The UW Defense & Luck Revisited

This summer I wrote a FanPost looking at whether the UW Defense was susceptible to regression to the mean. I’ll quote the article throughout but if you want to refresh your memory you can find it here. With a little bit of time before the Alabama game, let’s take a look at how accurate my conclusions were.

Turnovers

What I said:

“UW certainly has the personnel to lead the countries in turnovers if everything breaks right but the numbers suggest a bit of caution. Dropping from 29 to 27 to something in the range of 23-24 seems reasonable for this year. That number would drop UW to around the 35-40 range nationally but would still be above average.”

What actually happened:

It turns out I should have just shut up after 11 words as UW did indeed lead the country in turnovers with 33 total. The Huskies are now the only team in the country to have at least 27 turnovers in each of the past three seasons. Many teams still have a bowl game left to play but no one can match that number. San Diego State has forced at lest 25 turnovers in all three years while Louisville and Clemson each have forced at least 24 turnovers. Regardless, the Dawgs are in rarefied air here.

In August, I said that UW had shown an ability to force a large number of fumbles but that the ability to recover them at more than 50% is still a matter of luck. UW forced 1.8 fumbles per game in 2014, 1.8 in 2015, and 1.9 this year. How’s that for consistency? They ended up recovering 56% of all the fumbles so slightly lucky but not incredibly so. The coaching staff has emphasized stripping the ball through sure tackling and that mindset has certainly taken hold.

Red Zone Performance

What I said:

“The numbers aren't encouraging for UW to remain among the truly elite as the stingiest defenses in the country...There were 8 teams that increased by at least 30 yards per TD between 2013 and 2014 and only one (San Diego St.) saw another increase the next year. The average team in that group lost 21 yards per TD.”

What actually happened:

I nailed this one pretty much exactly. For those that didn’t read the original, I laid out a metric called stinginess which is equal to a defense’s total yards divided by offensive touchdowns given up. It tells you how many yards an opposing offense needs to gain on average in order to score an offensive touchdown. Large numbers are good and small numbers are bad.

Last year the Huskies were 6th in the country in stinginess. They gave up a touchdown every 176 yards. This season they fell down to 19th with a stinginess of 152 yards per touchdown. I predicted about a 21 yard/TD drop and they ended up dropping by 24. Not bad.

There were a lot of people in the original article who disputed the idea that stinginess is subject to luck. They said that elite coaching and specific defensive styles could transcend natural rises and falls in defense. Since 2013, UW has had a stinginess of at least 143 yds/TD in all 4 seasons. The only other programs to do that: Alabama, Stanford, and Florida. Those are some very good defensive programs with very good coaches. I think it is definitely fair to conclude that certain programs can be consistently excellent in this regard and it appears the Huskies are one of them.

[Sidebar] In the previous three seasons the best stinginess score was Ole Miss in 2014 with 204 yds/TD. This season Alabama has 230 yds/TD!!! The most shocking thing is that Auburn has even higher at 245 yds/TD while Ohio State and LSU also have better numbers than 2014 Ole Miss. I certainly hope that playing Washington will bring down that number for Alabama but the talk of Alabama’s defense being one of the best ever is real.

Casey Sapio-USA TODAY Sports

Injuries:

What I said:

“It's not a projection I like making but by the end of the year it's probable that Washington is down multiple key contributors and is worse off than we were a year ago. The good news is that this defense is incredibly deep. This lessens the impact of any individual injury as the difference in quality between our starters and backups on defense has to be close to if not the smallest in the conference.”

What actually happened:

Basically that. The defense definitely sustained more injuries this year than they had in 2015. Both Joe Mathis and Azeem Victor were lost for the season in the middle of the year. The sack numbers before and after Mathis went down are stark enough that I don’t think it’s fair to say that the team held steady in his absence. However, the defense as a whole compensated for the reduced pass rush.

D.J Beavers has played very well in Victor’s place so far. Comparing Beavers to Azeem isn’t very fair but he has at the very least been solid to this point. There have been a few other nagging injuries costing starters some time but those have been about it.

I also predicted that we would see at least two incorrect targeting calls which would cause defenders to be missing in action for a half or two (with Victor the leading contender). I am pleasantly surprised to say that never happened this year. Good job Pac-12 Refs! (That is the first and only time that phrase will ever be typed in that order.)

Strength of Schedule

What I said:

“In addition to the schedule being easier overall it gets even more favorable when you look at what our opposing offenses lost in the off-season...If UW performs exactly the same from a talent and execution standpoint next year the improvement in their schedule could reasonably be expected to shave off maybe 30 YPG and a few points per game.”

What actually happened:

The schedule ended up breaking even better for the Dawgs than anticipated in August. Technically, the schedule almost cost them a shot at the playoffs but in general the SOS was more of a plus than a minus. I used Sagarin as the source for my SOS numbers back in August and predicted it would end up somewhere around 55. Sagarin has UW’s schedule ranked 41st right now but that includes a Pac-12 championship game which we didn’t know would happen in August.

I wasn’t exactly going out on a limb saying that the offense in generally would dip in the Pac-12 this year but that held true as well. In 2015 the top six offenses in the conference by yards per game were Oregon, Cal, Arizona, Arizona St, Wazzu, and UCLA. Those offenses changed respectively by -46.5 yards, -16, -81.8, -86.6, (+)28, and -86.1. The only one to improve was the one who got a full season out of their future NFL QB. That did not hurt the outlook for the defense.

As far as the stats go, Washington’s yards per game dropped by 36 and their points per game dropped by 1.6. My caveat was “if this team plays at the same level as it did in 2015”. I think it’s fair to say that this version of the Husky defense was better in 2016.

Conclusion

When people criticize the use of statistics to make predictions in sports it is usually because they don’t understand that stats provide a range of probabilities. The numbers said that it was unlikely for UW to get more turnovers than they did either of the last two years. But they didn’t say it was impossible and when combined with eyes and the human brain it was easy to imagine a best-case scenario where it happened like it did. Other times, like for the stinginess number, the action on the field plays out just as the numbers predicted. But if this year has taught us anything it’s that as fans we have every right to close the spreadsheet and say no to the cold heartless world of analytics. Boundless optimism is just as good of a prediction method sometimes as the 2016 Dawgs can attest to. Follow your heart and if you want to predict 49-0 on New Year’s Eve I’m not going to stop you.