I would be careful with the wording you use and also potentially edit your formula slightly. You say that Michigan is collecting nearly 92% of 'expected' points. From a fans perspective, we probably expect a TD every time we reach the redzone, but in actuality the expected payout of reaching the endzone is less than 7. I would make sure you refer to it as 'potential points' instead. Following that line of thought, I think it would make sense to have have the potential points per trip be 8 points, not 7. I think this will better quantify 2 pt conversions. Under the current formula a successful 2 pt conversion would mean that drive was worth more than the potential points available from that trip to the redzone. This should just shift everyones percentage down a little, but I believe it will be more accurate since in the truest sense a team could score 8 pts on every trip.
Week 5 Weighted Red Zone Efficiency (Michigan #1)
Previously: Week 4 includes discussion of weighted efficiency metric.
Michigan still leads the nation in weighted red zone efficiency. This week I've updated the metric to be slightly more accurate; we now use:
(Total Red Zone Points) / (7 * Red Zone Opportunities)
This takes into account missed extra points and two-point conversions.
|Rank||Name||Red Zone||Points||TD||FG||Weighted Eff|
|74||North Carolina St.||26||122||15||6||0.670|
|76||San Jose St.||9||42||4||5||0.667|
|77||San Diego St.||19||88||10||6||0.662|
|96||New Mexico St.||8||34||4||2||0.607|
Official source data here. One thing bothers me when I look at these stats and I wonder if it's an error or if there's something that I'm missing. After week 4, Michigan was 18/19 in the red zone. After week 5, we're apparently 20/21. But what about the fumble at the 2-yard line against Indiana? How does that not show up as a failed red zone opportunity (thereby putting us at 20/22)? This has got to be a mistake in the NCAA stats, right? What am I missing?
Erroneous data point(s?) or not, Michigan is proving to be very efficient when sniffing the endzone, collecting nearly 92% of expected points once the team gets inside the 20.
Mostly I don't like the idea of a team getting over 100% efficiency. I prefer metrics to make sense in all possible scenarios, not just realistic ones. On the other hand, we could look at points per redzone appearance and avoid the issue completely.
I definitely agree that "expected" was a poor choice of words there.
Maybe "potential points"? Or "optimal points"?
Yeah, I can't imagine a reason that 20/21 isn't totally bogus. It was only one play (and one IU penalty) and it ended with a fumble, but nothing about it makes it anything other than a Red Zone opportunity.
So, if that's the only bad data, I get M down in 4th with 0.877. I'll take it, especially considering the number of times Denard just flies right past the red zone.
The play-by-play must've confused the stat monkey adding RZ opps 1-by-1: MICH 24, IND 2, IND 1, FUMB. By the by, can RR (not that RR) please proceed directly to the end zone instead of stopping inside the 5?
Aside from Michigan, a quick scan reveals some interesting rankings on this list:
Penn State at #117 out of 120, with 6 TDs, 6 FGs, and 6 RZF's (Red Zone Fails) in 18 chances.
LSU at #115 (and it would have been #119 if they hadn't been bailed out by Tennessee's too-many-men penalty when time expired on Saturday)
Also, teams in the state:
- Michigan #1 (or maybe # 4, per jwschultz)
- EMU (!) #17
- Sparty #31 (14 TD/ 3 FG/ 3 RZF)
- Western and Central: Actually, I don't really care about Western and Central.
I was about to say - EMU in the top 20 was the biggest shocker. How are they winless when they're cashing in so often?
EMU has 13 RZ appearances in 5 games... that's 2-3 visits per game. Considering most of the teams they play blow them out, my guess is these RZ visits are occurring in the second half when the game is out of reach and the 2nd and 3rd stringers are in for their opponents. This is what is deceptive about RZ efficiency ratings ranked on percentages. You could make 1 appearance and convert and have a 1.000 rating.
Edit: EMU has 2 blowouts losses, 2 games lost within one score, 1 game lost within 2 scores. So, not all were blowouts.
We made a field goal?
We made one field goal in the UConn game. We arrived at 30 points via four touchdowns, three extra points, and a field goal.
How predictive is the type of data going forward? Do you know of any research on red zone offenses?
It would seem that red zone efficiency would be victim to small sample size issues. Its great in looking back and seeing how great we have been doing in this particular stat. But, how repeatable is this?
I'd actually suggest you take the 2-pt conversions out and make them worth a single point and keep your methodology. If you get the 2 point conversion vs the extra point that doesn't really mean anything in relation to your ability to punch it into the end zone and it artifically inflates the efficiency of teams that fall behind and are forced to go for 2 to tie rather than a team that's ahead or has no reason to go for two. I don't know how time intensive that task is but I think it'd be more indicitive. Well, either that or take our 2 pt conversions and extra points entirely and base it off 6 points (although I don't like that method as much).
Whether potential or optimal, this offense is showing that "expected" might not be that far off.
I wonder if our lack of a field goal team helps us in the red zone. It puts pressure on the O to score since there isn't an alternative.