What If We're Just Lucky?

Nov 14, 2023

It's a great card to play in any college football debate. There are two Boston College fans arguing about their season, and the guy who doesn't like Jeff Hafley tells the guy who does like Jeff Hafley "sure, we're 6-4, but we're maybe five plays from 1-9 - we beat Virginia, Army, and Holy Cross by three and only beat UConn and Syracuse by seven."

And you'll see it used the other way, too. We saw it in 2006. "Sure, we went 2-10, but this team was maybe four plays from 6-6." Flip all the close games the other way and you can claim that the coach you like (or dislike) is simply unlucky (or lucky).

The great equalizer for that, at least in my mind, is Post-Game Win Expectancy (PGWE). It's a way to quantify whether you were actually lucky or unlucky. It traces all the things that historically have led to college football wins, sprinkles in some expected turnovers (you should recover 50% of all fumbles; you should intercept 21% of all passes defended), and comes up with a percentage. Play this game with those stats 100 times and you should expect to win XX out of 100.

If we look at Boston College this season, uh, yes, they've been lucky. They have six wins but when you add up their PGWE they're at 4.1 (meaning, a team that should have won 4.1 games has won 6). They beat Holy Cross with a 34% PGWE and they beat Army with a 17% PGWE. They are very, very fortunate to be 6-4.

(I often think of it in terms of an outdoor party and rain in the forecast. Against Holy Cross, there was a 66% chance of rain for BC's party and it didn't rain. Against Army, the forecasters put it at an 83% chance of rain yet they still somehow held an outdoor party with no rain. Very, very fortunate. Don't expect it to happen again. Odds will catch up to you very soon.)

I've used this stat for years. You'll probably remember that during the 2019 season, I drew a very clear line between the Wisconsin win and the Michigan State win. Wisconsin? Wasn't luck. A 54/46 game that we won. Michigan State? Luck. An 86/14 game where 14 hit.

So let's talk about the last four weeks. We won on a walkoff field goal, then we lost on a touchdown with 27 seconds left, then we won on a long pass play with 50 seconds left, then we won in overtime. Are these all BC/Army-like games where we won by three but the stats suggest we should have lost by two scores?

In a word, no. Here's the PGWE for each game:

Maryland: 61.6% - should have won by 2.4, won by 3
Wisconsin: 34.7% - should have lost by 3.3, lost by 4
Minnesota: 68.2% - should have won by 3.9, won by 1
Indiana: 91.5% - should have won by 11.3, won by 3

Add it all up and we're a team that should have 4.9 wins and we have 5 wins. We're not lucky, we're not unlucky, we're just... a 5-5 football team that's 5-5.

Again, the point of this statistic isn't to replace the actual results. This exists in the realm of predictive stats. The kinds of stats that Vegas uses to set lines. When you see a 7-3 team that's only a 4-point favorite over a nobody opponent, go check their PGWE. I'll bet that it's saying "right, but this team is very fortunate to be 7-3."

Why is the Indiana number so high in a game that went to overtime? Especially a game where turnovers were even? Well, Illinois was by far the most explosive team (9.3 yards per play to 5.9). And the team that's more explosive usually comes out on top. We did come out on top, just... barely.

All of that makes our adjusted scoring margin one of our most interesting stats this year. We should have beaten Toledo by 9 and we won by 2. We should have lost to Penn State by 7 and we lost by 17. We should have beaten FAU by 18 and we won by 6. Put it all together and, on Bill Connelly's chart, we're second only to Arkansas State in scoring margin difference. Our scoring margin on the season: -5.5. What our scoring margin should be for the season: -0.5.

And that's not "should" as in "a team as good as Illinois should beat a team like FAU by 18". That's "in this game, with these statistics, Team A usually beats Team B by 18." That doesn't mean it will always happen that way. Football is a funny game. But that's the most likely outcome in a game with those statistics.

So no, we're not lucky. Yes, winning all these one score games is a bit fortunate (those come down to... One Or Two Plays) but in terms of "luck" as quantified by PGWE, we haven't been lucky. We're not a great team, we're not an awful team, we're just... a 5-5 football team.

A 5-5 football team with... a great offense? Let's get to part two of this post. If you thought our Success Rate numbers looked good before...

Here's a chart I tweeted yesterday. I made the chart on CollegeFootballData.com. X-axis is Success Rate, Y-axis is offensive explosiveness.

We've covered both stats recently but a quick synopsis: Success Rate measures whether you got 50% of the required yards on first down, 70% of the required yards on second down, and 100% of the required yards on third or fourth down. It's pass-fail for every play. Why grade plays like that? Here's Parker Fleming's description of Success Rate. It's the best way I've seen it described:

All success rate does is strip out some noise and present a view of how two teams compared in the down to down business of moving the football. Games are won and lost with explosives, special teams, turnovers, etc, and looking at success rate just shows us how close the margins were. It's a good diagnostic to go back and look at a game and learn something.

Explosiveness, on the other hand, is an EPA-based stat (Expected Points Added). It's simply the average EPA on plays marked as successful. The bigger the plays (deep passes, long runs), the higher the average. Pretty simple.

Put that together like this and it says that Illinois has the third-best offense in the Big Ten. Maryland is much more explosive (but also significantly worse at Success Rate). Penn State is nearly the same as Illinois when it comes to Success Rate (although we're slightly ahead) but Penn State is also below Iowa when it comes to explosiveness.

As an aside, remember that reporter who asked James Franklin "why not just chuck it deep" to see if a receiver could make a play on it? That's why he asked that. Penn State is 133rd out of 133 teams in offensive explosiveness. The most boring offense in America.

Back to Illinois. In terms of points per game, Illinois is first in the Big Ten West. It terms of total offense, Illinois is first in the Big Ten West. When you look at Success Rate, the stat I've been pointing to all season, Illinois is not only 1st in the Big Ten West, we're 3rd in the Big Ten (behind only Michigan and Ohio State) and 23rd nationally. And no, that's not just because of the Indiana game. We were 30th last week (4th in the Big Ten) and moved up to 23rd this week (3rd).

So let's talk about that. And please allow me to dislodge my shoulder from its socket by patting myself on the back.

I wrote this after the back-to-back Kansas and Penn State losses:

Where I said this:

Or perhaps it means that we're a little better than we've shown so far. And that's why I'm writing this article. This is just one metric among dozens of metrics. It's not to be pointed to as "see, here's the truth about this team." But it might - might - suggest that we'll be able to hang in every game from here on out. (I mean, not if we go -5 in turnovers every game, but you get the point.)

Winning through yards is sustainable. Winning through penalties and turnovers is not. If you're consistently "winning", say, 60% of the plays in a game (as defined by Success Rate), then you probably have a sustainable team. You can move the ball. You can stop the ball. If you're only "winning" 40% of the plays but you're still winning games because the other team is turning it over and killing drives with penalties, then no, that's probably not sustainable.

And then on October 5th, I wrote this:

I had teased that headline the day before by ending my Returning Production article this way:

Yesterday and today, while reviewing all of the updated college football stats (and watching Devon Witherspoon dominate the Seahawks/Giants game, realizing that special players can make, like, five game-changing plays in a single game), my brain settled on two things:

One, we get our table assignment at the 2023 college football wedding and we see that we're seated with Arkansas and Pitt and Oklahoma State and Baylor. And we're wondering why they sat us all together and then we realize we were placed at the Returning Production Woes table. None of us can stop making mistakes. The groom's cousin Notre Dame walked by and said something smarmy like "oh look at that - the Mistakes Table" and Alabama had to hold us back because we wanted to punch him.

Two, this is still the best Illini offense of the last 10 years.

What?? ROBERT. You're not going to end at that, are you? Have you seen this offensive line? How can you POSSIBLY say that?

See you tomorrow.

And then I wrote "What If I Told You This Offense Is Good?", laying it all out there using Success Rate and other metrics. And then we lost to Nebraska 20-7 the very next day and I was dragged up and down the streets of Twitter. And then I quadrupled down and wrote "PFICK" and noted how the offense was still good and it was every single little thing that was costing us games. And people said "poor guy just can't admit that Barry Lunney sucks."

And then we got better at the little things (and a little better on defense) and won three of four and the offense climbed to the top of the West in points and yards (and now all the way up to 23rd in the country at Success Rate). And, believe it or not, not a single person that dragged me has said a word. Whatever happened to the days of "hey man, I gotta hand it to you..."?

This does not mean "Illinois is great!" Illinois is average because the Illinois defense is below average for the Big Ten. If you sort Big Ten defenses by Success Rate, we are... 13th out of 14 teams, only ahead of Indiana. It's an average-to-decent defense nationally, but that's like having an average-to-decent offense in the Big 12. You're seriously hamstrung given the way football is played in your conference.

Where does that leave us as a team after ten games? I think this EPA chart from @JBudDavis has us nailed:

(EPA/play is Expected Points Added Per Play)

I think this tells our season story very concisely. If this was last year, the Illinois logo would have been right were the word "good" is in "it's a very good defense". Fantastic defense, subpar (by national standards) offense.This year, we're in the "good offense" quadrant but have dropped below the "good defense" line. It's not some massively concerning drop like Baylor, Minnesota, Wake Forest, Mississippi State, Michigan State, or Pitt. It's just... a team that got a lot better on offense but lost too much on defense.

Lucky? No. A 5-5 team that should be 5-5. We still struggle with field position, we still struggle with turnover margin, we still struggle with penalties, and so our offensive gains haven't been muted. We went from 24 interceptions in 13 games to 5 interceptions in 10 game (YIKES) and that sent our defensive stop rate (and field position advantage) into the tank. Toss in all these penalties and... we're 5-5.

So now let's get that sixth win. I don't care which game. Figure it out, find a way, and let's go to Detroit on December 26th.

Maybe Connor Stalions will be running the video crew.


DB50 on November 15, 2023 @ 12:55 PM

Two different phrases I think of when someone mentions luck; "Luck is the residue of design" or "Luck is when opportunity meets preparation "!

Refugee from Illinois on November 15, 2023 @ 05:42 PM

Enthusiastically hoping we “go to Detroit December 26th” is a quintessential Robert thing. Love it.

Speak Your Mind

Please login or register to post comments on the IlliniBoard.

Post Preview