~ ~

Sunday, July 10, 2011

The Big Play and Impact on Wins/Loses

I've been using this website to check articles (links to all Spartan media stories in the footer at the bottom of the page). This one reminded me of a Spartan friend named Mark - he's a stats/math guy. I found this post at
"The Only Colors" Blog (again, links at the bottom)

Football Stats I Like- Net Big Plays

Like just about everyone else, I love big plays and I hate big plays. Specifically, I love big plays when my team is making them, and I hate them when my team is giving them up. As it turns out, it seems like victories also have a similar love-hate relationship with the big play. In a close game, little helps more than one of these big turn of events to get some momentum back. And when your team is trying to dig out of a hole, a big play given up can break your team's back. As such, I wanted to take a look and see just how important the 'Big Play' was to a team's success (measured in wins).
This was done by adding up all of a team's offensive running plays of 20+ yards (about 4-5 times the average carry) and all of their offensive passing plays of 30+ yards (about 3-4 times the average pass), and the same plays given up by the same team's defensive unit (the reason that the running plays are 4-5 times, and passing plays are only 3-4 times, is an attempt to adjust for a discrepancy in attempts. Big Ten teams in 2010 ran the ball essentially 4 times for every 3 passes).
Next added in were all of a team's made and surrendered punt returns of 30+ yards and kick off returns of 50+ yards. Then added in were all of a team's fumbles and interceptions, gained and lost. Then added were any defensive or return touchdowns scored or given up. Finally, any blocked kicks or punts, and any long field goals of 50+ yards made or conceded.
So for each column: [(Runs of 20+) + (Passes of 30+) + (PRs of 30+) + (KRs of 50+) + (Fumbles) + (INTs) + (return TDs) + (Def. TDs) + (Blkd Kicks) + (50+ FGs)
Then, Big Plays given up were simply subtracted from Big Plays made in order to provide a team's net score. Like so:

2010 Net Big Plays
TeamMadeGiven UpNetWins
Ohio State82433912*
Michigan State65442111
Penn St.535127

*Well, you know, technically...

Further analysis, and the full tables, after the jump.
Full Made Table:

2010 Big Plays Made
TeamRuns and Passes50+ yd FgsBlocksReturn TDsDefensive TDs30+ yd PR50+ yd KRInterceptionsFumblesTotalWins
Ohio State3903441119118212
Michigan State351210201776511
Penn St.31012011107537
Full Given Up Table:

2010 Big Plays Given Up
TeamRuns and Passes50+ yd FGsBlocksReturn TDsDefensive TDs30+ yd PR50+ yd KRInterceptionsFumblesTotalWins
Ohio State141441131324312
Michigan State201102011094411
Penn St.21111510174517
Further thoughts:
- It's clearly not enough to just have half the big play equation. 2010 Michigan was ridiculously explosive on offense, but even more combustible on defense and special teams. 2010 Purdue was extremely stingy at giving up the big play, but its offense and special teams were only slightly better at making them on the other side of the ball.
-Ohio State is about where you'd you expect to see the conference's elite team (of cheaters), but man, hats off to Illinois on tying for third in net during a season where they went 7-6. Thankfully, a. we probably don't have to play them this season, and b. they lost Liuget, Wilson, and Leshoure among others.
-File this one as another entry into the 'Iowa was probably the second best team in the Big Ten last year, statistically speaking.' folder. Losses to OSU and Wisconsin are totally understandable. The devastating time-zone death trap of traveling to Arizona can excuse that loss. But how that team lost to jNorthwestern and Minnesota (Minnesota!), I'll never know. The mysteries of college football...
- Like all stats, this one isn't perfect, but it still tells a pretty good story. Over a quarter (12 out of 47, or 25.53%) of MSU's 2010 offensive touchdowns could be categorized as 'Big Play Touchdowns'. In other words, Michigan State's offense was explosive enough to score a fourth of the time, without even having to enter the red zone. Furthermore, out of MSU's total offensive big plays, over a third (12 out of 35, or 34.29%) went for touchdowns, making the big play a good payoff from a probability perspective.

 -On the other side of the ball, only a fifth (7 out of 35, or 20%) of MSU's defensive touchdowns were of the big play variety. This meant teams facing the Spartans mostly had to work to get their points. However, when the defense did break, it often broke big, giving up a touchdown on a big play over a third of the time as well (7 out of 20, or 35%).

- MSU was also a healthy +5 in turnover margin. The only thing holding them back from teams like Wisconsin and Illinois, curiously, for a team with players like Martin, Bates, Conroy, Worthy, Jones, Adams, and Rucker, were various big defensive and special teams plays.

-Finally, all of the top six teams in net big plays finished above .500 overall, and all six finished .500 or better in the B1G. Meanwhile, only two of the bottom five in net finished above .500 overall, and none of those five finished above .500 in the B1G.
Statistical correlation between wins and net big plays is a very healthy .794 for the 2010 season (for the record, that's a better correlation of 'stat-to-wins' than 2007-2010 correlations on total offense, total defense, scoring offense, or scoring defense. Meaning this stat would've been more likely to correlate to a winning team than any of the 'Big Four' of college football stats. I'd be curious to see if that held up if we stretched the Big Play data out to four years as well)

So, I know: small sample size, somewhat arbitrary cut-off points for what is, and isn't, a 'Big Play', and all that. But there's a good argument to be made that effectiveness at making, and stopping, the 'Big Play' (at least as it's defined here) usually leads to wins in the Big Ten. Here's to thirty yards and a cloud of fireworks.

[Author's Note: This stat is one that I enjoy quite a bit, but, as it is listed here, it's a little more flawed than I'd ultimately like. Unfortunately, to get a more accurate sampling and net would require going through 11 team's 12-13 game play-by-play logs for last season. And I- I just can't do that right now.Still, there's certainly something to chew on here.
For the record, the missing adjustments would be adding in safeties,  and (the bigger adjustment) shifting long passes from 30+ yards to 35+ yards. I also would've appreciated more years of data, for a better sample size. As it is, I worked with what cfbstats , NCAA's stat websiteand ESPN gave me. This is something I'm likely going to be tracking, hopefully with these adjustments added in, when the 2011 season rolls about.)


(A call for technical help: Is the simplest way to insert a scatter plot graph into one of these posts to just host the graph to a image hosting website and use the image URL importer tool? Or can I somehow insert it in through the HTML view? For some reason, the interface won't let me copy and paste the graph directly from my excel program. Thanks!)

Thanks for reading. Additional thanks goes to Cold Hard Football Facts and their Big Play Index for the inspiration.

-Heck Dorland

No comments:

Post a Comment

Please sign in using the method most convenient for you. We do not receive your login information. This function is provided by Blogger.