Logarithmic Decay in the NFL Draft: Do Busts Matter?

Welcome back for Part 2 of my pre-draft stat series wherein I attempt to test my own theories about the importance of avoiding busts and finding diamonds in the rough. In Part 1, I took a stab at developing objective definitions for "bust" and "diamond in the rough." Based on an analysis of all NFL draft picks from 1994-2005, the definitions I arrived at were as follows:

• An NFL draft bust is a player who was selected in the 1st or 2nd round and played 67% or more below the expected performance of his specific pick number.
• An NFL draft diamond in the rough is a player who was selected after the 2nd round and played 200% or more above the expected performance of his specific round.

Today in Part 2, I'll be focusing on busts (insert juvenile humor here). Specifically, I'll be telling you who the biggest busts of the 1st and 2nd rounds were from 1994-2005, tell you how each team has done with respect to avoiding busts, and finally answer the question, "Do good teams avoid picking busts?" Given that this is Niners Nation, I'll be focusing especially on how the 49ers have fared bust-wise.

After the jump, I dive deep into the anatomy of busts...

PHANTOM BUST

Before I get started, I just want to rehash an issue that arose during Part 1; namely, the uncertainty about whether specific pick numbers matter. I think I pretty convincingly showed that they matter for picks in Rounds 1 and 2, but two issues make me feel like I need a qualifier in my bust definition:

1. The logarithmic decay function did a better job of explaining draft pick performance when I used rounds rather than picks.
2. The trendline for my pick-by-pick analysis, although awesomely predictive overall, did worst at the top of the draft (i.e., Round 1).

So, with this kind of conflicting evidence about the correct equation to use, I'm going to call pre-flop rather than pushing all-in. That is, I'm going to create a 2nd, less-pick-dependent class of busts, which I'll call "mild busts." These mild busts are players who played 50% or more below the expected performance of his pick number, which is slightly better than the 67% underachievement exhibited by full-on, no-doubt-about-it busts in my definition.

ATTACK OF THE BUSTS

Excluding 2 Ks and 1 P, there were a total of 748 total picks in the 1st and 2nd rounds from 1994-2005. Of these 748 picks, 70 of them were full-on busts, and another 40 were mild busts, which adds up to a total of 110 total busts, or 14.7%. Given how much people like to describe the draft as a crapshoot, I think that a non-bust rate of 85.3% is pretty darn good and suggests that NFL GMs actually know what they're doing for the most part. After all, 14.7% translates into about 1 bust every 6 picks.

Looking at individual rounds, there have been only 25 full-on busts among the 370 picks from 1994-2005, for a bust rate (BR) of 6.8%. Again, given the volume of bust-related discussion around water coolers and message boards, you'd think that percentage would be way higher. If you had in the mild busts, the NFL BR in Round 1 increases to 11.1%, which is still not too shabby.

Either way, we're talking about 1 full-on bust for every 7 first-round picks or 1 overall bust for every 10 first-round picks. Alternatively, we're talking about between 2 and 4 first-round busts per draft (depending on whether you include the mild ones). I'd venture to guess that that's far fewer than most people assume. I think, given the commentary, it seems like most fan bases consider their team's first-round pick a bust unless he becomes the next Peyton Manning.

REVENGE OF THE BUSTS

So who was the cream of the bust crop from 1994-2005? Who were the busts to end all busts? Below is a table showing the Top 20 1st-round busts during that period:

Top 20 First-Round Busts
Rk	Player	Year	Pk	Pos	Tm	NFL Yrs	AV Avg	Exp AV Avg	Under
1	Andre T. Johnson	1996	30	T	WAS	2	0.00	3.84	100.0%
2	Akili Smith	1999	3	QB	CIN	4	0.25	7.12	96.5%
3	Craig Powell	1995	30	LB	BAL	4	0.25	3.84	93.5%
4	Reggie McGrew	1999	24	DT	SF	3	0.33	4.16	92.0%
5	Trezelle Jenkins	1995	31	T	KC	3	0.33	3.80	91.2%
6	Ryan Leaf	1998	2	QB	SD	4	0.75	7.70	90.3%
7	Jerome McDougle	2003	15	DE	PHI	6	0.50	4.83	89.6%
8	Jamal Reynolds	2001	10	DE	GB	3	0.67	5.41	87.7%
9	Jim Druckenmiller	1997	26	QB	SF	2	0.50	4.05	87.6%
10	Dimitrius Underwood	1999	29	DT	MIN	2	0.50	3.89	87.1%
11	Marcus Nash	1998	30	WR	DEN	2	0.50	3.84	87.0%
12	Ki-Jana Carter	1995	1	RB	CIN	9	1.33	8.68	84.6%
13	John Avery	1998	29	RB	MIA	6	0.67	3.89	82.9%
14	Charles Rogers	2003	2	WR	DET	3	1.33	7.70	82.7%
15	Willie Middlebrooks	2001	24	DB	DEN	5	0.80	4.16	80.8%
16	Heath Shuler	1994	3	QB	WAS	4	1.75	7.12	75.4%
17	Wendell Bryant	2002	12	DT	ARI	3	1.33	5.15	74.1%
18	Trev Alberts	1994	5	LB	IND	3	1.67	6.39	73.9%
19	Rashaun Woods	2004	31	WR	SF	1	1.00	3.80	73.7%
20	Troy Williamson	2005	7	WR	MIN	5	1.60	5.91	72.9%

As always, some housekeeping first. In the table, AV Avg basically translates as actual career performance, whereas Exp AV Avg is what performance was expected given that player's pick number (Pk). The Under category tells you how much the player underachieved performance expectations.

At first glance, what jumps out to me is the fact that the 49ers had 3 of the Top 20 busts, and 2 of the Top 10. Also, they stand out as the only team to have selected a player that busted out after only 1 season in the NFL.

Oh, and one other thing of note about the Niners' busts. Each draft regime got in on the Top-20 bust party. Druckenmiller was taken by the Policy regime, McGrew was taken by the 2nd Walsh regime, and Woods was taken by the Donahue regime. To boot, if we expanded this out to include all busts (full-on or mild), it would be 4 for 4: Alex Smith (taken by the McNolan regime), has been a mild bust thus far in his career; ranking 27th with an underachievement rate of 63.1%.

A NEW BUST

Now let's move to Round 2. Below is a table showing the Top 20 second-round busts from 1994-2005:

Top 20 Second-Round Busts
Rk	Player	Year	Pk	Pos	Tm	NFL Yrs	AV Avg	Exp AV Avg	Under
1	Leon Bender	1998	31	DT	OAK	0	0.00	3.80	100.0%
2	Nathan Davis	1997	32	DE	ATL	3	0.00	3.75	100.0%
3	Charles Fisher	1999	33	DB	CIN	1	0.00	3.71	100.0%
4	James Manley	1996	45	DT	MIN	1	0.00	3.27	100.0%
5	Israel Ifeanyi	1996	46	DE	SF	1	0.00	3.23	100.0%
6	Paul Toviessi	2001	51	DE	DEN	0	0.00	3.09	100.0%
7	Pat Riley	1995	52	DE	CHI	1	0.00	3.06	100.0%
8	Jacob Rogers	2004	52	T	DAL	1	0.00	3.06	100.0%
9	Dan Cody	2005	53	DE	BAL	1	0.00	3.03	100.0%
10	Eric Shelton	2005	54	RB	CAR	1	0.00	3.01	100.0%
11	Michael Boireau	2000	56	DE	MIN	0	0.00	2.95	100.0%
12	Terrence Murphy	2005	58	WR	GB	1	0.00	2.90	100.0%
13	Jimmy Oliver	1995	61	WR	SD	0	0.00	2.83	100.0%
14	Jesse James	1995	62	C	STL	2	0.00	2.81	100.0%
15	Shane Hannah	1995	63	G	DAL	0	0.00	2.79	100.0%
16	Marquise Hill	2004	63	DE	NE	3	0.00	2.79	100.0%
17	Marques Tuiasosopo	2001	59	QB	OAK	8	0.13	2.88	95.7%
18	Alex Van Dyke	1996	31	WR	NYJ	5	0.20	3.80	94.7%
19	Andrew Greene	1995	53	G	MIA	4	0.25	3.03	91.8%
20	Jacoby Shepherd	2000	62	DB	STL	4	0.25	2.81	91.1%

As you can see, almost all of the Top 20 consists of players who literally added no value to their teams during their careers. Furthermore, 5 of the Top 20 never even played a single game in their NFL careers. I don't know what's worse: playing 0 years or playing 1 or more years without ever adding any value to your team. For the sake of simplicity, I ranked all those 100% underachievement guys in descending order of Exp AV Avg.

As far as the 49ers go, they're far less prominent in the 2nd-round bust list: only 1 of the Top 20, and only 1 bust total (full-on or mild). Rather than the Niners, the team that stands out to me here is their cross-town NFL-mates, the Oakland Raiders. Specifically, they treated their massive busts very differently. Whereas they took the #1 2nd-round bust from 1994-2005, a DT who ended up never playing for them, they also took the #17 bust, but he ended up playing 8 seasons for them. Go figure. I think Marques Tuiasosopo was the Raider equivalent of Arnaz Battle in that I find myself wondering, "How did he stay on their roster for that long?"

THE BUST STRIKES BACK

Although that's a question I often ask myself, the one that I'm specifically here to answer is, "Do busts matter?" That is, do good teams do a good job of avoiding busts? To find that out, I tallied up all the busts for each team, got their winning percentages from 1994-2009, and calculated the league-wide correlations of winning percentage to (a) total busts, and (b) BR. Below is a table showing every team's bust totals along with the league-wide totals and correlations:

Team	Win Pct	Pks1	Pks2	Tot Pks	FB1	TB1	FB2	TB2	FB	TB
MIN	0.566	16	15	31	3	3	2	7	5	10
CIN	0.393	13	12	25	3	5	1	2	4	7
KC	0.508	10	9	19	1	2	3	4	4	6
SF	0.520	13	12	25	3	5	1	1	4	6
ARI	0.391	13	14	27	1	2	3	3	4	5
CHI	0.469	11	11	22	0	2	2	3	2	5
MIA	0.523	8	12	20	2	3	1	2	3	5
PIT	0.627	12	11	23	0	2	2	3	2	5
BUF	0.469	11	15	26	1	2	2	2	3	4
DAL	0.547	8	17	25	0	0	3	4	3	4
DEN	0.598	9	13	22	2	2	2	2	4	4
DET	0.344	15	12	27	1	2	1	2	2	4
NYJ	0.465	13	10	23	0	0	2	4	2	4
TEN	0.535	10	14	24	0	1	2	3	2	4
BAL	0.518	15	9	24	1	1	1	2	2	3
GB	0.625	12	7	19	1	1	2	2	3	3
NE	0.641	14	13	27	0	0	2	3	2	3
NYG	0.521	11	12	23	0	1	1	2	1	3
OAK	0.414	14	12	26	0	0	2	3	2	3
PHI	0.566	12	16	28	1	1	0	2	1	3
STL	0.430	15	15	30	0	1	2	2	2	3
WAS	0.439	12	8	20	2	2	1	1	3	3
ATL	0.471	9	10	19	1	1	1	1	2	2
CAR	0.488	12	11	23	0	0	2	2	2	2
NO	0.449	12	12	24	0	0	1	2	1	2
SD	0.508	7	18	25	1	1	1	1	2	2
SEA	0.500	15	11	26	0	0	0	2	0	2
HOU	0.383	5	3	8	0	0	1	1	1	1
IND	0.625	12	12	24	1	1	0	0	1	1
JAC	0.521	13	14	27	0	0	1	1	1	1
CLE	0.335	7	8	15	0	0	0	0	0	0
TB	0.496	11	10	21	0	0	0	0	0	0
Total	--	370	378	748	25	41	45	69	70	110
r(Win Pct)	--	0.084	0.243	0.219	0.070	-0.010	0.110	0.204	0.129	0.130

More housekeeping. Pks1 and Pks2 are the total number of picks each team had in Rounds 1 and 2, respectively. FB stands for "full-on busts" and is the total number of full-on busts for each team, whereas TB stands for "total busts" and is the total number of busts (full-on plus mild) for each team. I've listed the teams in descending order of total busts.

Looking at the correlations, you can see that my theory is basically destroyed when it comes to totals. The only totals that are even remotely associated with winning percentage are (a) total # of 2nd-round picks, (b) total number of combined 1st- and 2nd-round picks, and (c) total busts in the 2nd round. Of course, the problem here is that these 3 are all positive correlations, meaning that, as the total goes up, so does winning percentage. So if anything, as it relates to busts, it turns out that we can venture to say that total busts don't matter, but winning is associated with selecting more 2nd-round busts. Obviously, that makes no sense. In other words, epic theory fail.

RETURN OF THE BUST

What about bust rates (BRs)? Maybe it's not the totals that matter, but rather the rate at which a team selects busts. Do teams that waste a higher percentage of their 1st- and 2nd-round picks on busts lose more than teams that avoid such a fate? Below is the relevant table, which is the same as the one above except it substitutes rates for totals:

Team	Win Pct	FBR1	TBR1	FBR2	TBR2	FBR	TBR
MIN	0.566	18.8%	18.8%	13.3%	46.7%	16.1%	32.3%
KC	0.508	10.0%	20.0%	33.3%	44.4%	21.1%	31.6%
CIN	0.393	23.1%	38.5%	8.3%	16.7%	16.0%	28.0%
MIA	0.523	25.0%	37.5%	8.3%	16.7%	15.0%	25.0%
SF	0.520	23.1%	38.5%	8.3%	8.3%	16.0%	24.0%
CHI	0.469	0.0%	18.2%	18.2%	27.3%	9.1%	22.7%
PIT	0.627	0.0%	16.7%	18.2%	27.3%	8.7%	21.7%
ARI	0.391	7.7%	15.4%	21.4%	21.4%	14.8%	18.5%
DEN	0.598	22.2%	22.2%	15.4%	15.4%	18.2%	18.2%
NYJ	0.465	0.0%	0.0%	20.0%	40.0%	8.7%	17.4%
TEN	0.535	0.0%	10.0%	14.3%	21.4%	8.3%	16.7%
DAL	0.547	0.0%	0.0%	17.6%	23.5%	12.0%	16.0%
GB	0.625	8.3%	8.3%	28.6%	28.6%	15.8%	15.8%
BUF	0.469	9.1%	18.2%	13.3%	13.3%	11.5%	15.4%
WAS	0.439	16.7%	16.7%	12.5%	12.5%	15.0%	15.0%
DET	0.344	6.7%	13.3%	8.3%	16.7%	7.4%	14.8%
NYG	0.521	0.0%	9.1%	8.3%	16.7%	4.3%	13.0%
BAL	0.518	6.7%	6.7%	11.1%	22.2%	8.3%	12.5%
HOU	0.383	0.0%	0.0%	33.3%	33.3%	12.5%	12.5%
OAK	0.414	0.0%	0.0%	16.7%	25.0%	7.7%	11.5%
NE	0.641	0.0%	0.0%	15.4%	23.1%	7.4%	11.1%
PHI	0.566	8.3%	8.3%	0.0%	12.5%	3.6%	10.7%
ATL	0.471	11.1%	11.1%	10.0%	10.0%	10.5%	10.5%
STL	0.430	0.0%	6.7%	13.3%	13.3%	6.7%	10.0%
CAR	0.488	0.0%	0.0%	18.2%	18.2%	8.7%	8.7%
NO	0.449	0.0%	0.0%	8.3%	16.7%	4.2%	8.3%
SD	0.508	14.3%	14.3%	5.6%	5.6%	8.0%	8.0%
SEA	0.500	0.0%	0.0%	0.0%	18.2%	0.0%	7.7%
IND	0.625	8.3%	8.3%	0.0%	0.0%	4.2%	4.2%
JAC	0.521	0.0%	0.0%	7.1%	7.1%	3.7%	3.7%
CLE	0.335	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%
TB	0.496	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%
Average	--	6.8%	11.1%	11.9%	18.3%	9.4%	14.7%
r(Win Pct)	--	0.106	0.042	0.011	0.106	0.096	0.117

I've listed teams in this table in descending total-bust-rate (TBR) order.

Looking again at the correlations, it's an even bigger epic theory fail. Clearly, busts don't matter; even when we're talking about them as a percentage of total bust-selecting opportunities. Also clear is that I'm going to start having to polish off my apology lines as promised.

BOTTOM LINE

Based on what I've presented in Part 2, here's what you should take away going into tonight's draft:

1. There aren't as many busts selected in the NFL as you think there are.
2. Whereas a team's 1st-round pick is almost guaranteed to play one day for them, the NFL graveyard is littered with 2nd-round picks who never even played a down.
3. The Niners sucked immensely from 1994-2005 when it came to avoiding busts, but...
4. Avoiding busts doesn't matter when it comes to winning in the NFL.

Stay tuned for Part 3 sometime tomorrow, wherein I'll pray that the diamond in the rough part of my theory doesn't suffer the same fate as the bust part did.

p.s. There was a lot more I could have discussed here, but time is of the essence. The draft is nigh, so feel free to continue the discussion in the comments section.