Stat of the Day: Best Games of 2009

As part of our ongoing Stat of the Day series, we're digging deep into our spreadsheets to run a new stat every weekday until Super Bowl XLIV.

Today, let's look at the best single-game DVOA ratings of 2009. Of course, since these are opponent-adjusted, blowing out a good team has more power than blowing out one of the dregs of the league. However, the opponent adjustments are not perfect; as of yet, they don't adjust for how teams develop over the course of the year, so the Eagles get a little bit of a "bonus" for beating the Panthers before the Panthers got their act together. They also don't adjust for who is playing in the game, and you'll notice two Week 17 "sit starters" games on our first list.

The best game of 2009, the Patriots' 59-0 win over Tennessee, ends up as the number five game of the DVOA Era. We've gone over this list a couple times in the last year; for those who never saw it and are curious, the best DVOA game ever was Pittsburgh's victory over the expansion Browns in Week 1 of 1999 (43-0, 145.6% DVOA).

Best Total DVOA Games of 2009

Team

Week

vs.

Score

DVOA
NE 6 TEN 59-0 134.0%
NYJ 17 CIN 37-0 116.8%
NO 12 NE 38-17 116.3%
PHI 1 CAR 38-10 108.0%
BUF 17 "IND" 30-7 104.0%
BAL 18 NE 33-14 97.6%
ARI 13 MIN 30-17 96.4%
PHI 8 NYG 40-17 95.5%
PHI 5 TB 33-14 95.3%
CAR 17 "NO" 23-10 92.3%

Arizona's Wild Card victory over Green Bay is the top offensive game, and number three all-time. The games that surpass it are the Chargers' 45-23 win over the Giants in Week 3 of 2005 (113.7% DVOA) and Kansas City's 45-17 win over Detroit in Week 15 of 2003 (104.1% DVOA).

Best Offensive DVOA Games of 2009

Team

Week

vs.

Score

DVOA
ARI 18 GB 51-45 (OT) 102.8%
IND 2 MIA 27-23 99.5%
PIT 15 GB 37-36 91.9%
NO 12 NE 38-17 89.3%
DAL 1 TB 34-21 78.5%

The Jets' 37-0 win over the Bengals in the final game of the season ends up fourth all-time. The three better defensive games were that Steelers-Browns game from 1999 (-124.1% DVOA), Baltimore's 34-7 win over the Giants in Super Bowl XXXV (-120.2% DVOA), and Tennessee's 31-0 victory over Dallas in the final game of 2000 (-114.4% DVOA).

Best Defensive DVOA Games of 2009

Team

Week

vs.

Score

DVOA
NYJ 17 CIN 37-0 -112.6%
PHI 1 CAR 38-10 -85.7%
BAL 18 NE 33-14 -80.0%
NE 6 TEN 59-0 -79.1%
SF 14 ARI 24-9 -76.7%

We'll close things out with special teams.

Best Special Teams DVOA Games of 2009

Team

Week

vs.

Score

DVOA
CHI 4 DET 48-24 44.7%
MIA 8 NYJ 30-25 38.6%
CLE 4 CIN 23-20 (L) 34.4%
CLE 15 KC 41-34 30.6%
BAL 8 DEN 30-7 29.7%

Comments

24 comments, Last at 27 Jan 2010, 10:54pm

1 Re: Stat of the Day: Best Games of 2009

Nice to see Cleveland do something right this year. ST was great all year. Wonder where the Pittsburgh vs CLE game was in ST dvoa that CLE won and Cribbs returned a kick for TD?

2 Re: Stat of the Day: Best Games of 2009

Also impressive for the Cleveland special teams was that both Dawson and Zastudil missed significant games at times during the season. Especially since Zastudil was replaced with Reggie hodges who was one of the worst punters in the league as shown in a previous stat of the day.

4 Re: Stat of the Day: Best Games of 2009

You've got your CLE/PIT games confused: Cribbs returned a KO for a TD in their week 6 loss to Pittsburgh, but returned a punt to the PIT 8-yard line in their week 14 win.

You can access these with FO Premium database, but I will tease you with the fact that their ST DVOA was about 2.5x better in the win without the TD than it was in the loss with the TD.

3 Re: Stat of the Day: Best Games of 2009

the opponent adjustments are not perfect; as of yet, they don't adjust for how teams develop over the course of the year

………… but one day …… (soon?)……

This is always a fun feature. Because of the opponent adjustments, the teams and units on the wrong end are often good teams having bad days. This year we have the Patriots twice, plus of course the Packers' defense. I would have expected Dallas-Philly to be on here somewhere: that was a typical example.

6 Re: Stat of the Day: Best Games of 2009

Yeah, when I saw the headline, I thought it would be about the closest/most exciting games according to (D)VOA. (This could be done a number of ways -- for example, the smallest single-game differences in VOA.) It would be a great stat for another day.

Of course, seeing a listing of "best" games in terms of the most dominant was interesting, too.

16 Re: Stat of the Day: Best Games of 2009

The problem with subtracting DVOAs is that it might not have been a particularly close game, on the surface.

Let's say Team A's season-long DVOA is +20%, and Team B's is -20%. If they were to play each other to a draw, both single-game VOAs would be 0%, but Team A's DVOA would be -20%, and Team B's would be +20%. This would have been a close game, on the surface, but the difference in DVOA would have been 40%!

Working backwards, let's say we get that each team's single-game DVOA was 0%. This means that Team A must have had a VOA of 20%, while Team B must have had a VOA of -20%. It would not have been a close game, in all likelihood, but the difference in DVOA would be 0%, a close as you can get!

14 Re: Stat of the Day: Best Games of 2009

That would be skewed towards games between good teams.

Now, if you want to measure team quality in there, that's fine; however, I feel that dramatic games like this year's Browns-Lions match should be represented. That's why I think measuring the difference in VOA, rather than DVOA, is a good method for determining the "best games". Maybe call it "most dramatic" games, or "closest games"?

12 Re: Stat of the Day: Best Games of 2009

However, the opponent adjustments are not perfect; as of yet, they don't adjust for how teams develop over the course of the year, so the Eagles get a little bit of a "bonus" for beating the Panthers before the Panthers got their act together

Wouldn't the Patriots over the 0-5 Titans (featuring Kerry Collins, then-30th in DVOA and something like 7th on the all-time lowest single-game DVOA list) also be a good example of that effect, i.e. getting a bonus for beating a team that later turned things around with a different QB after their bye week? At the time, it didn't even crack the top-10 games list, at 117.5% (which is still good enough for #1 on the season list).

(I also like the Eagles)

17 Re: Stat of the Day: Best Games of 2009

I just thought that one was particularly interesting, since the Week 6 article also discussed the single-game DVOAs for NE/TEN and their place in history.

http://www.footballoutsiders.com/dvoa-ratings/2009/week-6-dvoa-ratings

Looking back at the Week 1 page, PHI only jumps from 102.3% to 108.0%; NE 117.5% to 134.0%.

I guess what's really interesting is that NE comes in #1, is on the receiving end of #3 and #6, but is "only" 24th in variance.

(I also like the Eagles)

18 Re: Stat of the Day: Best Games of 2009

The Jets were 10 point favorites over the Bengals in week 17. That alone should be enough to give that game an asterisk.

There of the top 10 games of the season came when an opponent either did not care or was possibly incentivized to lose.

Perhaps we can do a top 10 list with those games removed.

21 Dynamic opponent adjustments

However, the opponent adjustments are not perfect; as of yet, they don't adjust for how teams develop over the course of the year, so the Eagles get a little bit of a "bonus" for beating the Panthers before the Panthers got their act together.

There are a couple things I can think of that might be worth trying. Most of these would only work retroactively -- they would be of no use until after the season was over -- but it could still be helpful for descriptive / evaluative purposes. Also, I don't have an tremendous background in stats, so there might be technical reasons why these methodologies would be invalid.

(1) You may have already tried this, but you could simply find the best-fit curve for each units' DVOA over the course of the season, and use the "expected" value as the basis for adjustments for opponents of that team. I'm pretty sure you already calculate the best-fit lines, but it would require doing so for more than just overall team DVOA, since you break opponent adjustments down by at least passing and rushing.

(2) As a way to fine-tune the rough methodology given above, in instances where there's an obvious qualitative reason to expect a major shift in play (e.g., a QB change or major injury), you could use a Chow test for structural breaks. Of course, if you're looking at full games, you only have a sample size of 16, so you could consider using quarters as your units of analysis.