Got my hands on some good play-by-play data, and wanted to try making an advanced box score from scratch. Here's my first attempt:
Definitions, Callouts, etc:
- Scoring Opportunities = Trips inside your opponents 40 yards line
- Success Rate = Percentage of plays where you get:
- 50% of first down yardage
- 70% of second down yardage
- 100% of third or fourth down yardage
- A 'Standard Down' is any down where the defense does not know if you will run or pass (*insert offensive coordinator joke*), such as:
- Any first down
- Any second down with 7 or less yards to gain
- Any third or fourth down with 4 or less yards to gain
- A 'Passing Down' is any down where the offense will likely pass (because they need to pick up significant yardage to stay 'on track') - this includes any offensive snap that is NOT a standard down
- A 'Power Down' (AKA 'Money Down') is 3rd or 4th down from 2 yards or less in which an offense either converted into a 1st down or scored a TD. 1st and 2nd down and goal plays within the 2 yard line are also included.
- Explosiveness is calculated based on Collegefootballdata.com's PPA metric (same thing as EPA). Basically:
- Every possible down, distance, and yard line is assigned a point value based on historical data (I think 5 years of every FBS snap goes into this particular model), this is called Predicted Points (or Expected Points). Here's a good piece if you're interested in the specifics of calculating PP.
- The delta between your starting PP and ending PP is PPA. It can be positive or negative
- Explosiveness is the average of the positive PPA; a small PPA represents a minor situational improvement. A big PPA represents a significant situational improvement. For context, Tuten's 27 yard rush on 2nd & 9 from our own 42 had a PPA = 2.73. Jaylin Lane gaining 7 yards on 1st & 10 from the Marshal 39 had PPA = 0.41.
Callouts:
- None of this is opponent adjusted (like SP+, for example)
- Like I said, I'm creating this from scratch from play-by-play data. I'm seeing some (small) differences between sites, including:
- Yardage - you'll see a difference between the yards I got vs what's on ESPN, not sure why (yet)
- Success Rate - Collegefootballdata.com success rates are all off a few percent from mine. Not sure why (yet)
- Collegefootballdata.com only considers rushing plays when calculating Power Success Rate. I included passes and punts, because I wanted to see how good each team was at high pressure short yardage scenarios.
Forums:
DISCLAIMER: Forum topics may not have been written or edited by The Key Play staff.
Comments
0 success rate on Power downs.
For what it's worth, we were 0/2 on power downs, one was a punt (4th & 2 from the 55 yard line in Q2), the other was the 3rd & 2 that came right before the punt.
Maybe I need to break out this category a little more?
EDIT: I broke this category down more to and renamed it 'money downs'
My eye test has told me that our issue on defense is primarily explosive run plays, especially on so-called power downs, which indicates problems with interior linebacker and safety play.
Question for bar: are the splits (1-4) by quarter or by down?
Quarter! Good question!
2.7 PPA on money downs is... bad. That's basically the equivalent of giving up a field goal every single time the offense has 2 yards or less to go.
Would be interesting to see breakdown by down. What I see is poor 3rd & 4th down explosiveness prevention (this year and last year) wonder if the numbers back that up even outside of power downs.
I think Std downs vs passing downs data should address that?
Think less about the down and more about the situation - basically, 2nd & 6 isn't that different from 1st & 10 or 3rd & 2, because all every situations the defense has to respect the run and the pass.
However, 2nd & 10 is significantly different from 3rd & 1; the offense can take risks and dictate the defense in the latter. In the former, it's the opposite.