Revisiting Havoc Rate and the New Defensive Coordinator

VTM member mohave wrote this post and I'm publishing it.

Texas Tech is looking for a new defensive coordinator. That is a sentence that Red Raider fans thought they were finally finished hearing, with Matt Wallerstedt coming into 2014 season with two offseasons and 13 games coaching the Red Raider defense to his credit. What happened next is a story that Tech fans probably would prefer not to relive in its particulars, and are all too familiar with in its generalities: results on the field that ranged from lackluster to disastrously bad, and intra-program friction that culminated with Wallerstedt's dismissal amid media rumors of personal misconduct.

In the midst of this chaos, Seth authored an extended brainstorm session on how to fix the defense, and in it he referenced a statistic called havoc rate, a metric that?Bill Connelly uses as a very raw, very simple measure of how disruptive a defense is. I'll let him explain:

Havoc rate is a pretty simple method for looking at how much hell a defense is raising. Add up tackles for loss (which includes sacks), forced fumbles, and defensed passes (picks and break-ups), divide it by total plays, and voila: havoc rate.

It's a great little stat because it captures a very specific type of data- while it comes nowhere close to measuring the full story of a defense's quality, what it does do is give you a sense of how successful a defense is at actively mucking up the offense's plan. This is valuable information to us because at its most basic level, football defense is about the interplay between two tactics: trying to force the offense to make mistakes, and sitting back and allowing the offense to make mistakes on its own. In an era when offenses are more efficient and higher scoring than ever, the more disruptive your defense is, generally speaking, the better. Having a strong havoc rate does not on its own make you a good defense, but there is a strong correlation between havoc rate and defensive efficiency. In?Football Outsider's Defensive S&P+ rankings for the 2014 college football season, only five teams that were ranked in the top 50 for Defensive S&P+ had a havoc rate below 15%. By the same token, of teams ranked 76th and worse (128 teams in FBS this year), only 13 had havoc rates above 15%. So, while it is possible to have an atrocious defense and a decent havoc rate, it is rare to have a good or excellent defense that does not.

One other reason I like this stat is because it lets us play along at home without having to keep miles of play-by-play data spreadsheets updated. What I want to do is follow up on Seth's thoughts, given a full season's worth of results, and meditate a little on what havoc rate can tell us about a season when two different defensive coordinators got to take a crack at coaching the same group of players with the same general scheme.

So, to see where we're at, let's look at where we've been. I went back and looked at 2013's defense, and calculated a havoc rate of 16.26%, which is slightly above 2013's FBS-wide average of 15.9%. With several impact seniors exhausting their eligibility at the end of 2013, it would have been reasonable, if pessimistic, to expect that number to take a step back along with the overall quality of the defense, but the first three games of 2014 were still a shock for Red Raider fans.

Seth calculated the havoc rate for Wallerstedt's three games in 2014 as part of the piece I linked, and I'm taking the liberty of reproducing them here, complete with his very nice table:

I'll go ahead and tell you the average havoc rate of those three games was 8.7%, but you don't need the exact number to see that it was terrible and trending in the wrong direction. Kingsbury, in probably the toughest decision he's had to make in his short time as a head coach, made the call to move on from Wallerstedt, and give Mike Smith the task of trying to salvage the remainder of the season. We know that in terms of results, he, the defense, and the team as a whole had a rough go of it, winning only two more games, and finding themselves on the wrong end of a few record-setting offensive performances by conference rivals. We also know that his defense performed admirably in some tough spots, often without much support from a sputtering, turnover-prone offense, and that certain players like Pete Robertson, who seemed invisible to start the year, experienced amazing turnarounds that almost perfectly coincided with Smith taking over.

What we want to know is, despite the lack of wins down the stretch, has Mike Smith made enough of a mark on this defense to earn the job? And how would you even quantify that, when he's locked into a framework and depth chart another coach had final say on? Let's recall what havoc rate measures: when fumbles are forced, when passes are tipped or picked, when the quarterback or ballcarrier are tackled behind the line of scrimmage, these are all indicators of a defense disrupting the offense's plan. Excellent athletic talent can produce these kinds of plays independent of scheme or coaching, but what if we have a set of data where the player talent stays the same or erodes in quality due to injuries/suspensions, and the only variables we have are in how the players are coached and motivated? In this case, havoc rate can provide an insight into just how much more effective Smith was than his predecessor.

I went ahead and put together a spreadsheet to calculate havoc rate for the season, individual games, as well as before and after the coaching change. It's got all the information you could possibly require, but here are the essentials:

What jumps out immediately is that right from the start, Mike Smith's version of this defense is much more disruptive overall than Wallerstedt's, even when they aren't playing well. This isn't surprising in and of itself, as Smith runs less base 3-4, calls more blitzes, and generally has a more dynamic approach to defense than Wallerstedt did. What is surprising is that this wasn't Wallerstedt's approach as well. It makes all the sense in the world as a defensive strategy for a team that's thin on depth and talent: over the long haul, you are probably going to get ground down and lose too many individual matchups, but with the right application of aggression you can potentially spring big plays that will make up for some of your deficiencies. We saw ?this a few times during the season when Sam Eguavoen and Pete Robertson were able to get pressure on passing downs by selling out on a blitz- when you have players who aren't succeeding in the base defense, a creative coach can still get some production from them by giving them an easier job than read and react: get into the backfield and ruin someone's day.

We also see, unfortunately, that this strategy doesn't always pay off. Perhaps the two worst losses on the Smith-coached end of the schedule, the Kansas St. and TCU games, both saw encouraging havoc rates from the defense, but due to big plays and high offensive efficiency from the opponent, it's quite possible that Smith's more aggressive approach may have made these final scores look worse than they might otherwise have been. Then again, it's also possible that pursuing the aggressive strategy was the only way to have even a low-percentage chance at a victory, considering what excellent form both of these teams have been in all year. That's certainly supported by the result against Baylor, another case of a strong havoc rate in the face of poor yardage and points totals, where big plays by the defense helped almost steal a game from a superior opponent.

Another takeaway is that when the defense faced an opponent with a dominant rushing attack, the havoc rate goes down no matter who is in charge. Arkansas and Oklahoma both enjoyed mostly havoc-free offensive outings against Texas Tech's defense. This is mainly due to Tech's havoc plays being predominantly composed of sacks and passes defensed: they force relatively few fumbles and rarely tackle runners behind the line of scrimmage. As we've seen for years, the lack of depth or talent on the defensive line makes this something of an inevitable outcome: until further notice, Texas Tech is not set up to stop a power running game.

On a more encouraging note, Mike Smith's defense showed that when they had the opportunity, they could victimize a bad offense. The Kansas, Texas, and Iowa St. games all were strong defensive performances both in havoc rate and other measurements (at least by this team's standards). It's good to know that, even when your defense is bottoming out, you can make that happen. Perhaps most encouragingly, Mike Smith's extended audition had an average havoc rate of 15.14%- and he did it without Kerry Hyder, Dartwan Bush, Will Smith, Terrance Bullitt, Bruce Jones, or Tre Porter. It's pretty stunning: 8.7% in their first three games, 15.14% in their last nine, and against far better competition. It was a rough year any way you slice it, but this was an area where we saw unambiguous improvement with the change in management.

Ultimately, Mike Smith, his defensive assistant coaches, and even Matt Wallerstedt don't deserve to be judged solely by a statistic as simple as havoc rate. Their jobs are too complicated, the factors that determine their success too numerous to allow for such an easy answer. Bill Connelly calls havoc rate a "personality stat", because it can and does capture the type of philosophy a defense pursues, but in our case I think it captures something else: a personality match. Texas Tech defensive coordinators have known for as long as fans have that this team is up against it in terms of defensive talent and depth, but unlike his predecessor, Mike Smith responded by going more exotic and blitz-heavy with his schemes, encouraging more aggressive play from his defenders, and staking his employment on the idea that it is possible for his coaching to make the difference. Havoc rate shows us that not only is this good football strategy, but that his players responded to that approach. I don't know if Mike Smith is going to lose his interim tag or if Texas Tech moves on to its 7th defensive coordinator in 7 years, but if he's not, I hope like hell we can bring someone in who can cause as much havoc as he did on short notice.