The UCF Knights were handed their first loss of the young 2020 season this past weekend versus Tulsa. After two pretty strong offensive performances to open up the season, UCF had arguably their worst offensive showing of the Heupel era in their 34-26 loss on Saturday.
Aside from the obvious (penalties and turnovers), how was Tulsa able to slow down this dynamic passing attack that we’re so used to seeing?
I’ll try to break it down as best as I can, in my opinion, with some context to back up my point.
Using the middle of the field
It’s easy to simply say UCF struggles against a 3-3-5 base look. Tt seemed to be common last season and again vs. Tulsa.
But I tried to look a little more in-depth than that.
This photo (courtesy of @PFF_Seth on Twitter) is a heat map of the frequency of routes run in an area compared to every other team in the country from 2019, essentially showing the utilization of space.
The two areas that UCF is above average come 5-10 yards downfield, to the sidelines. If you watch UCF at all, this doesn’t come as a surprise.
Where UCF is below average is the entire middle of the field. Where they are way below average is the short middle.
So what does this have to do with Tulsa’s defense?
In my opinion, Tulsa used this to their advantage. They forced UCF to play using 2/3rd’s of the field as much as they could and tried to completely forget about the middle of the field. Sounds simple.
In my opinion (and this might be wrong and probably is), UCF’s route tree is based off defensive coverage. By this, I mean UCF runs out-breaking routes vs. an inside technique defender with the exact route determined by coverage depth and number of high safeties, and vice-versa vs. an outside technique defender, which would be an in-breaking route (slant, post, dig, etc.).
Again, that is just an opinion I have, and I could be wrong.
Back to the point:
You can see here, the LB forces the inside slot receiver outside. He’s playing in an underneath zone, and when the ball is thrown here to Marlon Williams, the safety 6 yards deep is the one who closes in on him.
This was a 3x1 formation, and you can see the inside technique being played on both slot receivers. In this play Marlon ran an out route and was the target, but it was just too well defended.
You can kind of see the exact same thing happen to Marlon on this play here:
It seemed as if Tulsa played inside technique on the most inside receiver almost every snap as well as on the outside.
This is also allows Tulsa to play a bit softer coverage and limit the amount of one-on-one deep chances they’re going to give UCF.
A throw from the left hash to right sideline is about 36 yards (!) long even if it is 0 yards downfield (Hashmarks are 17.3 yards from the sideline, the field is 53.3 yards wide).
This is (a) an extremely difficult throw to make consistently, as we are marveled on 36-yard throws downfield, and (b) gives the defenders time to close on orjump routes.
On 46 drop-backs, UCF threw three (3) passes to the middle of the field.
Those three passes resulted in a 15-yard pass interference call, a 14-yard reception, and a 22-yard reception. That’s 55 yards on three plays, gaining 1.364 expected points per play. Now, obviously that’s an extremely small sample, but I’m more trying to illustrate how only 6% of passes were thrown to the middle of the field.
On UCF’s first TD, Ryan O’Keefe ran a post down the middle of the field. Look at the attention it drew:
This left Jacob Harris one-on-one on a corner route for a TD, which is exactly what UCF wants.
I kind of look at it as a fight for space.
Here, Jaylon Robinson was being played to the sideline, and just by faking the slant route (going inside) he was able to get the DB to open.
Now, one thing I believe UCF did well to make Tulsa commit to the middle of the field and kind of open up the space to the outside was by going 12 personnel (1 RB, 2 TE’s and 2 WR).
Outside of the red zone, when UCF lined up in 10, 11 or empty looks, they faced an average of 5.11 men in the box. When they lined up in 12 personnel, that number went up to 6.5.
It’s really just numbers: You have to move men in the box vs. 12 personnel and you’re forced to spread out vs. 10/11/empty, give or take one guy for the most part.
It’s not the best gif from me, but you can see just how packed in Tulsa is here.
12 personnel also indicates run. Two tight ends would make you assume a run play is coming.
UCF only went 12 personnel outside the red zone nine times. This is still a high number compared to season and past season averages.
On those nine plays, they threw on six of them and averaged a gain of 1.045 expected points per play. They gained 76 yards on five pass attempts (four completions) as well as drawing a pass interference call downfield.
Tulsa seemed to come in with a game plan and executed it to perfection, and for that you have to give them the credit they deserve.
UCF did some things that were a nice response at times, but in my opinion, they never were able to force Tulsa to adjust to them.
I’ll fully admit I don’t know much of anything about X’s and O’s from a scheme point so I could be totally wrong on all of this, but to me, that’s what Tulsa seemed to do to neutralize what UCF loves to do.
Obviously penalties are a killer and can kind of derail an offense. Clearly they hurt UCF offensively, but it’d be hard to say that the offense didn’t struggle outside of the penalties.
On 13 offensive penalties, UCF lost a total of 5.7 expected points. That is ... a lot. For those that don’t know, expected points factors in time, score, and field position to give an expected value of each down and distance at a given position on the field. The numbers range from -2 (safety) through 7 (touchdown).
Using data from Power 5 + AAC games (a ~60,000 play data set), I looked at the conversion rate for a set of downs based on down and distances.
1st and 10’s were converted into a 1st down 73.41% of the time, but that drops to 59.42% on 1st and 15 and all the way down to 52.43% on 1st and 20. It just gives you a small idea of how costly penalties are.
The only situation better than a 1st and 10 is a 3rd and 1. That should be enough to show you how important avoiding penalties is.
4th Down Decision Making
Something I enjoy doing is looking at the math behind some 4th down decisions between kicking or going for it. There were two situations that I think would have given UCF a better chance at winning the game by going for it rather than kicking.
But before that, it was interesting to see Heupel take timeouts into the locker room with ~45 seconds and the ball at the end of the second quarter. I have no math behind this to make a point, so it’s all kind of just my opinion.
However, for a team that prides itself on going fast and being explosive, it was kind of head scratching. They were coming off 10-play and 13-play drives, which were probably the best two drives they had all game.
It hasn’t bit UCF really at all, but sometimes playing not to make a mistake comes back to hurt. Obviously they would’ve loved the possession in hindsight. 48 seconds in college football with timeouts in your pocket is just a lot of time.
On to the 4th down plays:
The first one:
Up by 14, 8:31 left second quarter, 4th and Goal on the opponents’ 4.
Result: Obarski 21-yard FG missed
The things that matter:
- EP of successful conversion: 3.57
- 4th and 4 conversion rate: ~43%
- EP if you don’t get 4th and 4 (1st and 10 down 14 on own 4): -0.5408199 (+ from a UCF perspective)
- Field Goal Probability (20-25 yards for larger sample size): 95.7%
- Field Goal EP (3-factoring in Tulsa getting the ball): 1.91
- EP of missed FG starting field position (1st and 10 down 14 on own 20): 0.69975321
Factoring all of this information, the value of going for it is 1.8473 expected points, while the value of kicking a field goal is 1.7978. Pretty close to a toss-up here.
UCF kicked the field goal, which you can make the case for on either side.
The one advantage, in my mind, of going for it is the fact Tulsa is starting inside their own 5 if you don’t convert. This puts any offense at a major disadvantage.
While a successful kick gives you 3 points, it’s still probably looked at as a win for the defense considering how successful of a drive the offense had, but it also allows the offense to start at their own 25 (assuming touchback). Even on a missed FG, kicks missed inside the 20 give the ball to the opponent on the 20 as opposed to the previous LOS.
This image from @statsowar shows how few drives result in TDs starting between the 10 and 20, let a lone inside the 10. (Labeled 23-26 = 10-40)
10:59 4th quarter, 4th and 7 at opponents’ 43, Down 5
Result: Osteen punts 43 yards for a touchback
This one hurt because we’ve seen Heupel go for in essentially this exact same down, distance, yard line combination with the lead.
The things that matter;
- EP of successful conversion (worst case, 1st and 10 at opponents 36): 0.97048353
- Conversion Rate: 40.7%
- EP of Failed conversion (1st and 10 on own 43 up 5): 0.91233182
- EP of punt (Using a punt down at the 17): 0.47066596 (average punt from 43 was downed at 11 in ‘19, but this was the closest EP number I could get)
Going for it has an expected value of (-)0.146 points and punting has an expected value of (-)0.471 points. To be fair, the gap is a little smaller with those 6 yards of field position, but there’s still more value in going for it, although smaller.
UCF’s offense wasn’t moving the ball too well, but the defense was very banged up as well, so it’s kind of a give-or-take if you want to throw in the game feel.
All in all, I don’t expect every decision made to favor the math because of the other “real” factors that are going on, but in close games you realize how important these decisions can actually be.
In the end, you have to give Tulsa the credit they deserve, but at the same time I felt UCF helped out Tulsa by almost playing into their game plan (in my opinion). When you fall behind the chains and maybe make a conservative decision that you wouldn’t make if you could do it again, it makes it extremely difficult to win, and I think we saw that.
All EPA data from @cfbscrapr.