When Does Average Depth of Target Stabilize for Wide Receivers?

So far I’ve found that average depth of target (aDOT) stabilizes in 10 games (or 286 aimed throws) for QBs and in 14 games (or 30 targets) for RBs. Today, I’ll present and discuss the aDOT stabilization point for WRs.

Methods

It’s been a while since my last post, so below is the standard procedure I use to determine a stat’s stabilization point; this time applied to WR aDOT:

1. I collected aDOT data for all WRs that had at least 8 games played for the same team from 2006 to 2016.
2. Starting with WRs that played 8+ games for the same team, I randomly selected two sets of 4 games for each WR and calculated their aDOT in both sets.
3. I calculated the split-half correlation (r) between the two randomly-selected sets of games.
4. I performed 25 iterations of Steps 2 and 3 so that r converged.
5. I repeated Steps 2-4 in 8-game intervals, from 16+ games all the way to 72+ games.
6. For each “games played” interval, I calculated
   1. the number of games at which the aDOT variance explained, R², would mathematically equal 0.5. ((The formula is (Games/2)*[(1-r)/r].))
   2. the “true” aDOT for a hypothetical WR that’s had an observed performance of 12.00 aDOT through that number of games. ((The formula is [(Observed Performance * Games) + (League-Average Performance * Stabilization Point)] / (Observations + Stabilization Point) ))
7. 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 WR aDOT:

Games	n	r	R2 = 0.50	Avg aDOT	Obs 12.50 aDOT
4	625	0.47	4	12.04	12.27
8	429	0.61	5	12.02	12.26
12	308	0.71	5	12.01	12.25
16	220	0.79	4	12.00	12.23
20	164	0.85	4	11.94	12.22
24	114	0.86	4	11.87	12.18
28	79	0.87	4	12.01	12.26
32	57	0.91	3	11.90	12.20
36	49	0.91	3	12.02	12.26
Wtd Average			4	12.00	12.25

As always, the bottom row labeled “Wtd Average” reveals my main finding: WR aDOT takes only 4 games to stabilize. Given an average of 5.3 targets and 26.1 routes run per game in my sample, this translates to only about 24 targets and only about 117 routes. Notably, this is faster than any WR stat I’ve analyzed thus far:

• Yards per Target (YPT) = 39 games (or 205 targets or 1,027 routes)
• Touchdowns per Route Run (TDPRR) = 33 games (882 routes)
• Yards per Route Run (YPRR) = 14 games (351 routes)
• Receptions per Route Run (RPRR) = 7 games (188 routes)
• Targets per Route Run (TPRR) = 7 games (184 routes)

That first bullet point isn’t a typo: aDOT does, in fact, stabilize in nearly 1/10th the number of games/targets/routes it takes YPT. What’s more, this is now the second time I’ve found YPT to be far less reliable than a mathematically related WR stat (i.e., YPRR previously). Given the overwhelming popularity of YPT in both fantasy and non-fantasy football circles, these results are troubling, to say the least. Using it for describing or explaining the past is fine. For predicting WR production in the future, however, it’s becoming increasingly clear that relying on YPT is ill-advised. And I’m not the only one who’s come to this conclusion.

To drive this point home, it’s worth showing the reliability timelines of aDOT, YPRR, and YPT in graphical form:

Along the x-axis is number of games, while the y-axis represents how much each stat reflects “true” skill as opposed to random luck. The dots represent each stat’s stabilization point, defined as the point at which it represents 50 percent skill and 50 percent luck. There are two major revelations when displaying my results this way. First, although I used “1/10th” earlier to describe the ratio of YPT stabilization to aDOT stabilization, reliability trajectories are non-linear. It may take 1/10th as long, but at no time is YPT’s skill percentage 10 times (or more) lower than aDOT’s. That said, the differences are still considerable:

• After 4 games, a WR’s actual aDOT represents 50.0% skill, but his actual YPT represents only 9.3%.
• After 14 games, a WR’s actual aDOT represents 77.8% skill, but his actual YPT represents only 26.4%.
• After 39 games, a WR’s actual aDOT represents 90.7% skill, but his actual YPT represents only 50.0%.

Now, one might argue, “OK. I get that YPT’s bad. But aDOT isn’t an analogous alternative because it’s more indicative of a WR’s offensive role than of his individual ability to gain yardage.” While I’m sympathetic to this logic, I do have two counterarguments. First, there’s the simple arithmetic that YPT = aDOT + YAC + a bunch of zeroes for yardage on incomplete targets, as well as the empirical truth that aDOT constitutes about 80 percent of YPT for the average WR. Indeed, among 130 WRs that played at least 25 percent of offensive snaps in 2016, only Cordarelle Patterson had a higher YAC per target (5.1) than aDOT (4.8). In short, YPT and aDOT are inextricably linked due to math.

Second, a WR’s offensive role is inextricably linked to his individual skill set in a chicken-and-egg sort of way. For instance, was DeSean Jackson’s consistent Top-20 aDOT the past three seasons the result of his role as Washington’s deep threat or was his role as Washington’s deep threat the result of an individual skill set best suited for high air-yard targets? What about Torrey Smith in San Francisco or Ted Ginn in Carolina? On the other end of the spectrum, what about the Cowboys’ Cole Beasley or the Rams’ Tavon Austin? What I’m getting at with these rhetorical questions is that aDOT is an indicator of individual yardage-gaining skill because WRs fill certain roles based on how their skill sets help maximize the team’s goal of gaining yardage (and ultimately scoring points). This is an unavoidable aspect of how the game of football works.

Finally, below is a table showing actual and “true” aDOT for every WR in 2016 that played at least 25 percent of snaps per PFF:

Player	Tm	G	Targ	aDOT	Rk	True aDOT	Rk
Sammie Coates	PIT	11	49	21.6	1	18.5	1
J.J. Nelson	ARI	14	67	18.1	3	16.5	2
Robbie Anderson	NYJ	14	75	17.5	4	16.2	3
DeSean Jackson	WAS	15	89	16.9	5	15.9	4
Mike Evans	TB	16	168	16.0	9	15.5	5
Marquise Goodwin	BUF	14	63	16.6	7	15.3	6
Roger Lewis	NYG	9	19	19.3	2	15.2	7
Josh Bellamy	CHI	12	38	16.9	5	15.0	8
Will Fuller	HOU	14	87	15.8	11	15.0	9
Terrelle Pryor	CLE	16	132	15.5	16	15.0	10
Corey Coleman	CLE	10	62	16.0	9	14.9	11
Dez Bryant	DAL	12	89	15.6	13	14.8	12
Michael Floyd	ARI	12	67	15.8	11	14.8	13
Kenny Stills	MIA	16	75	15.6	13	14.7	14
Alshon Jeffery	CHI	12	92	15.3	17	14.6	15
Julio Jones	ATL	14	125	14.5	24	14.1	16
John Brown	ARI	14	66	14.8	18	14.1	17
Brice Butler	DAL	14	30	15.6	13	14.0	18
Breshad Perriman	BAL	15	64	14.7	21	14.0	19
Rishard Matthews	TEN	16	104	14.3	27	13.9	20
Marvin Jones	DET	15	101	14.3	27	13.9	21
Sammy Watkins	BUF	8	50	14.6	23	13.8	22
Russell Shepard	TB	12	39	14.8	18	13.7	23
Allen Robinson	JAX	16	144	13.9	33	13.6	24
Torrey Smith	SF	12	45	14.4	25	13.6	25
Chris Hogan	NE	15	54	14.2	29	13.5	26
Cody Latimer	DEN	7	14	16.1	8	13.5	27
Phillip Dorsett	IND	15	56	14.1	31	13.5	28
Devin Funchess	CAR	14	54	14.1	31	13.5	29
Deonte Thompson	CHI	9	36	14.4	25	13.4	30
Ted Ginn	CAR	16	89	13.8	34	13.4	31
Jaron Brown	ARI	6	24	14.8	18	13.4	32
Andre Roberts	DET	13	25	14.7	21	13.4	33
Brandon Marshall	NYJ	15	121	13.6	36	13.3	34
T.Y. Hilton	IND	16	143	13.5	37	13.3	35
Cobi Hamilton	PIT	11	28	14.2	29	13.2	36
Tajae Sharpe	TEN	16	77	13.5	37	13.1	37
Jermaine Kearse	SEA	16	76	13.5	37	13.1	38
Brandin Cooks	NO	15	113	13.3	41	13.1	39
Travis Benjamin	SD	14	71	13.3	41	13.0	40
Vincent Jackson	TB	5	30	13.7	35	12.9	41
Emmanuel Sanders	DEN	15	131	13.1	44	12.9	42
DeVante Parker	MIA	15	82	13.2	43	12.9	43
A.J. Green	CIN	10	95	13.1	44	12.9	44
Paul Richardson	SEA	11	32	13.4	40	12.8	45
Jordy Nelson	GB	16	146	12.8	47	12.7	46
Charles D. Johnson	MIN	13	37	13.1	44	12.7	47
Tyrell Williams	SD	16	111	12.8	47	12.7	48
Mike Wallace	BAL	16	109	12.8	47	12.7	49
Brian Quick	LA	16	75	12.7	50	12.5	50
Kelvin Benjamin	CAR	16	112	12.6	52	12.5	51
Kenny Britt	LA	15	109	12.5	53	12.4	52
Chester Rogers	IND	9	31	12.7	50	12.4	53
DeAndre Hopkins	HOU	16	138	12.4	55	12.3	54
Davante Adams	GB	16	118	12.4	55	12.3	55
Jordan Taylor	DEN	11	24	12.5	53	12.3	56
Aldrick Robinson	ATL	13	31	12.4	55	12.2	57
Terrance Williams	DAL	15	61	12.2	58	12.1	58
Kamar Aiken	BAL	16	46	12.2	58	12.1	59
Marqise Lee	JAX	16	101	12.1	60	12.1	60
Brandon Coleman	NO	13	38	12.0	61	12.0	61
Victor Cruz	NYG	14	68	12.0	61	12.0	61
Dontrelle Inman	SD	16	89	12.0	61	12.0	61
Brittan Golden	ARI	7	14	11.9	64	12.0	64
Jeremy Maclin	KC	12	72	11.8	65	11.8	65
Bennie Fowler	DEN	10	21	11.5	66	11.8	66
Walt Powell	BUF	7	23	11.4	68	11.7	67
Ryan Grant	WAS	11	18	11.3	70	11.7	68
Taylor Gabriel	ATL	12	46	11.5	66	11.7	69
Allen Hurns	JAX	11	67	11.4	68	11.6	70
Tyler Lockett	SEA	14	62	11.3	70	11.5	71
Robert Woods	BUF	13	71	11.3	70	11.5	72
James Wright	CIN	10	19	10.8	82	11.5	73
Jeff Janis	GB	10	17	10.7	83	11.5	74
Ricardo Louis	CLE	7	32	11.0	79	11.4	75
Michael Crabtree	OAK	16	142	11.3	70	11.4	76
Odell Beckham Jr.	NYG	16	161	11.3	70	11.4	77
Adam Thielen	MIN	16	88	11.2	75	11.4	78
Kendall Wright	TEN	11	41	11.0	79	11.4	79
Nelson Agholor	PHI	15	62	11.1	76	11.3	80
Donte Moncrief	IND	9	54	11.0	79	11.3	81
Demaryius Thomas	DEN	16	141	11.1	76	11.2	82
Antonio Brown	PIT	15	151	11.1	76	11.2	83
Andre Holmes	OAK	14	23	10.3	90	11.2	84
Malcolm Mitchell	NE	11	45	10.6	84	11.1	85
Corey Brown	CAR	15	45	10.5	85	11.0	86
Danny Amendola	NE	11	28	10.0	94	10.9	87
Chris Conley	KC	16	65	10.5	85	10.9	88
Dorial Green-Beckham	PHI	14	67	10.4	88	10.8	89
Justin Hardy	ATL	11	30	9.8	96	10.8	90
Pierre Garcon	WAS	16	110	10.5	85	10.8	91
Jordan Matthews	PHI	14	109	10.4	88	10.7	92
Brandon LaFell	CIN	16	101	10.3	90	10.6	93
Braxton Miller	HOU	9	27	9.3	102	10.6	94
Charone Peake	NYJ	11	33	9.5	99	10.5	95
Cameron Meredith	CHI	14	95	10.1	92	10.5	96
Sterling Shepard	NYG	16	103	10.1	92	10.5	97
Amari Cooper	OAK	16	127	10.0	94	10.3	98
Jaelen Strong	HOU	6	23	8.5	110	10.3	99
Eli Rogers	PIT	13	65	9.6	97	10.2	100
Quinton Patton	SF	13	59	9.4	100	10.1	101
Quincy Enunwa	NYJ	16	99	9.6	97	10.1	102
Jordan Norwood	DEN	14	32	8.6	109	10.1	103
Seth Roberts	OAK	16	75	9.4	100	10.0	104
Rod Streater	SF	13	27	8.2	114	10.0	105
Tyler Boyd	CIN	16	76	9.2	103	9.9	106
Cody Core	CIN	6	26	7.5	121	9.7	107
Steve L. Smith	BAL	14	102	9.0	105	9.6	108
Jeremy Kerley	SF	16	108	9.0	105	9.5	109
Julian Edelman	NE	16	146	9.1	104	9.5	110
Doug Baldwin	SEA	16	117	8.8	107	9.3	111
Stefon Diggs	MIN	13	109	8.7	108	9.3	112
Andrew Hawkins	CLE	16	50	8.0	118	9.3	113
Tyreek Hill	KC	16	81	8.4	112	9.2	114
Mohamed Sanu	ATL	15	78	8.3	113	9.2	115
Golden Tate	DET	16	131	8.5	110	9.0	116
Jamison Crowder	WAS	16	92	8.2	114	9.0	117
Albert Wilson	KC	14	50	7.5	121	9.0	118
Michael A. Thomas	NO	15	119	8.2	114	8.8	119
Eddie Royal	CHI	9	44	6.9	124	8.7	120
Bryan Walters	JAX	8	31	6.1	128	8.7	121
Larry Fitzgerald	ARI	16	147	8.1	117	8.6	122
Willie Snead	NO	15	96	7.8	119	8.6	123
Tavon Austin	LA	15	101	7.7	120	8.5	124
Randall Cobb	GB	12	80	7.3	123	8.4	125
Anquan Boldin	DET	16	88	6.6	126	7.7	126
Jarvis Landry	MIA	16	122	6.8	125	7.6	127
Cole Beasley	DAL	16	92	6.5	127	7.6	128
Adam Humphries	TB	15	77	6.1	128	7.5	129
Cordarrelle Patterson	MIN	16	64	4.8	130	6.8	130

DT : IR :: TL : DR

Deploying my normal procedure for reliability analysis, I found that a WR’s average depth of target (aDOT) stabilizes in only 4 games (or 24 targets or 117 routes run). This is the fastest of any WR stat I’ve examined thus far and about 10 times faster than yards per target (YPT). Comparing the reliability timelines of aDOT and YPT reveals that, by the time a WR’s actual YPT represents half skill and half luck, his actual aDOT represents about 90 percent skill and only 10 percent luck. Taken together, my results suggests that we should be using aDOT — or at the very least, yards per route run (YPRR) — rather than YPT for the purposes of predicting WR yardage production.