Check The Tape - Rutgers (Second Order Wins Edition)
You have currently viewed 1 story this month.
IlliniBoard now offers two free stories per month, for more please subscribe.
Fridays are all about combining articles apparently. I'm en route to Minneapolis (choo choo), and I want to write about two things, and I have to write about 8,000 words for the basketball preview (comes out Monday), so this will be a combo post. Go with the cheddar-pretzel.
The two things I want to write about:
- Some of the gifs I captured when re-watching the Rutgers game this week.
- Bill Connelly finally adding everyone's Postgame Win Expectancy to his public SP+ spreadsheet in Google Sheets.
The second one is a big deal for me. As you may know, I believe his Postgame Win Expectancy (plus the resulting Second Order Wins) to be the best barometer in college football, and when Connelly moved to ESPN in 2019, his sortable stats all went away. You had to wait for him to publish them on ESPN (or wait for him to tweet individual games). As of this morning, he's added them to his public spreadsheet. Merry Christmas to me.
I've explained the stat many times, but if you're new, a quick recap. It takes all of the stats from a game - how many explosive plays you had, the success rate of your offense and defense, turnovers in the game, and several other factors - and says "if the game goes like this, Team X will win 73% of the time". Meaning, if that game was played 100 times, Team X wins 73 of them and Team Y only wins 27 of them. When Team Y wins, they were a bit fortunate to win a game like that. And when a game is 98%/2% and the 2% wins, it's flat-out luck.
There's a lot of misconceptions about the statistic, especially when it comes to turnovers. People (and by "people" I mean "Iowa fans") hate the concept of turnover luck, and so they exaggerate and say things like "he considers all turnovers to be luck so this statistic is idiotic". He does not consider all turnovers to be luck. For the turnover portion of this statistic, he looks at turnovers above normal rates.
Fumbles recovered usually come out to right around 50/50 (offense grabs 49%, defense grabs 51%). If there are five fumbles in a game - three by you, two by the opponent, and your opponent recovers all five - that's fairly lucky for them (more on this in the images below). About 19-21% of all passes defended (deflected at the line, batted away, etc) result in interceptions. If you tipped three passes and they all fell directly to your safety, that's crazy luck. So, any departure from the norm says that you were a bit fortunate to win the game.
Add that with all of the other factors and you get to his Postgame Win Expectancy. Wisconsin game in 2019? 54/46. Even game won by Illinois. Michigan State game in 2019? 86/14 Michigan State, Two tipped passes fell to Sydney Brown, one tipped pass fell to Stanley Green, Michigan State outgained us, out-explosived us, and did everything you usually do to win a game... and we won.
(I feel like I've used this same exact explanation about 20 times.)
ANYWAY, here's the PGWE for 2021. Based on the stats in the game, the percentage chance that Illinois would win that game:
Nebraska - 59.2%
UTSA - 12.7%
Virginia - 0.0% (sounds about right)
Maryland - 3.3%
Purdue - 17.4%
Charlotte - 37.4% (whoa)
Wisconsin - 0.0% (totally)
Penn State - 86.6%
Rutgers - 54.1%
Rutgers is the most interesting here. Outgained by 90 yards, no turnovers, yet had the (slightly) higher PGWE. I'm guessing a lot of that is the explosive plays. Just look at the yards per play - Illinois 6.1, Rutgers 5.0. Run that out over a full game and the team with the higher YPP is usually going to win in a turnover-less game. Something something three penalties killing three drives.
Purdue and Maryland are the other surprises. Penn State we've already discussed - Penn State would have needed 2019 Michigan State game fortune to win that game (and they almost did). But Maryland being 3.3% is quite a surprise. We have the ball, 3:00 remaining, with a chance to drive 40 yards, kick a FG, and win, and that was a 3.3% win expectancy game. Of course, if I recall correctly, we recovered all four fumbles in that game, and we were outgained significantly, so, that does make at lot of sense.
All of that combines together into Connelly's Second Order Wins statistic. Combine all of those percentages and how many wins should you have?
And "should" is where most fans get hung up. Tell some Northwestern fan that their Second Order Win total in 2018 was 6.2 (meaning it was a 6-6 team that needed every ounce of fortune and luck to win nine games) and they'll throw things at you. They just know "how to win" and "Fitz magic" and blah blah blah. I actually agree to a point. A 6-6 team that goes 9-3 is a lot of fun. I'd love to see it happen in Champaign. Some teams do win every close game (like 2001 Illinois). Some teams lose every close game (like 2006 Illinois). And quite often a team with a very high Second Order Wins total (like 2006 Illinois, which was 5.5, meaning that was a 5-7 or 6-6 team that went 2-10) can speak of good things to come.
So what's the current Second Order Wins for this Illini team? 2.7. This is a 3-6 football team that should have won 2.7 games. So more or less the results are as expected.
I thought it might be higher than that given the down-to-the-final-drive results in the Maryland, Purdue, and Rutgers games (all losses), but both Purdue and Maryland fall in the "lucky to be close at the end" category. Stats leaned heavily in Maryland's favor (but the three fumble recoveries helped us a ton), and Purdue is the kind of game that was only close because of the two interceptions (one in the endzone). Also, thinking across the whole season, it's not hard to think of dropped interceptions by our opponents. Those drive that "passes defended but not intercepted" stat in the wrong direction.
Overall, there's not much to learn here. Our lack of offensive explosiveness is going to make wins few and far between. And we already knew that because we watch the games.
OK, so let's launch into the (abbreviated) CTT for Rutgers. And let's tie these two together by talking about turnover luck.
There were no turnovers in the Rutgers game. But there were almost two. Let's rate them based on 0 to 5 Turnover Fairies. 0 means no luck involved at all. 5 means all was lost without the Turnover Fairy stepping in to help.
Fumble One. 4.25 Turnover Fairies
The punchout here? 100% not luck. Great defensive play. This is where everyone always gets hung up when it comes to turnover "luck". They'll scream and scream and scream that punching a ball out like that is not luck. And they are 100% right.
But where the ball goes after that? Completely up to the turnover fairy. Oblong ball bouncing on the turf and where she stops, no one knows. In the case of that Chase fumble, I'd say there was about a 17% chance it bounced out of bounds. Get out your protractors.
The ball is punched towards the Illinois endzone so there's about 180 degrees it can go. Needs to be between 0 and 32 degrees to get out of bounds. Anywhere from 33 degrees to 180 degrees it's gonna be Rutgers ball. It got out of bounds. Massive turnover luck (especially since the bomb to Isaiah Williams was a few plays later).
OK, next play,
Fumble Two - 3.5 Turnover Fairies
If we're just going with the bounce there, it's damn near 5.0 turnover fairies. Haven't seen a bounce that fortunate since Joe McKnight in the Rose Bowl. Could have been absolute disaster for Rutgers yet it bounces to him in stride.
The thing that keeps this at only 3.5 Turnover Fairies is the location of the fumble. Once that ball is out, there's still 5 Rutgers players closer to the ball than any Illini player. It's like the shotgun snap that the quarterback drops. It's a "fumble", but 90% of the time the QB or the RB falls on it. (The snap over the QB 's head is a different story).
Here, it's a 1-in-25 lucky bounce but also in one of the least damaging places to fumble a ball. When the ball comes out, the closest Illini defender is still at the 25. So he had time to recover. If this is a punt returner muffing a kick like that, there's probably four opposing players within five yards and it changes the math.
Anyway, that's how I see Turnover Luck. Both teams were lucky not to lose the ball. Chase Brown was more fortunate.
The worst thing about losing to Rutgers at home is that you cannot "LOL Rutgers" because they just beat you. But I'm going to do it anyway. This play, to me, was even more confusing than the fourth-down call for Illinois at the end of the game.
Third and six. This is just before Rutgers kicked the FG to go up 10-7. They choose to run the ball (??) and then they call this:
OK, yes, they weren't expecting Isaiah Gay out there. That's one of the benefits of this OLB position. A typical defensive end is rushing the QB there. So they're expecting Gay to head towards the QB and then they're going to try to get the pulling guard to chip Tarique Barnes coming from MLB.
Instead, Gay drifts out and is basically standing and waiting for the tailback. And the pulling guard, um, gets blocked by his own guy (no, seriously - go watch the Rutgers offensive linemen block each other). This leaves absolutely nowhere for the tailback to go. There's honestly eight guys who could have made the tackle and prevented the first down. LOL Rutgers.
(LOL Rutgers, the team that just got their first Big Ten win in Memorial Stadium.)
Out, Damned Spot
I tweeted all of these images the other day, but I wanted to go through it all one more time. It just doesn't make any sense.
This is the third down play. The one before the fourth down play where we lost four yards and Rutgers won. We'll start with the gif. Here's the forward progress for Jakari Norwood. The line to gain for the first down is the 33, so the second hash mark.
The arrow here is pointing at the hash mark he needed to reach. Just short. Somewhere between the 34 and the 33 and closer to the 33 than the 34.
And the Line Judge comes running in and puts his foot here. I think this is pretty much the exact spot. Fourth and a foot.
Now, all spots are a combination of the Line Judge on one side and the Head Linesman on the other side. This play - outside the hashes on the west side, right at the numbers - is going to be the call of the Line Judge on this side. Sometimes he's blocked out and has to rely on the Head Linesman across the field (they work together to get the right spot).
I bring that up to say "just because he put his foot there, that's not the final spot". 98% of the time, when a guy on that side of the field runs in and puts his foot on a spot, that's the spot, but it can be adjusted for various reasons. Here, it was the right spot. But then...
When we come back from the timeout that Rutgers took before the 4th down play, here's where the ball was spotted:
And then the center judge picks up the ball and moves it even further back.
One more time.
AND THEN THE CENTER JUDGE PICKS UP THE BALL AND MOVES IT EVEN FURTHER BACK.
So when Illinois snaps the ball on the all-important fourth down, it's being snapped a full yard behind the original spot.
As I said in the Twitter thread, this was not why we lost. Yes, we probably QB sneak if it's 4th and a foot, but we ran a play that lost four yards. We're not getting a first down there if it was fourth and an inch if that's our play choice (and that's how Rutgers plays it.)
But sweet fancy Moses, what do we have to do to get an accurate spot around here? I've covered bad spots in six of the CTT posts so far. I know that officials spot the ball on hashes now to reduce the number of measurements and speed up the game (have we had a single chain measurement this season?), but when they get addicted to spotting balls on the last hash it crossed - even if it advanced 2.95 feet beyond the previous hash - it creates situations like this.
I mean, let's just say that the ball was spotted where the Line Judge originally put his foot down. And then we sneak it. And then we go the other 33 yards and punch it in to win the game. If that were to have happened - let's say there's a 41% chance we punch it in after picking up the first down - then that spot cost us a game. That's a group of officials costing Illinois one win.
(Wait, I already went though "this is not the reason we lost". Forget that last paragraph.)
Anyway, we lost to Rutgers. It was a 54/46 game - 0.9 points in Illinois' direction, according to Connelly's formula - and Rutgers won. We're now 3-6 in a season where we should probably have right around 2.7 wins after nine games.