Writers of Pro Football Prospectus 2008

23 Apr 2014

FO Mailbag: "Team-Adjusted" Run Stop Rates

Tuluse: I was reading through the comments for the article on Run Stop Rate, and realized that it's a flawed stat because it doesn't take into account a player completely whiffing a tackle. I know you have other stats to try to account for this based on the charting project. However, I wonder if you would indulge me and post a stop rate based on the total number of runs a team faced. Getting 20 stops is a lot more impressive on a good run defense that doesn't see many runs vs the Bears for example where teams were running whenever and however they could.

I know this would be a flawed stat as well because it doesn't account for plays when the player is not on the field, but you could adjust for games where the player actually played and maybe even use the participation data (if a defender plays for 80 percent of snaps and the team faced 20 runs, assume they were on the field for 16 runs). I know it's not perfect, but football stats never are and it would be interesting to see.

I've actually been sitting on this a while. Suddenly got the urge to answer it today, perhaps because I'm excited to finally be done with broken tackle stats, which will run sometime in the next couple weeks and sort of tie into the idea here.

We'll run these quick, so not a lot of commentary from me. I'll let the readers provide commentary by comparing these numbers to the Run Stop Rate numbers I ran a few weeks ago, which again are here.

I started here with each player's percentage of defensive snaps by dividing defensive snaps by the number of defensive plays his team faced only in games where he was active. Then we multiply that number by the number of runs a team faced (including scrambles, but not including kneeldowns). That gives us the estimated number of runs when each player was on the field. I used that to create two stats. Adjusted Percent of Team Run Plays divides a player's total run tackles (including assists) by that estimated number of runs faced. Adjusted Run Stop Rate divides a player's run Stops (plays that prevent a successful offensive gain) by that estimated number of runs faced.

We're going to get very different lists here because now we are more measuring how many plays a defender was involved in rather than how often he made a successful tackle. But there are differences. For example, the list of the top linebackers shows how strong Lavonte David was against the run this year.

"Adjusted Run Stop Rate" for Linebackers
Player Team Pct Def Snaps Adj Pct
Team Run Plays
Adj Run
Stop Rate
53-N.Bowman SF 97.7% 23.5% 16.1%
50-K.Alonso BUF 100.0% 25.6% 15.8%
54-L.David TB 96.5% 18.6% 15.2%
55-V.Burfict CIN 95.9% 22.8% 14.5%
59-L.Kuechly CAR 97.6% 23.9% 14.3%
94-L.Timmons PIT 100.0% 17.9% 13.8%
51-P.Posluszny JAC 98.7% 19.8% 13.8%
55-S.Tulloch DET 99.0% 22.1% 13.4%
50-C.Lofton NO 95.8% 18.7% 13.4%
52-D.Jackson CLE 99.9% 19.8% 13.2%
52-D.Harris NYJ 99.8% 17.9% 13.0%
56-K.Dansby ARI 99.6% 19.2% 12.8%

With the defensive linemen, you get a good idea once againg of how awesome J.J. Watt is, but also how important the defensive ends were to the run defenses for the Jets and Patriots.

"Adjusted Run Stop Rate" for Defensive Linemen
Player Team Pct Def Snaps Adj Pct
Team Run Plays
Adj Run
Stop Rate
99-J.Watt HOU 94.1% 13.6% 12.5%
50-R.Ninkovich NE 95.5% 14.6% 10.7%
93-C.Campbell ARI 89.0% 11.7% 10.2%
99-L.Houston OAK 95.0% 11.7% 8.8%
91-S.Richardson NYJ 80.0% 11.8% 8.4%
95-Cha.Jones NE 97.9% 13.1% 8.3%
94-A.Clayborn TB 88.0% 11.1% 8.2%
96-M.Wilkerson NYJ 94.5% 10.2% 8.1%
96-C.Dunlap CIN 87.4% 9.5% 8.1%
95-K.Williams BUF 82.0% 9.7% 8.0%

Things get more interesting with defensive backs. With linebackers and defensive linemen, the leaders in "Adj Pct Team Run Plays" were basically the same as the leaders in "Adj Run Stop Rate," just in a different order. However, because a lot of defensive backs are stuck making clean-up tackles, you get some real differences with cornerbacks and safeties. I've taken these top tens and added (in italics) a couple of players who made a huge number of run plays for their teams but didn't make a lot of run Stops because those plays were so far downfield.

I have no idea why Josh Wilson played such a huge role in the Washington run defense this year, and why he made his plays so much closer to the line of scrimmage than DeAngelo Hall.

"Adjusted Run Stop Rate" for Cornerbacks
Player Team Pct Def Snaps Adj Pct
Team Run Plays
Adj Run
Stop Rate
26-J.Wilson WAS 94.5% 7.7% 4.3%
38-T.Williams GB 98.7% 6.5% 3.5%
23-T.Porter OAK 91.7% 5.9% 3.5%
41-C.Munnerlyn CAR 95.8% 6.3% 3.1%
21-L.Webb BAL 90.3% 5.1% 2.8%
20-P.Amukamara NYG 95.5% 6.1% 2.4%
22-W.Gay PIT 85.0% 3.9% 2.2%
33-C.Tillman CHI 84.4% 3.8% 2.1%
24-D.Revis TB 89.5% 3.6% 2.1%
25-C.Harris DEN 92.3% 3.8% 2.0%
29-B.Ross OAK 88.8% 5.9% 1.9%
32-T.Mathieu ARI 88.1% 5.4% 1.7%
23-D.Hall WAS 97.6% 5.2% 1.6%

I think I counted Mathieu as a safety when I originally ran Run Stop Rates. Now we've decided to mark him as a cornerback. It's hard to put him in one place or the other, because he played so much of both. But with almost every offense in the NFL now using three or more wide receivers on a majority of plays, I figure he's more a nickelback than a free safety. 

"Adjusted Run Stop Rate" for Safeties
Player Team Pct Def Snaps Adj Pct
Team Run Plays
Adj Run
Stop Rate
43-T.Ward CLE 96.9% 14.6% 8.6%
31-K.Chancellor SEA 96.5% 13.6% 7.8%
25-R.Clark PIT 99.2% 15.6% 7.1%
32-J.Ihedigbo BAL 99.5% 13.4% 7.1%
26-D.Landry NYJ 98.3% 11.1% 5.8%
41-A.Bethea IND 97.2% 14.4% 5.8%
31-B.Pollard TEN 98.9% 10.0% 5.3%
37-Y.Bell ARI 98.9% 10.5% 5.2%
42-M.Burnett GB 99.8% 13.2% 5.0%
23-M.Barron TB 89.2% 10.0% 4.7%
42-B.Church DAL 89.1% 14.0% 3.4%
29-E.Thomas SEA 96.6% 13.3% 3.8%

Posted by: Aaron Schatz on 23 Apr 2014

4 comments, Last at 23 Apr 2014, 5:48pm by Mike B. In Va


by nat :: Wed, 04/23/2014 - 2:13pm

The link to the previous article is broken. The url should be http://www.footballoutsiders.com/stat-analysis/2014/run-stop-rates-2013

Congrats on a much improved stat analysis. The old article compared defeats to tackles, which was wrong and gave mostly meaningless results. After all, once the tackle happens, there is little the defender can do to make it a defeat or not.

Basically, a defender could get a 100% rating in the old stat by running across the line badly out of position and getting an occasional tackle by pure luck. Giving up on plays is a plus under the old stat, which is a sign it was broken from the beginning.

This new, better stat is much closer to the mark. Each running play with the defender on the field is an opportunity for him to make a tackle and to make a defeat. Your estimate of the play count may not be perfect, but it's the right kind of denominator for a good rate stat.

To get a high score, the defender has to have a knack for getting to the ball carrier quickly, and making the tackle without losing a lot of ground.

Well done.

by theslothook :: Wed, 04/23/2014 - 3:45pm


by tuluse :: Wed, 04/23/2014 - 5:01pm

Thanks a bunch for answering my question Aaron!

Looks like the new list weeds out the Bears linebackers who were failing to make tackles at all on an alarming number of plays.

I'd be really curious to see the cornerback stats over a number of years and see how consistent they are. Some of the Lovie Smith years it felt like the run defense was designed around Briggs, Urlacher, and Tillman each being responsible for 1/3 of the field.

by Mike B. In Va :: Wed, 04/23/2014 - 5:48pm

This makes me wonder if moving Alonso outside might be a mistake. I would not have guessed he'd be that high.