Friday, August 07, 2015

Calculating Success Rate

Overnight temperatures in the low 50s at this time of year result in fog on the Connecticut River. Somewhere in the thick soup down there this morning is the Dartmouth campus. Up here on the shoulder of Moose Mountain the skies were crystal clear while in the valley cars needed their headlights. Rising in the background are Killington and Pico ski resorts. Click the photo to enlarge it.
A few years after I graduated my high school had a small, quick and blazingly fast running back. It seemed as if he would spin off a run of 70 or 80 yards every single game. His other 12-15 carries would net maybe 20 yards at most.

If you looked at his stats, his average per carry was healthy and so was his average per game, but we used to debate whether he should keep getting the ball in the hope that he'd break another one, or whether it would be better to hand it to someone else who could grind out consistent yardage. (Being a Penn State guy I was in the latter category.)

You had to know in an era of analytics and Moneyball someone would come up with an algorithm to assess a back's relative value beyond yards gained per game and per carry, and now somebody has.

Football Outsiders brings us the "Success Rate," which has been adopted by some NFL teams. Here's the idea:

• A run on first down is considered a success if it achieves 40 percent of a first down. So, on first-and-10 any run of four or more yards is a success.

• On second down a successful run picks up 60 percent of the remaining yardage. In other words, on second-and-10 a run that picks up six yards is deemed successful. On second-and-eight a runner would have to pick up five yards for the run to be a success.

• On third or fourth down the runner must achieve a first down for the run to be a success.

Is the formula flawed? Of course. For example, on a third-and-20 you can run for 19 yards and the run is not considered a success. But I digress . . .

In the NFL, a runner who grades out at 50 percent for a season is considered to have had a pretty good year.

After reading about the Success Rate, I started to wonder how Dartmouth's backs would have been graded last year. That got me wondering about the top backs returning in the Ivy League – as well as last year's Ivy rushing leader. And eventually I started to wonder how Dartmouth's best backs in recent years compared.

In college I wrote a Fortran computer program to determine something I developed for baseball called an REA, a Run Efficiency Average. It was a pretty novel formula. Too bad I never followed up on it because I was well ahead of the curve ;-)

Unfortunately, whatever programming literacy I once had long ago disappeared. I could have developed a spreadsheet to calculate Success Rates but to be completely honest I think I would have spent more time making sure it worked properly and inputting the numbers incorrectly than using pen and paper. Which is what I used.

Every single-season carry for nine players was charted – and I have the blisters and blurry vision to prove it.

In the end I calculated the overall Success Rate as well as the Success Rate on first down, second down and third/fourth down for Dartmouth rushing leader Kyle Bramble as well as the three backs who finished above him in yards-per-game last year: Yale's Tyler Varga, Harvard's Paul Stanton, Jr., and Cornell's Luke Hagy.

Because he also finished in the top-10 in Ivy Leaguer yards per game I calculated the numbers for Dartmouth quarterback Dalyn Williams.

(I also worked up the numbers for Dartmouth's Brian Grove and Ryder Stone, but because their number of carries was significantly fewer than the others I did not include them in the main charts below.)

Finally, for context I went back and worked out the senior year Success Rates for Nick Schwieger '12 and Dominick Pierre '14, Dartmouth's two all-time rushing leaders.

Before going any further, here are the top four rushers in the Ivy League last fall in yards per game (yards per carry in parentheses):

1. Tyler Varga, Yale – 142.3 (6.1)
2. Paul Stanton, Harvard – 110 (6.7)
3. Luke Hagy, Cornell – 81.6 (4.9)
4. Kyle Bramble, Dartmouth – 72.8 (5.2)

Here are the same numbers for Schwieger and Pierre in their senior seasons:
Schwieger – 131.0 (5.4)
Pierre – 92.2 (5.0)

Keeping in mind that no one proofed my work and that while I went to college as a math major I ended up graduating with an English degree . . . here are the numbers I came up with. All mistakes are mine and mine alone ;-)

First Down Success Rate
61.4 – Tyler Varga, Yale '15
61.1 – Nick Schwieger, Dartmouth '12
59.4 – Dalyn Williams, Dartmouth '16
58.9 – Dominick Pierre, Dartmouth '14
57.1 – Kyle Bramble, Dartmouth '16
55.1 – Paul Stanton, Jr., Harvard '16
46.2 – Luke Hagy, Cornell '16

Second Down Success Rate
66.7 – Varga
58.8 – Williams
56.7 – Pierre
54.2 – Stanton, Jr.
46.7 – Bramble
42.7 – Schwieger
23.9 – Hagy

Third/Four Down Success Rate
82.1 – Varga
65.9 – Pierre
58.7 – Schwieger
54.2 – Hagy
51.7 – Stanton, Jr.
50.0 – Williams
46.7 – Bramble

Overall Success Rate
65.4 – Varga
58.9 – Pierre
56.7 – Williams
54.6 – Schwieger
53.4 – Stanton, Jr.
51.9 – Bramble
39.6 – Hagy

Brian Grove and Ryder Stone were Dartmouth's leading running backs after Bramble last year. Grove had 221 yards on 38 carries (31.6 yards per game, 5.8 yards per carry). Stone had 175 yards on 25 carries (21.9 yards per game, 7.0 yards per carry). Because their carries were limited the numbers are skewed but here goes:

First Down Success Rate
75.0 – Stone
50.0 – Grove

Second Down Success Rate
60.0 – Stone
41.7 – Grove

Third/Four Down Success Rate
66.7 – Grove
50.0 – Stone

Overall Success Rate
66.7 – Stone
51.3 – Grove

What does it all mean? You tell me by clicking on the comments link below ;-)

No comments: