Week 5: Colorado Recap/UCLA Preview
A Colorado recap, a look at ASU’s run/pass balance, and a UCLA preview that includes a breakdown of the pin-and-pull run concept.
Dear Fellow Smart Sun Devils,
Apologies for the lack of a second newsletter last week. I’ve gone back and forth about whether it’s a better use of my time to do one big newsletter in the middle of the week, or two shorter ones at either end of the week. For now, I’m going back to one newsletter per week that will include both a recap of the previous game and a preview of the coming game. This will force my analyses to be more focused and hopefully unique. My aim is to provide something you won’t find other places - not just generic recaps and previews that you can find in several other places. As always, if you like what you read, sign up for the free email newsletters, and share with your friends.
ASU handled Colorado on Saturday night, 35-13. ASU built a 14-0 lead in the first half, with a few hiccups on offense, before Colorado hit a field goal near the end of the half, then scored a touchdown to start the second half, cutting the lead to 14-10. The offense then got on a roll, scoring the next 21 points before the Buffaloes finished out the scoring with a field goal.
ASU out-gained Colorado 439-250 in total yardage, holding the Buffs to only 67 passing yards on 3.9 yards/attempt, while averaging 10.1 yards/pass attempt and 7.44 yards/play overall themselves.
Here’s the advanced box score, from our friends at CFB Numbers:
This was a game that, as the post-game win expectancy tells us, ASU throughly dominated (which means that given the other advanced stats in the game not knowing the score, if you guessed that ASU won, you’d be right 99.8% of the time). If you saw the score early in the second half, and were frustrated that ASU was struggling with a bad Colorado team, this should disabuse you of that feeling. ASU had a healthy EPA of .313 (better than a 90th percentile performance), but was only moderately efficient, with a 55% success rate for the game, mostly due to some offensive hiccups in the first half, in which they had a success rate of only 40%. The offense did put together an efficient second half, scoring touchdowns on three of their four possessions (not counting their final possession when they ran out the clock), with a success rate of over 70% and 8.4 yards/play. This is against a defense that is considered mediocre by SP+, ranked 67th out of 130 FBS teams, so more meaningful than the offensive performances against bad defenses like SUU or UNLV.
On defense, the key was throughly shutting down Colorado’s passing game, limiting them to 3.9 yards/attempt, and an EPA/pass of -.293. On average, each pass attempt for Colorado made it much more likely that ASU would be the next team to score, and that’s without any turnovers. This was a really good performance by the ASU pass defense, and there was virtually no amount of run success that Colorado could have had that would have overcome such a deficiency in the passing game.
Here’s the QB box score:
This was pretty easily Jayden Daniels’ best game of the season with an EPA/play of .477, a percentile performance in the upper 80s (he had a better EPA/play against SUU, but that was against far inferior competition). He threw for almost 10 yards/attempt (9.4), he was efficient with a healthy success rate, committed no turnovers and was not sacked. He was judicious about when he ran (as opposed to the SUU and UNLV games, when he often took off too quickly), and was very effective running, carrying seven times for 75 yards, for 10.7 yards/rush and two touchdowns, and a rush EPA/play of 1.1. Like the rest of the offense, he was especially efficient in the second half, with a success rate of 77% (both runs and passes included). If there’s something to criticize, it’s that he was only 3-8 on passes thrown more than 10 yards beyond the line of scrimmage. However, the passing game was still quite effective despite the lack of downfield throws. We will continue to monitor his downfield passing as the season goes on.
ASU’s dominant performance helped give them a big bump in the SP+ rankings, from 23 last week all the way up to nine this week. Another factor is that the weight given to preseason data decreased from week three to week four. SP+ considers ASU to be a well-rounded team at this point, ranking them 21st in offense, 16th in defense, and 8th in special teams (remember how concerned we were with special teams after week one?).
Some Thoughts on Run/Pass Balance
Several years ago (2006 to be precise), Chris Brown (the Smart Football guy, not the hip-hop artist) wrote an intriguing article on his site on run/pass balance and game theory - you can find it here. Read it, it’s really good and was pretty forward-thinking in 2006. He posits that rather than equal amounts of rush yards and pass yards, or an equal number of run plays and pass plays, true balance is when a team’s first and second down yards/rush is one yard less than their net yards/pass attempt on those downs, as this is where yards/play is maximized. (Where net yards/pass equals yards/rush is actually the optimization point, but passing carries more risk of a turnover, so he places a one yard/play premium on pass plays. Additionally, he throws out third down, since the goal on third down is to maximize the probability of getting a first down, not to maximize yards). The idea is that as you run (or pass) successfully, teams adjust to that, and your yards/rush (or pass) will go down, and you’ll actually have success passing (or rushing), since the defense is focusing on your run game (or pass game). Using this reasoning, mathematically, the run/pass mix that maximizes yards/play is where yards/run equals yards/pass (but remember, we’re placing a one-yard premium on passes, since they’re more risky). On occasion, I like to check out how ASU’s run/pass balance looks using this methodology.
What got me thinking about this is this chart that CFB Numbers posted on Twitter:
So, instead of looking at yards/rush and yards/pass, we’ll use EPA/rush and EPA/pass to evaluate ASU’s balance. Along the X-axis, we have EPA/pass minus EPA/rush - a positive number would indicate a team has a higher EPA/pass than EPA/rush. Along the Y-axis, we have pass rate. If we find ASU, we can see that on first and second downs, they have a slightly higher EPA/pass than EPA/rush (about .2 higher), and throw the ball less than 30% of the time. Since EPA/play takes into account turnovers (yards/pass attempt does not), we won’t require an EPA/play premium on passes. As ASU’s EPA/pass is higher than their EPA/rush, this would indicate that the Sun Devils aren’t actually passing the ball enough. They run the ball a lot on first and second downs, more than 70% of the time, and because opponents know that, they load the box to stop the run, leaving opportunities for ASU to successfully pass the ball, which they don’t quite take advantage of enough to maximize their EPA/play. They aren’t that unbalanced, as the difference in EPA/play is only about .2, but they could do a little better. So next time a less-informed ASU fan is complaining on social media or a message board that ASU should just run the ball more, since they have a good offensive line and good running backs, you can use some data to show that ASU should actually pass the ball a little more instead.
Additionally, and this is a little beside the point, the fact that ASU runs the ball so often on first and second downs and is still close to maximizing EPA/play is a strong indication that ASU is really good at running the ball.
ASU will take on UCLA in Pasadena this Saturday, in what might be a matchup between the top two teams in the PAC-12 South Division. As noted earlier, SP+ has ASU ranked #9 currently, and UCLA is right behind them at #10. So while the traditional polls have this game as a contest between one top-20 team and one unranked team, SP+ sees this as a matchup between two top-10 teams.
UCLA is not as well balanced between offense and defense as ASU, having the 2nd-ranked offense and 55th-ranked defense per SP+. They average 6.6 yards/play; 4.9 yards/rush and 10.3 yards/pass (they really don’t pass the ball enough, given this amount of success), and are one of the most run-heavy teams in FBS, only passing 37.8% of the time. They are definitely a team that wants to run a lot, then hit you with play-action pass. On defense, they give up 5.3 yards/play; 2.4 yards/rush and 7.2 yards/pass.
Bill Connelly’s SP+ ratings project a 31-29 UCLA victory. FEI (at BCFToys.com) has UCLA as a 3.7-point favorite and the Las Vegas Line is UCLA -3.5 points. These teams are widely seen as evenly matched, with UCLA being favored as the home team.
Out of the UCLA Playbook: Pin-and-Pull With a Backside Read
This is something I haven’t done yet: occasionally I’ll break down a specific play or play concept from either the Sun Devils playbook, or from the playbook of an opponent. Since Chip Kelly is pretty much a genius when it comes to scheming up ways to get numbers advantages in the run game, I thought I would take the opportunity to breakdown a concept that the Bruins frequently use in their run game: the pin-and-pull.
The pin-and-pull is a sweep concept that Kelly helped popularize in the early 2010s while he was the head coach at Oregon. It is always run to the strong side with an in-line tight-end, as it requires a guard, tackle and TE on the play side to execute. Pin-and-pull rules for those three players are thus: if you have a defender lined up inside of you, you block down (pin), and if not, you pull. The center always pulls, unless he has a nose tackle lined up directly over him.
Here’s a diagram of how it wold work against a typical four-man front:
As you can see, both the play side TE and T have defenders lined up directly over them, or to the inside of them, so they block down. The play side G is uncovered, and the C doesn’t have a NT directly over him, so they pull. The first puller is responsible for either sealing the first second-level defender he encounters to the inside, or kicking him out, depending on where he lines up (here, he’s sealing the S linebacker). The second puller will turn upfield and will typically block the M linebacker. The back side G and T follow zone blocking rules as if it were an outside zone to the opposite side with a backside QB read - the backside DE is left unblocked, the QB reads him, and will keep the ball if that backside DE crashes down on the RB. Often, it will not actually be a read (the QB will execute the hand-off no matter what), since the backside DE shouldn’t be able to chase down the RB on an outside run like this, but will carry out a fake as if it were a read.
Here’s how the play looks against a three-man front:
This time, it’s the play side G and T that pull, since they are both uncovered, while the TE blocks down, as does the C, since he has an NT directly over him. Against a three-man front, it’s the backside outside backer that is the key for the QB to read, if there’s a read at all.
There are countless other variations, depending on the number of defensive lineman and how they align, but the rules stay the same. Often, an RPO backside bubble screen is added to it, and the QB will just count the number of defenders over the receivers, and if they are favorable, will pull the ball and throw the screen.
Another wrinkle that Kelly has added is to leave the backside DT unblocked, and have the QB read him. UCLA ran this variation against LSU for a touchdown:
Here’s how that looks diagrammed out:
Here’s another pin-and-pull look, this time with a backside slot receiver also pulling:
UCLA ran these plays with great effectiveness against LSU, on their way to racking up 210 total rushing yards in that game. ASU’s first task on defense will be slowing down UCLA’s rushing attack enough to put them into passing-down situations.
The consensus seems to be that UCLA is a slight favorite at home, and I’m having a hard time disagreeing with that. If both teams are at their best, I think ASU is better, but the Sun Devils have not given me confidence that they will play a clean game with few mistakes, or that they will slow down UCLA’s run game enough to overcome those mistakes. I think UCLA wins a close one, 27-24.
Question of the Week
What are your predictions for the game, and what do you think the Devils need to do to win?