A few months ago, I wrote a few posts here discussing what I hope to be a new level of football statistics...
A couple of weeks ago, I officially finished entering play-by-play data for the 2007 football season, and I've been transferring everything into database form. In the last couple of weeks at Rock M Nation, I've begun the offseason look at Mizzou opponents (here's Illinois, and here's Nevada) while still playing with and tweaking the numbers behind the scenes. What will probably happen is, as the summer progresses and I've continued to swim in the data for a while, my vantage point will change.
The first vantage point change occurred within the last day or so, as I've re-calibrated the 'EqPts' associated with each down and yardline. As a refresher, here's what I said about the 'EqPts' idea in my latest Beyond the Box Score Glossary (which I will probably need to update again toward the end of the summer):
If success rates are like on-base percentage, then Points Per Play (PPP) are like slugging percentage. Probably the coolest thing I’ve come up with completely independent of Football Outsiders or anything else is the EqPts measure. It’s based on the average number of points that could be expected when the offense is on any specific yard line. ... Possessions with a play at your own 1-yard line average about 0.9 points per possession. Possessions with a play at your opponents' 1-yard line average about 6.0 points per possession. Your likelihood of scoring doesn't go up much anywhere between your goal line and 20, but it goes up quite a bit between your 30 and 40...and then again between your opponents' 30 and 40. ... Looking at things this way gives you a lot better indication of who accounted for a team’s points than yards or TDs or yards per carry. It measures not only how many yards you gain at any one time, but also how important those yards are toward points actually being scored.
Now that I've got data on all 796 games from the 2007 season, I a) went back and refigured what would be expected from a given down and yardline, b) recalculated the EqPts figures for all the plays, and c) adjusted the scoring slightly to where the total number of EqPts 'scored' on the season through runs, passes, sacks, and penalties equaled the number of actual points scored during the season.
(One thing you'll notice here is that special teams aren't taken into account--just plays and penalties. I will take special teams into account when I look at the difference between the points scored and the EqPts 'scored'. When you think about it, the differences between those two figures should be accounted for mostly by luck and special teams, so I'll use that as a starting point.)
So now that I've reconfigured the numbers, the national averages that I discussed during my Nevada BTBS piece are, two whole days later, outdated. But that's fine.
Anyway, I thought I'd share some of my initial findings. Honestly, a lot of these findings could be discussed in their own post, but I'll just throw a bunch out at you at once.
(And as always, the center point for these discussions comes from three figures: success rates, Points Per Play (PPP), and S&P (success rates + PPP). As I've mentioned before (including in the glossary that you really should make sure you've read before going any further), S&P is, for all intents and purposes, an OPS figure for football. Check out my previous posts above for more info on Success Rates and PPP)
S&P FOR ALL PLAYS (Success Rate / PPP / S&P)
Rushing: 43.3% / 0.33 / 0.762
Passing: 40.5% / 0.29 / 0.697
ALL: 41.9% / 0.31 / 0.730
So as a whole, run plays are more productive/efficient than pass plays. This makes some sense considering that a) we've always known that pass plays are considered higher-risk, higher-reward, b) a lot of pass plays come on 3rd-and-long situations that aren't going to lead to success very often, and c) some of the biggest point values come near the goalline, where teams are much more likely to be running the ball.
S&P DURING CLOSE GAMES (i.e. scoring margin with 16 points, or two possessions)
Rushing: 43.3% / 0.32 / 0.751
Passing: 40.9% / 0.29 / 0.701
ALL: 42.1% / 0.31 / 0.726
Plays during close games end up spreading over the same point/success distribution as plays during blowouts. I honestly don't know if this is interesting or not, but...there you go.
This one, however, IS interesting.
S&P IN PASSING DOWNS AND NONPASSING DOWNS
Looking at sack rates and success rates for different down yardages, here's how I defined what constitutes a Passing Down:
* 2nd-and-8 or more
* 3rd-and-5 or more
* 4th-and-5 or more
There is a tremendous difference in sacks and successes for plays above and below those yardages. And I'd say the numbers back that up.
Rushing: 26.7% / 0.21 / 0.477
Passing: 31.7% / 0.17 / 0.486
ALL: 30.1% / 0.18 / 0.483
Rushing: 47.4% / 0.36 / 0.833
Passing: 47.4% / 0.39 / 0.864
ALL: 47.4% / 0.37 / 0.845
Now...logic would tell you that there's going to be a decent-sized difference between the two, but...that's a HUGE difference. The best offenses stay out of uncomfortable situations. One of the most interesting things about the Nevada breakdown the other day was how well they came through in passing downs. Relying on that seems like a mighty dangerous game to me.
S&P IN THE REDZONE
Rushing: 47.2% / 0.45 / 0.917
Passing: 41.8% / 0.37 / 0.793
ALL: 45.0% / 0.42 / 0.866
First of all, I define the "REDZONE" as anything inside the opponent's 25 yardline instead of the 20. Why? Because in OT you get the ball at the 25, and you're expected to score on every possession. It's silly to think of teams "entering the redzone" when they've got the ball in OT.
This is some nice verification of long-established conventional wisdom--it gets much harder to throw the ball when you get close to the goalline. It will be interesting to see how Mizzou's play-calling changes this year, as they were just as likely to (successfully) throw that inside screen to Martin Rucker from the 2 as they were to run the ball. But that's another post for another time.
S&P BY DOWN
(Success on 1st down = 50% of necessary yardage)
Rushing: 41.2% / 0.30 / 0.716
Passing: 45.9% / 0.38 / 0.836
ALL: 43.1% / 0.33 / 0.766
Always throw the ball on first downs, kids. Okay, that's a lie. This shows you the gamble that teams run with every first down. You have a higher percentage of success and efficiency throwing the ball on first down. But if you're unsuccessful throwing the ball on first down, then you find yourself in a "Passing Down" situation on second (and likely third) down...and we just touched on how difficult it is to succeed in those situations.
(Success on 2nd down = 65% of necessary yardage)
Rushing: 42.6% / 0.31 / 0.737
Passing: 39.6% / 0.33 / 0.722
ALL: 41.1% / 0.32 / 0.730
(Success on 3rd down = 100% of necessary yardage...duh)
Rushing: 43.3% / 0.33 / 0.762
Passing: 40.5% / 0.29 / 0.697
ALL: 41.9% / 0.31 / 0.730
First of all, the initial goal with success rates was to have equal chance of success on every play, and with the heavy load of data (141,000 plays), I'd say that more or less held up. Your chances of success are 43.1% on 1st down, 41.1% on 2nd down, and 41.9% on third down. That's pretty much as close as it's going to get.
Second, it does seem like your best chances of success, from both a Success Rates and PPP perspective, come on 1st down, which makes a bit of sense--what you do on 1st down dictates how the rest of the set of downs will go.
S&P BY QUARTER
Rushing: 43.1% / 0.28 / 0.706
Passing: 42.3% / 0.29 / 0.711
ALL: 42.7% / 0.28 / 0.709
Rushing: 43.7% / 0.37 / 0.808
Passing: 40.9% / 0.31 / 0.715
ALL: 42.2% / 0.34 / 0.759
Rushing: 45.0% / 0.30 / 0.749
Passing: 40.2% / 0.28 / 0.682
ALL: 42.7% / 0.29 / 0.717
Rushing: 41.4% / 0.37 / 0.784
Passing: 38.5% / 0.30 / 0.681
ALL: 40.0% / 0.33 / 0.734
This is odd. At the start of each half (i.e. in Q1 and Q3), plays are more likely to be successful but less likely to be explosive (higher success rates, lower PPP) compared to what happens toward the end of each half (Q2, Q4). From an S&P perspective, offenses become more efficient as each half progresses. This is different than what numbers were telling me before my recalibration--it suggested that the first three quarters were relatively equally productive, and games tended to become a big slog in Q4. Maybe that's still the case as far as close games go, but not overall.
Looking at it by Half...
Rushing: 43.4% / 0.32 / 0.757
Passing: 41.5% / 0.30 / 0.713
ALL: 42.4% / 0.31 / 0.735
Rushing: 43.1% / 0.34 / 0.767
Passing: 39.3% / 0.29 / 0.681
ALL: 41.3% / 0.31 / 0.726
Not a huge difference there by any means, but it appears that rushing gets a little easier and passing a little harder in the second half. Take that for whatever it's worth.
Now here's something interesting...
Rushing: 41.8% / 0.29 / 0.706
Passing: 40.0% / 0.15 / 0.553
ALL: 41.1% / 0.23 / 0.644
So you know how I said that the Redzone starts at the 25? Well, that basically means that all OT possessions start in the Redzone, so you would expect OT numbers to resemble overall Redzone numbers, right? Very wrong. Apparently sphincters start puckering in OT.
So that's all the S&P numbers you should need to see in one post...let's move on to some other tidbits...
LINE YARDS & SACK RATES
Line Yards/carry: 2.89
Sack Rates (Nonpassing downs): 4.82%
Sack Rates (Passing downs): 8.83%
Click here for more info on Line Yards, and keep this in mind as we look at other teams...during the summer
Average point value of a defensive penalty: 0.914 points
Average point value of an offensive penalty: -0.229 points
Suggests that defensive penalties are more costly than offensive ones...
Average point value of a turnover: 3.17 points
HOME FIELD ADVANTAGE
Taking away games against 1-AA teams (which, granted, didn't always lead to success for the 1-A team), I looked at the overall home vs road scoring average to get an idea of home much home field is truly worth:
Home Team Average: 30.5
Road Team Average: 25.9
So that suggests that home field is worth about 5.6 points. I read that 7 points was a reasonable estimate, but this is less. But here's the thing: North Texas played at Oklahoma (so did Utah State); Florida International played at Penn State...and Kansas...and everybody else; Rice played at Texas. You get my drift. Most preordained non-conference blowouts take place at the homefield of the preordained blower-out, not of the preordained blowee-out.
So let's take a look at just conference games. There are plenty of horrid mismatches in conference games too, but whether they take place at home or on the road is a tossup.
Home Team Average (conference games only): 29.5
Road Team Average (conference games only): 26.2
Home field advantage: 3.3 points
I must say, that's surprising. Home field isn't worth nearly what I would have thought. A measly field goal.
What I hope to do over the remaining 70something days until the season starts is take a look at not only upcoming Mizzou opponents and other Big 12 teams, but...well, everybody in some regard or another. I want to look at the best performers in each of the categories I listed above, but I also want to look at who they were performing against (meaning, if you put up big numbers against Washington's or Oklahoma State's schedule, it should mean more than if you did it against Hawaii's or Kansas'). I accept that I won't get to as much as I want, but we'll see where we can go with this...let me know what you think...