29 replies to this topic

[soxfan121]

It takes 3 years to properly evaluate whether a team has succeeded or failed with a particular draft class. By the end of Year Three, a player should be on the way to a long career, a second contract, a starting spot, a contributing role OR he's out of the league, never to be seen again.

Round 2, Pick #34 - Patrick Chung, SS
Round 2, Pick #40 - Ron Brace, DT
Round 2, Pick #41 - Darius Butler, CB (Carolina Panthers)
Round 2, Pick #58 - Sebastian Vollmer, RT
Round 3, Pick #83 - Brandon Tate, WR/KR (Cincinnati Bengals)
Round 3, Pick #97 - Tyrone McKenzie, LB (Minnesota Vikings)
Round 4, Pick #123 - Rich Ohrnberger, OG
Round 5, Pick #170 - George Bussey, OT
Round 6, Pick #198 - Jake Ingram, LS
Round 6, Pick #207 - Myron Pryor, DT
Round 7, Pick #232 - Julian Edelman, WR/CB/PR
Round 7, Pick #234 - Darryl Richard, DT

1. Eight of the twelve players picked by the Patriots in 2009 are still in the NFL, technically. Ron Brace is still on the roster but was inactive for most of the year and figures to be cut during this offseason. Myron Pryor spent 2011 on Injured Reserve, after finishing 2010 on Injured Reserve. Rich Ohrnberger spent 2011 on IR after bouncing between the active roster and the practice squad for his first two years. Tyrone McKenzie was on the Vikings practice squad in 2011. Brandon Tate was limited to kick return duties in Cincinnati, not catching a pass on offense all season. Both Chung and Volllmer missed significant portions of the 2011 season to injury. Butler started 6 games and played in 13, netting 33 tackles and 7 passes defensed.

It is somewhat fitting that the most productive on-field player in 2011 from this draft class was Julian Edelman, with 4 receptions, 28 kick returns, 4 rushing attempts and 18 tackles, playing in all three phases of the game.

2. Provided a return to health, both Chung and Vollmer have earned starting roles, with Vollmer being a borderline all-pro candidate at RT. Edelman is a valuable player who contributes everywhere, including as the emergency 4th QB, scout team QB and disproven assault allegations. I am more optimistic that Pryor claims a role with the 2012 Patriots as a rotation DL but his health is a huge question. I have no doubt Ron Brace will be looking for work elsewhere.

McKenzie has an outside shot at a roster spot in MIN; Tate seems a lock to have one as a KR specialist somewhere, most likely CIN. Butler might have carved himself a niche in Carolina - maybe he regained his confidence there, and if so, good for him. His problem in NE wasn't skill, it was mental.

3. Although 2 starters, a do-everything guy and a rotation DL is not TERRIBLE, it's reasonable to say that this was a less than perfect draft for the Patriots. The expectations for 4 2nd round picks was higher than "2 starters, a washout CB and a DT who never found the field". That said, you cannot "FAIL" a draft if you net multiple starters and both Chung & Vollmer have shown enough to be considered NFL starters, with Vollmer showing the potential for much more than that. If Vollmer's back is healthy and he returns to the Pro Bowl a couple more times (not unreasonable), then this was a better draft than it appears now.

As it is, the results merit nothing more than a solid "C".

**I'm gonna go through every draft and re-grade them. In the end, will any other team net 2 starters, a jack of all trades and a rotation guy from this draft? I'm curious to find out - and to compare everyone's "washout" percentages at the end of the exercise. Stay tuned.**

EDIT: Thanks to SuperNomario for catching the Ohrnberger mistake, now corrected.

Edited by soxfan121, 07 February 2012 - 10:28 AM.

[soxfan121]

Thanks to KiltedFool, who referred me to the link - The 2009 Steelers Draft Review

This is a really interesting draft. There were some definite hits, and seemingly a lot of misses. Less than half the picks are still on the roster. On the other hand, they picked up an incredible "class" of UFAs. At least I think it is pretty incredible when four of your UFAs are still on the team, and two are starters.

[Morgan's Magic Snowplow]

I think you're being a little harsh on this draft. You have to look at the outcomes of the other available players taken in the same round. And from that perspective, the Patriots did pretty well in 2009, a draft that produced very few good players after the first round.

Chung and Vollmer were two of the best players taken in the 2nd round (behind a few other guys, but still very, very good). Shit, they might be among the best ten players taken in the entire draft after the first round. Pryor and Edelman are also pretty clearly two of the best guys taken in rounds 6-7. We didn't get much out of rounds 3-5 but those rounds were total wastelands in this draft. Scan the list of picks and you'll only find a couple players with any real value at all.

Whether they should have traded out of the first round to begin with is another issue. But its been beaten to death so I'm just focusing here on the actual picks that were made.

Edited by Morgan's Magic Snowplow, 07 February 2012 - 09:49 AM.

[ragnarok725]

The expectations for 4 2nd round picks was higher than "2 starters, a washout CB and a DT who never found the field".

Is it, though? I know we've had this conversation all over the place, but how much can really be expected of 2nd rounders? What's the bust rate?

My perception has always been that far less than half of 2nd rounders ever become starters, much less above average ones like Chung and Vollmer.

Edited by ragnarok725, 07 February 2012 - 09:49 AM.

[underhandtofirst]

Shouldnt we be able to determine what the average draft should net based on number of teams in the NFL and avg length of career? We should be able to calculate how many new players a team must put on the field to be able to keep up with normal attrition. Also, will you include the Pats performance on Undrafted FA?

[soxfan121]

Is it, though? I know we've had this conversation all over the place, but how much can really be expected of 2nd rounders? What's the bust rate?

My perception has always been that far less than half of 2nd rounders ever become starters, much less above average ones like Chung and Vollmer.

Yes and no. First, remember that the picks were accumulated through multiple "Trade-downs", which I'm going to ignore in the review of actual picks because it's two different things. But suffice to say that while the trade downs that netted these picks eventually turned themselves into other players (like Gronk) they also removed the Pats from picks that would have given them impact 1st round picks (like Matthews). Since these things have been debated here before, I was hoping to just look at the picks.

As for the value of 2nd rounders, three of the Pats picks were in the top 50; Vollmer was #58. What's the bust rate on the top 50 players taken in any draft? It's one of the questions I've been looking at with these reviews as I've done them the last few years and the numbers fluctuate wildly depending on the year/class. For example, the Brandon Merriweather draft is one of the worst in recent memory and the bust rate is well over 50% - and Merriweather is a "success" in that calculation. But other years are deeper and thus, have fewer outright busts in the top 50. Generally, you expect to get a starter with 1st & 2nd round picks, contributors with picks 3 & 4 and lottery tickets down the line. However, everyone who read the last sentence is jumping to say that guys like Tom Brady get picked in the 6th round all the time and shitty busts are picked in the Top 50 all the time. And it's only sorta true - lots of productive players come from the bottom of the draft (or UDFA) and many top picks wash out. But it's magnified - most 6th rounders never make the league and most 1st and 2nd rounders end up starting on Sundays.

I want hard numbers on this too, which is why I'm systematically investigating it. This is part of that exercise. Any help would be appreciated.

[soxfan121]

Shouldnt we be able to determine what the average draft should net based on number of teams in the NFL and avg length of career? We should be able to calculate how many new players a team must put on the field to be able to keep up with normal attrition. Also, will you include the Pats performance on Undrafted FA?

Not in this thread. It's a good investigation to undertake and I encourage anyone with the time to do it. It's an important component of the equation, for sure. But it gets REALLY complicated and messy and I want to get through all 32 teams before the end of the month - impossible if I include UDFAs.

But please - anyone can help. Pick a team, do the analysis and post it here. As a group, we might be able to gather the data needed to get a real answer.

[Shelterdog]

Is it, though? I know we've had this conversation all over the place, but how much can really be expected of 2nd rounders? What's the bust rate?

My perception has always been that far less than half of 2nd rounders ever become starters, much less above average ones like Chung and Vollmer.

First round picks have about 75% chance of becoming starters, second round picks have about a fifty percent chance of becoming starters, third rounders one in three, fourth and fifth rounders about one in five, and after that about a 10% chance. Here's a link to a story on the topic that Casserly wrote. http://www.boston.co...rly_billic.html

It's probably a slightly below average draft class based on historic norms although the 2009 draft had a gigantic number of busts.

[Fishercat]

Is it, though? I know we've had this conversation all over the place, but how much can really be expected of 2nd rounders? What's the bust rate?

My perception has always been that far less than half of 2nd rounders ever become starters, much less above average ones like Chung and Vollmer.

I think you're right, but as Soxfan mentions, while they were four second round picks, three of them were early in that round. I think if you go to the bottom of the second round where we usually are, you're right. But let's take the top of R2 2009 as an example

33 - Louis Delmas (borderline pro bowl safety, solid player)
34 - Patrick Chung (we know)
35 - James Laurinitis (Not a Rams fan, but I hear good things and he seems to be all over the field whenever I see him)
36 - Brian Robiskie (second team in three years)
37 - Alphonso Smith (Can't speak to his quality, haven't seen him enough, but he at least still starts)
38 - Rey Maualuga (Starts most games for a solid Cincy LB corps)
39 - Eben Britton (It's really hard to judge O-Line play for me, I know he's had injury issues and been moved to guard, so probably a bust as of right now)
40-41: Brace and Butler

After that, there are some good players (Lesean McCoy, Jairus Byrd, Connor Barwin, Sebastian Vollmer) and a lot of guys we haven't really hard much from. I'm guessing that draft just had some circumstances that resulted in the drop-off.

Personally, if you told me that our four second rounders that year would give us a plus OT and a good safety (albet injury prone), it's a pretty good day. I guess we could have ended up like the Browns with Massaquoi, Robiskie, and Veikune. Shelter beat me to the Casserly article, though I wish we had a little more in-depth info. A starter after four years: does that mean a player who started at some point within those four years, or someone who started on Day 1, Season 4?

[soxfan121]

First round picks have about 75% chance of becoming starters, second round picks have about a fifty percent chance of becoming starters, third rounders one in three, fourth and fifth rounders about one in five, and after that about a 10% chance. Here's a link to a story on the topic that Casserly wrote. http://www.boston.co...rly_billic.html

It's probably a slightly below average draft class based on historic norms although the 2009 draft had a gigantic number of busts.

I think those approximations can be firmed up by really digging in and doing the analysis; it's one of the few things in football that could be investigated systematically. Further, the 50% percentage is for the whole second round, not for the top 50 players drafted (where I'd guess the number is closer to late first round than it is to late second round). Anyway, maybe Casserly is right and those 75/50/33 numbers prove to be true across a decade of drafts.

[Shelterdog]

I think you're right, but as Soxfan mentions, while they were four second round picks, three of them were early in that round. I think if you go to the bottom of the second round where we usually are, you're right. But let's take the top of R2 2009 as an example

33 - Louis Delmas (borderline pro bowl safety, solid player)
34 - Patrick Chung (we know)
35 - James Laurinitis (Not a Rams fan, but I hear good things and he seems to be all over the field whenever I see him)
36 - Brian Robiskie (second team in three years)
37 - Alphonso Smith (Can't speak to his quality, haven't seen him enough, but he at least still starts)
38 - Rey Maualuga (Starts most games for a solid Cincy LB corps)
39 - Eben Britton (It's really hard to judge O-Line play for me, I know he's had injury issues and been moved to guard, so probably a bust as of right now)
40-41: Brace and Butler

After that, there are some good players (Lesean McCoy, Jairus Byrd, Connor Barwin, Sebastian Vollmer) and a lot of guys we haven't really hard much from. I'm guessing that draft just had some circumstances that resulted in the drop-off.

. Shelter beat me to the Casserly article, though I wish we had a little more in-depth info. A starter after four years: does that mean a player who started at some point within those four years, or someone who started on Day 1, Season 4?

It's not clear exactly what Casserly's criteria were--it looks like it was guys he thought counted as starters in their fourth season. Without a doubt someone could improve on his numbers. (Something like clumping the draft into 10-pick clusters and seeing the average number of games played in each cluster would be a pretty good way to judge it over time).

Alphonso Smith got cut from Denver, and a bunch of the players picked in the 40s that year (Everette Brown, Pat White, Clint Sintim, Darcel McBath) have done zero as pros.

EDIT: Oops, Smith got traded not cut. Nice catch.

Edited by Shelterdog, 07 February 2012 - 10:30 AM.

[Super Nomario]

Ohrnberger I think it technically still on the team; he was on IR this year.

[Fishercat]

FWIW, Alphonso Smith was traded from Denver, but it was for a nearly cuttable player (Dan Gronkowski) so I think your point's fair. He's found his way on to the field since then (although I think I was wrong when I said he starts, that was 2010) in Detroit. I'm guessing you would want more from the 35th pick, seems a bit comparable to Darius Butler in that sense.

Edited by Fishercat, 07 February 2012 - 10:28 AM.

[soxfan121]

Ohrnberger I think it technically still on the team; he was on IR this year.

Correct - thank you. First post updated.

[Shelterdog]

Ohrnberger I think it technically still on the team; he was on IR this year.

It seems that the view him as a potential center of the future. I haven't really seen it in his limited play myself but who knows.

[bowiac]

I think those approximations can be firmed up by really digging in and doing the analysis; it's one of the few things in football that could be investigated systematically. Further, the 50% percentage is for the whole second round, not for the top 50 players drafted (where I'd guess the number is closer to late first round than it is to late second round). Anyway, maybe Casserly is right and those 75/50/33 numbers prove to be true across a decade of drafts.

I'm actually doing a bit of work on this right now. Still waiting for some 2011 data to be added to the dataset before I can get really in depth, but Casserly's numbers are broadly in line with what I've found. I get slightly lower numbers across the board, since I'm differentiating between 3 year starters and 6 year starters for two guys drafted in 2006, but that ends up being pretty uniformly between 5-10% lower.

Something like 63% of seasons from 1st round picks result in that player being a starter. That number drops to around 43% for 2nd rounders, and down to about 25% for third round picks.

I'm going to try to have a bigger more complete thing done soon though that will differentiate between being a starter, being a contributor, and being a star. It's probably more useful, because I don't really trust the database I'm using for whether someone is a starter or not. Trying to create a modern draft value chart as well to analyze trades.

[soxfan121]

I'm actually doing a bit of work on this right now. Still waiting for some 2011 data to be added to the dataset before I can get really in depth, but Casserly's numbers are broadly in line with what I've found. I get slightly lower numbers across the board, since I'm differentiating between 3 year starters and 6 year starters for two guys drafted in 2006, but that ends up being pretty uniformly between 5-10% lower.

Something like 63% of seasons from 1st round picks result in that player being a starter. That number drops to around 43% for 2nd rounders, and down to about 25% for third round picks.

I'm going to try to have a bigger more complete thing done soon though that will differentiate between being a starter, being a contributor, and being a star. It's probably more useful, because I don't really trust the database I'm using for whether someone is a starter or not. Trying to create a modern draft value chart as well to analyze trades.

Looking forward to it. PM me if I can fit my effort into your larger project; sounds like you're further along and I'd be up to help.

[ZMart100]

I'm not sure how we are defining a player being a "starter." I do have data on player's first four years from the 1995-2008 drafts (the most recent eligible players). I have done a regression of the games started by draft pick in those first four years. I have plotted the first four years with the regression and converted it into average starts/season for the first 240 picks and put that under the second spoiler.


Some words of caution and some words of math:

Spoiler

1) I have made no adjustment for positional value (though if I can figure out a nice way to do it, I may in the future). In other words a fullback who starts 62 games = a quarterback who starts 62 games. An adjustment would likely increase the value of high picks and depress the value of lower picks.
2)I made no attempt to adjust for $$$. This obviously matters due to the constraints of the salary cap.
3)I have discarded P and Ks.
4)I kept both QBs and DBs. This matters because I suspect that injuries requiring missing an odd game or two are not distributed among positions in the same way.
5) The standard deviation is high, a little over 1 game. Also, note that the estimates of the first 10 picks are about 1 game under their actual production. On the other hand picks 85-95 are overestimated by about a game. It is possible (although untested) that the first 10 picks are more likely to see the field when they shouldn't because they are a top 10 pick. It is also possible that the model is not well specified or that it is just a fragment of a small sample size. I remain agnostic with respect to those possibilities.
6) The actual math. I'm essentially taking the functional form from the weibull distribution [wikipedia.org]. Its a little tricky to estimate. I have tried a number of forms and find this both the most intellectually satisfying as well as the best fitting. That said, I'm still not 100% satisfied. Any ideas would be appreciated.
7)Starts are a good predictor of free agent salary, so I am using them as a proxy for skill and health. I am aware that a Tom Brady start is not the same as a Julien Edelman on defense start, but I see no better way to measure talent across positions and skill sets.

A picture:
Circles represent average games started for each draft position


A chart:

Spoiler

Average Starts per Season - First 4 Years

Pick #	Starts/Season	Pick #	Starts/Season	Pick #	Starts/Season	Pick #	Starts/Season
1	11.16	61	6.23	121	3.43	181	1.86
2	11.06	62	6.17	122	3.40	182	1.84
3	10.95	63	6.11	123	3.36	183	1.82
4	10.85	64	6.05	124	3.33	184	1.80
5	10.74	65	5.99	125	3.30	185	1.78
6	10.64	66	5.94	126	3.26	186	1.76
7	10.54	67	5.88	127	3.23	187	1.75
8	10.44	68	5.82	128	3.20	188	1.73
9	10.34	69	5.77	129	3.17	189	1.71
10	10.25	70	5.71	130	3.14	190	1.69
11	10.15	71	5.65	131	3.10	191	1.67
12	10.05	72	5.60	132	3.07	192	1.66
13	9.96	73	5.55	133	3.04	193	1.64
14	9.86	74	5.49	134	3.01	194	1.62
15	9.77	75	5.44	135	2.98	195	1.60
16	9.68	76	5.39	136	2.95	196	1.59
17	9.59	77	5.34	137	2.92	197	1.57
18	9.50	78	5.28	138	2.89	198	1.55
19	9.41	79	5.23	139	2.86	199	1.54
20	9.32	80	5.18	140	2.84	200	1.52
21	9.23	81	5.08	141	2.81	201	1.50
22	9.14	82	5.03	142	2.78	202	1.49
23	9.05	83	4.98	143	2.75	203	1.47
24	8.97	84	4.94	144	2.72	204	1.45
25	8.88	85	4.89	145	2.70	205	1.44
26	8.80	86	4.84	146	2.67	206	1.42
27	8.71	87	4.79	147	2.64	207	1.41
28	8.63	88	4.75	148	2.61	208	1.39
29	8.55	89	4.70	149	2.59	209	1.38
30	8.47	90	4.66	150	2.56	210	1.36
31	8.39	91	4.61	151	2.54	211	1.35
32	8.31	92	4.57	152	2.51	212	1.33
33	8.23	93	4.52	153	2.49	213	1.32
34	8.15	94	4.48	154	2.46	214	1.30
35	8.07	95	4.43	155	2.43	215	1.29
36	8.00	96	4.39	156	2.41	216	1.28
37	7.92	97	4.35	157	2.39	217	1.26
38	7.84	98	4.30	158	2.36	218	1.25
39	7.77	99	4.26	159	2.34	219	1.23
40	7.70	100	4.22	160	2.31	220	1.22
41	7.62	101	4.18	161	2.29	221	1.21
42	7.55	102	4.14	162	2.27	222	1.19
43	7.48	103	4.10	163	2.24	223	1.17
44	7.41	104	4.06	164	2.22	224	1.15
45	7.34	105	4.02	165	2.20	225	1.14
46	7.27	106	3.98	166	2.17	226	1.13
47	7.20	107	3.94	167	2.15	227	1.12
48	7.13	108	3.90	168	2.13	228	1.10
49	7.06	109	3.86	169	2.11	229	1.09
50	6.99	110	3.82	170	2.09	230	1.08
51	6.92	111	3.79	171	2.06	231	1.07
52	6.86	112	3.75	172	2.04	232	1.05
53	6.79	113	3.71	173	2.02	233	1.04
54	6.73	114	3.68	174	2.00	234	1.03
55	6.66	115	3.64	175	1.98	235	1.02
56	6.54	116	3.60	176	1.96	236	1.01
57	6.47	117	3.57	177	1.94	237	0.99
58	6.41	118	3.53	178	1.92	238	0.98
59	6.35	119	3.50	179	1.90	239	0.97
60	6.29	120	3.46	180	1.88	240	0.96

[dbn]

One can use ZMart100's results (nice job, by the way) to do two quick-and-dirty -- but probably informative -- tests of the draft. First, compare the actual total number of starts to the total expected number for the actual draft positions using ZMart's regression. This says something about how well they did at picking the players given when in the draft they chose. (N.B. that this isn't the same as ability to evaluate talent because a pick might have been wise but potential starts lost to injury.) Alternatively, one can compare the actual total number of starts to the expected number given where their original assigned + compensatory picks were. This tells you something about the quality of the draft as a whole, including the wisdom (or lack thereof) of any trades that were made.

edit: after some quick googling, it seems the Patriots were assigned the 23rd, 58th, 89th, 97th (compensatory), 124th, 170th (comp), 199th (how could we trade away the 199th!!!), 207th (comp) and 234th. Using the numbers from ZMart100's chart, one would predict: 9.05 + 6.41 + 4.70 + 4.35 + 3.33 + 2.09 + 1.54 + 1.41 + 1.03 = 33.91 starts per season.
Actual starts/3: Chung = 7.33; Brace = 2.33; Butler = 4.67; Vollmer = 9.67; Tate = 3.67; McKenzie = 0.00; Ohrnberger = 0.00; Bussey = 0.00; Ingram = 0.00; Pryor = 0.67; Edelman = 3.33; Richard = 0.00; sum = 31.67 which is close to what was expected, so a "C" grade might be pretty accurate.
Starts data according to NFL.com.

Edited by dbn, 09 February 2012 - 09:33 AM.

[Marbleheader]

This is the killer of this draft to me. Made the wrong choice, this draft would have been much different.

83. New England Patriots Brandon Tate WR North Carolina
84. Pittsburgh Steelers Mike Wallace WR Ole Miss

[Super Nomario]

82. Detroit Lions Derrick Williams WR Penn St.
83. New England Patriots Brandon Tate WR North Carolina
84. Pittsburgh Steelers Mike Wallace WR Ole Miss
85. New York Giants Ramses Barden WR Cal Poly - SLO

Added a little more context.

It would have been awesome to have drafted Wallace, but at that point in the draft it's pretty much a crapshoot.

[DaubachmanTurnerOD]

I'm very surprised to see in ZMart's chart that the average starts per draft position decreases steadily without a single outlier (that I noticed) where a later draft position had more average starts than an earlier position.

Is there an obvious explaination for this that I'm missing? Is that a chart of 'expected' starts? Or have teams, on average, just been that good at evaluating talent?

[Shelterdog]

I'm very surprised to see in ZMart's chart that the average starts per draft position decreases steadily without a single outlier (that I noticed) where a later draft position had more average starts than an earlier position.

Is there an obvious explaination for this that I'm missing? Is that a chart of 'expected' starts? Or have teams, on average, just been that good at evaluating talent?

The chart is expected starts by draft position based on his model, not the actual average starts by draft position (which would be very interesting to see, by the way).

[ZMart100]

The chart is expected starts by draft position based on his model, not the actual average starts by draft position (which would be very interesting to see, by the way).

Yeah, I could have made that clearer. Below are the observed average games started per season from drafts 1995-2008 for the first 240 picks.

Spoiler

Games Started per Season over First Four Years

Pick #	Starts/Season	Pick #	Starts/Season	Pick #	Starts/Season	Pick #	Starts/Season
1	12.18	61	7.89	121	2.73	181	3.15
2	11.98	62	5.46	122	3.64	182	0.96
3	12.84	63	5.75	123	3.02	183	2.02
4	11.91	64	6.86	124	4.84	184	1.11
5	11.52	65	4.63	125	3.73	185	1.64
6	10.80	66	8.04	126	5.45	186	0.73
7	12.18	67	4.98	127	2.38	187	2.91
8	11.77	68	5.82	128	3.34	188	1.79
9	11.86	69	4.29	129	2.41	189	2.19
10	10.63	70	5.02	130	2.55	190	2.46
11	11.25	71	7.43	131	2.95	191	2.04
12	10.32	72	3.61	132	3.05	192	0.81
13	11.13	73	4.71	133	1.98	193	1.02
14	11.21	74	2.46	134	2.82	194	1.13
15	9.36	75	5.36	135	2.05	195	0.96
16	8.59	76	2.61	136	3.89	196	1.93
17	8.85	77	2.59	137	3.65	197	1.70
18	8.25	78	6.23	138	3.38	198	2.27
19	11.70	79	2.93	139	3.07	199	3.16
20	11.16	80	5.23	140	3.66	200	2.84
21	10.20	81	4.61	141	4.00	201	0.75
22	7.61	82	1.94	142	1.56	202	1.54
23	10.48	83	4.21	143	3.13	203	0.50
24	10.04	84	4.46	144	2.43	204	1.27
25	8.63	85	3.68	145	2.14	205	0.93
26	7.96	86	3.54	146	2.93	206	1.16
27	8.36	87	3.38	147	1.30	207	2.71
28	9.57	88	3.21	148	1.04	208	1.93
29	7.34	89	4.13	149	3.98	209	2.13
30	8.50	90	3.37	150	0.66	210	0.71
31	6.38	91	4.89	151	1.95	211	0.30
32	7.86	92	5.48	152	3.09	212	0.84
33	8.95	93	3.63	153	2.50	213	1.20
34	10.32	94	4.82	154	2.39	214	1.00
35	10.13	95	2.50	155	2.29	215	1.93
36	7.45	96	4.55	156	2.89	216	1.17
37	8.80	97	4.36	157	1.86	217	0.48
38	8.45	98	7.12	158	2.09	218	1.11
39	8.79	99	2.73	159	1.75	219	2.39
40	7.80	100	4.89	160	4.06	220	1.16
41	6.11	101	3.25	161	1.64	221	1.30
42	9.52	102	4.19	162	1.70	222	0.77
43	9.39	103	1.64	163	0.50	223	1.14
44	8.13	104	2.98	164	2.46	224	4.40
45	7.07	105	3.57	165	1.55	225	0.50
46	6.09	106	5.23	166	1.93	226	1.48
47	6.35	107	1.73	167	1.80	227	1.16
48	9.63	108	3.64	168	2.71	228	1.35
49	4.86	109	5.67	169	1.88	229	2.19
50	7.75	110	3.16	170	1.23	230	2.45
51	5.96	111	3.96	171	0.59	231	0.77
52	7.00	112	3.46	172	1.29	232	1.23
53	7.18	113	4.25	173	2.95	233	1.14
54	5.45	114	3.27	174	2.87	234	0.34
55	8.43	115	2.48	175	2.00	235	0.46
56	4.79	116	2.94	176	0.75	236	0.38
57	7.18	117	2.07	177	2.32	237	1.41
58	6.70	118	3.98	178	1.42	238	1.13
59	7.04	119	5.34	179	1.61	239	1.54
60	7.14	120	3.62	180	3.67	240	1.29

edit: grammar

Edited by ZMart100, 09 February 2012 - 10:47 PM.

[weeba]

Pick 224 is quite the outlier there, with 4.4 starts. Are there a few players driving that?

Edited by weeba, 10 February 2012 - 08:25 AM.

[underhandtofirst]

Pick 224 is quite the outlier there, with 4.4 starts. Are there a few players driving that?

2000 Mark Tauscher T GB
2001 Eric Johnson TE SF
2008 Stevie Johnson WR Buf

[Morgan's Magic Snowplow]

I'm not sure how we are defining a player being a "starter." I do have data on player's first four years from the 1995-2008 drafts (the most recent eligible players). I have done a regression of the games started by draft pick in those first four years. I have plotted the first four years with the regression and converted it into average starts/season for the first 240 picks and put that under the second spoiler.

This analysis is awesome, ZMart, and hasn't gotten the attention and applause it deserves.

One small comment regarding your note about certain groups of picks being over or underestimated by the model: How many starts a player gets in his first four years is likely to depend somewhat on the quality of the team drafting him. At least in the first few rounds, I think picks at the top of rounds are likely to be underestimated by the model and then picks at the bottom of the round are likely to be overestimated by model. In addition to helping account for the patterns you mentioned around picks 1-10 and 85-95, this effect may also help explain why the start totals in the first half of the second round are higher than those in the last quarter of the first round. I assume that as rounds go by, this effect diminishes: At some point, everybody sucks and few players can win starting jobs even with bad teams.

[Shelterdog]

This analysis is awesome, ZMart, and hasn't gotten the attention and applause it deserves.

One small comment regarding your note about certain groups of picks being over or underestimated by the model: How many starts a player gets in his first four years is likely to depend somewhat on the quality of the team drafting him. At least in the first few rounds, I think picks at the top of rounds are likely to be underestimated by the model and then picks at the bottom of the round are likely to be overestimated by model. In addition to helping account for the patterns you mentioned around picks 1-10 and 85-95, this effect may also help explain why the start totals in the first half of the second round are higher than those in the last quarter of the first round. I assume that as rounds go by, this effect diminishes: At some point, everybody sucks and few players can win starting jobs even with bad teams.

An interesting-but difficult-question would be whether there is any correlation between having your draft picks start more games than expected and winning. At first glance you'd think the answer is yes-if your draft picks start a lot of games that means you made good picks-but then it could also mean that you're playing the wrong guys (you're playing a shitty second round pick rather than a better UDFA or waiver wire guy).

[ZMart100]

Pick 224 is quite the outlier there, with 4.4 starts. Are there a few players driving that?

Here is what my dataset looks for for pick 224. Keep in mind these are the total numbers for 4 years. Also notice that 2004 is absent because the pick was P Donnie Jones. I'm posting this partially to answer you're question about 224 and partially to give others an idea of what other questions I can answer. Data taken from multiple sources. Again, I'm hiding it beneath a spoiler to keep the post compact.

Spoiler

Pick 224

Draft.Year	Pick.No	Player	Pos	Games	Starts	Pro.Bowls	All.Pros	First.Round	School	Team
1995	224	Wesley Leasy	LB	28	0	0	0	0	Mississippi State	Arizona Cardinals
1996	224	Darius Smith	C	0	0	0	0	0	Sam Houston State	Oakland Raiders
1997	224	Omar Stoutmire	DB	60	19	0	0	0	Fresno State	Dallas Cowboys
1998	224	Ernest Blackwell	RB	0	0	0	0	0	Missouri	Kansas City Chiefs
1999	224	JoJuan Armour	DB	38	19	0	0	0	Miami (Ohio)	Oakland Raiders
2000	224	Mark Tauscher	T	50	48	0	0	0	Wisconsin	Green Bay Packers
2001	224	Eric Johnson	TE	57	47	0	0	0	Yale	San Francisco 49ers
2002	224	Derrius Monroe	DE	0	0	0	0	0	Virginia Tech	New Orleans Saints
2003	224	Taco Wallace	WR	4	0	0	0	0	Kansas State	Seattle Seahawks
2005	224	Jay Ratliff	DE	50	31	1	0	0	Auburn	Dallas Cowboys
2006	224	E.J. Whitley	C	0	0	0	0	0	Texas Tech	Dallas Cowboys
2007	224	Michael Johnson	DB	49	35	0	0	0	Arizona	New York Giants
2008	224	Steve Johnson	WR	48	30	0	0	0	Kentucky	Buffalo Bills

An interesting-but difficult-question would be whether there is any correlation between having your draft picks start more games than expected and winning. At first glance you'd think the answer is yes-if your draft picks start a lot of games that means you made good picks-but then it could also mean that you're playing the wrong guys (you're playing a shitty second round pick rather than a better UDFA or waiver wire guy).

That is a good question. There are a couple of complications to think about as well. 1) I have seen a study suggesting that there is no correlation (I think it was even slightly negative, but not significant) between years of a team's performance against expectations in the draft. So it would be hard to tell which draft was driving a team's win percentage. 2) My dataset might not be well suited to tackle this question because I aggregate games played for all teams over the first 4 years. So team A could have a "overperforming" draft but only because they had a 6th round pick who started 30 games for team B and I would not be able to see the difference. 3) I think the mechanism one would expect is a coach, or GM, or Al Davis-type thinking, "he's a 1st round talent, he must be a better player than the other guy!" So maybe the best way to look at the question would be to isolate performance against expectations in the first round or two rather than taking the whole draft. Also, it might make sense to look at it by coach/gm and not by team. Ideas?

[Super Nomario]

Ohrnberger I think it technically still on the team; he was on IR this year.

OK, never mind