
| Darren Fells | ARZ | 9 | 17 | 5.9 | 56 | 7.1 | 56 |
| Brent Celek | PHI | 10 | 17 | 5.8 | 58 | 7.0 | 57 |
| Martellus Bennett | NE | 16 | 66 | 6 | 55 | 6.8 | 58 |
| Lance Kendricks | LA | 16 | 74 | 5.5 | 59 | 6.5 | 59 |
| Marcedes Lewis | JAX | 10 | 28 | 4.8 | 60 | 6.4 | 60 |
| Brandon Bostick | NYJ | 8 | 10 | 2.8 | 61 | 5.7 | 61 |
DT : IR :: TL : DR
At the end of this reliability journey, it turns out that aDOT for TEs (9 games) stabilizes faster than for RBs (14 games) and QBs (10 games), but slower than for WRs (4 games). Taken together, these results suggest that RB aDOT muddies the otherwise pristine waters of QB aDOT based on WR aDOT and TE aDOT. In addition, in direct contrast to what I found for RBs, TEs seem to exhibit an aDOT survivor effect such that the lengthier their aDOT, the lengthier their tenure with a team.
Now that I’ve cycled through aDOT, next comes my series of analyses regarding the reliability of yards after catch (YAC) for QBs, RBs, WRs, and TEs. I hope you stay tuned.