Sunday, February 28, 2010

Overtime Omnibus

Lately, every time PFT has a post complaining about OT rules, I think of something else to look at regarding OT games.

Here's some more OT-related stats.

Overtime margin of victory, regular season and playoffs

This is somewhat of an overlap on what I've previously posted, but I want to have the complete stats for regulation and playoffs in one place.

This is for all 445 regular-season overtime games (going back to the 1974 season), and all 27 playoff overtime games (going back to the first one, in 1958).

Margin  Reg. Season%  Playoffs%
0173.8%00.0%
220.4%00.0%
3312 70.1%19 70.4%
611425.6%829.6%

The two overtime games won on a safety are here and here.

Last team to score in regulation

Back on my first post on OT rules, I suggested there is a correlation between OT winners and the last team to score in the regulation period. It turns out there is, though it's much stronger in playoff games than in the regular season.

Of the 27 playoff overtime games (the lst 24 listed here), 18 (66.7%) were won by the same team that scored last in regulation.

Of the 445 regular season overtime games, the last team to score in regulation was 224-204-17 (52.2%, counting ties as a half-win, which may or may not make any sense in this context). Not as marked as the playoffs, but still an advantage.

Home field advantage in overtime

In the playoffs, 15 of the 27 overtime games (55.6%) were won by the home teams, and 9 (44.4%) were won by the visiting teams.

In the regular seasons, 227 (51.0%) overtime games were won by the home team, 201 (45.2%) were won by the visiting team, and 17 (3.8%) ended as a tie. Counting ties as a half-win, that gives the home team a 52.7% win rate,

Note that since 1974, the home team has a 57.8% winning percentage in the regular season and a 66.0% winning percentage in the playoffs.

That 12-4 run for home teams in the playoffs

As noted at the above link, the last 16 playoff overtime games have gone 3-1 for the home team. I'd like to be able to say something about statistical significance here, but my one statistics class was a long time ago (and poorly understood even when it was more recent).

What I can do is run some empirical expermiments, under the following conditions:

  • Assign a home win a probability of 55.6%.

  • Simulate 27 games by generating a random number, and choosing a winner (Home or Visitor) based on that probability.

  • Run this simulation repeatedly, and count how many times the 27 games end with 12 (or more) wins for the home team.

Based on the above simulation (and assumption of probabilty for a home win), it lookes like the probability of a 12-4 run is about 9.2%. Which means it's unusual, but not ridiculously unlikely.

Certainly, this agrees with my intuiton, that 27 games (particularly over the course of 52 years) is pretty thin data to draw any strong conclusions from.

Future analysis

I've actually been avoiding the main complaint against the current overtime system—that the team that wins the toss often scores on their first drive.

I have (mostly complete) play-by-play listings for all regular season (and playoff) games going back to the 2001 season, so at some point I will take a look at the 145 regular season overtime games in this period and see how often the "win the toss and immediately score" scenario happens. (Analysis for the playoff games during this period is here.)

Conclusions

It's clear that (even without the numbers on "win the toss and score immediately") losing the coin toss doesn't automatically mean lose the game.

I'm not particularly sympathetic to arguments that the Vikings were somehow cheated in the NFC Championship Game because they never got the ball on offense in overtime. The fact is, if they had managed to hold on to the ball in regulation, they likely would've won easily. And starting on defense in overtime is a disadvantage? Tell that to the Cardinals.

Still, I'm not totally adverse to some kind of "first to score six" modification of the overtime rules. I just think the "unfairness" of the current system has been overstated.

Sunday, February 21, 2010

NFL: How points were scored in the 2009 regular season

I went through PFR's list of box score data for every regular season game played in the 2009 season, and compiled the following tallies:

UnitType#PointsTotal
OffensePassing TD 711 4266
Rushing TD 428 2568
2-point conversion 24 486882
DefensePick-6 48 288
Fumble-6 26 156
Safety 14 28472
SpecialField goals 756 2268
Teams Extra points 1163 1163
Kickoff return 18 108
Punt return 10 60
Blocked punt return4 24
Blocked FG return 2 123635
Total 10989

Comments:

  • PFR's data seems to be mostly OK, although there are some obvious errors. This game, for example, has a touchdown being scored on a "-21 yard fumble return".

  • All passing and rushing touchdowns are credited to the offense, even though there was at least one scored by special teams. There's no reliable way to tell when a touchdown is scored on special teams using PFR's score format.

  • Another deficiency of PFR's score format: when a conversion attempt fails after a touchdown, it doesn't indicate whether it was an extra point try (and whether it was a bad snap, a bad kick, or a block), or whether it was a 2-point try.

    There were 60 failed conversions. According to this article [nytimes.com], there were 53 2-point conversion tries this season, which means that 29 failed 2-point conversions and 31 failed extra points.

    That gives a 97.6% success rate for extra points, and a 45.3% success rate for 2-point conversions. That may not seem like much, but getting 2 points on success doubles the effectiveness (for an extra point, expected value is .976, for a 2-point conversion, expected value is .906). Of course, that's over the course of a season for 32 teams. Which doesn't always help in one game, one team, one play, for a single good-or-bad outcome.

  • I've said before I'm generally not a big fan of 2-point conversions; I was actually rather surprised by how infrequently they were attempted (53 tries/1163 TDs = 4.6% of the time).

  • I was also surprised by how low the defensive scoring was compared to offense and special teams. As important as a defensive score can be to the course of a game, in terms of total points scored, it's pretty minor.

  • More scoring data coming as soon as I get the box score information parsed out to a more convenient format.

Sunday, February 14, 2010

NFL Wrapup: APR Pick Reliability

2009 Navigation Links

Read this for a description of APR

Introduction

The following is an analysis of APR pick reliability for regular season games.

APR new margin power vs. classic margin power

You may remember at the beginning of the season I announced that I would be using a new margin power function for the APR power rankings this year. APR's picks for this year finished at 152-88. This turns out to be 6 picks weaker than the original 'classic' margin function for the same 2009 games.

The new and classic versions of APR were actually mostly in agreement on game picks; they only picked 8 games differently (correct picks are shown in bold):

WeekGame Pick
(current)
Pick
(classic)
3 Falcons 10,Patriots26Falcons Patriots
4 Jets 10,Saints 24 Jets Saints
5 Patriots 17,Broncos 20 PatriotsBroncos
10Bills 17,Titans 41 Bills Titans
11Chargers 32,Broncos 3 Broncos Chargers
12Giants 6,Broncos 26 BroncosGiants
12Steelers 17,Ravens 20 SteelersRavens
14Redskins 14,Raiders 13 Raiders Redskins

Over the course of the season, I've become increasingly dissatisfied with the new margin power function; this tends to confirm my suspicion that it's not really an improvement.

APR: Home vs Visitor

Here's the breakdown of pick accuracy for picking the visiting or home teams.

Visitor
Won
Home
Won
Total
Picked Right 66 86 152
Picked Wrong 37 51 88
Total 103 137 240

APR picked the home team to win 123 times (86+37), again noticably less than the 137 home teams that did win.

As with last year, this indicates that something needs to be changed so that the home team is picked more often. This year, I think I know what that change needs to be: each team needs to have separate home and away power rankings, and the games picked according to the appropriate power.

If you look at the regular season standings, there are some teams that have a noticable difference between home and road records:

Team Home RecordRoad RecordDifference
Patriots8-02-66
Vikings 8-04-44
49ers 6-22-64
Ravens 6-23-53
Steelers6-23-53
Falcons 6-23-53
Bears 5-32-63
Seahawks4-41-73

I will be very surprised if separating power rankings into home and away categories doesn't yeild at least some improvement in pick reliability.

APR vs. Pythagorean Projection

For week 1 of the season, I posted game picks based on the Pythagorean projections from last year. The Pythagorean picks for this year were 164-92-0 (64.1%), which beats APR (and even APR classic) this year. A 64.1% successful pick rate is best Pythagorean Projection has done at least since 1993 (see results for previous years here, along with an important update).

Conclusions

APR continues to be a work in progress. For next year, I will implement the home/away power rankings for game picks (the weekly power ranking posts will still show a combined power ranking, though).

I'm going to abandon the current margin weight function, either in favor of the classic function, or something better (if I can find something better).

I'm also going to adopt the Pythagorean Projection pick method for week 1 games, so that APR can pick the full season of games.

Friday, February 12, 2010

NFL: Rate of tie and overtime games

Just about two weeks ago, I posted some data analyizing the rate of playoff games that went to overtime from 1970 to 2009. That data shows a marked increase in overtime playoff games since 1999. This post takes a look at the rate of tie and overtime games for the history of the league.

How the data was gathered

  • I used the data as given on pro-football-reference.com.

  • Only NFL games were considered. This means the first two APFA seasons, the AAFC and AFL games are not included in this data.

  • The number of teams in the league and number of games played each season has, of course, fluctuated. Therefore, the numbers for each season are presented as a percentage of games played, in order to make them as comparable as possible season-to-season.

The Data

Here is a graph of the result. The red line represents tie games, the blue line represents overtime games.

Click picture to open full size image

Comments:

  • There's a surprising amount of season-to-season variation, even in the modern era, of games that are tied after four quarters.

  • The rate of overtime playoff games from 1970 to 1998 (3.7%) is at the lower range of regular season overtime games during that era; the rate of overtime playoff games 1998 to 2009 (14.3%) is well over the rate of regular season overtime games.

  • There doesn't seem to be an increase in the number of overtime games in the regular season since 1999. Certainly nothing as dramatic as what has happened in the playoffs.

Interesting stuff I found while looking at the data:

Tuesday, February 9, 2010

NFL Super Bowl Power Rankings

2009 Navigation Links
Read this for a description of APR.

RankTeamW-LPower
Index
This week
1 (↑2)New Orleans
Saints
16-36.8776
(LW × 1.156)
won by 14
at Colts
2 (-)Carolina
Panthers
8-86.3478
(LW × 1.054)
Bye
3 (↑2)Dallas
Cowboys
12-65.9514
(LW × 1.033)
Bye
4 (-)New York
Jets
11-85.8160
(LW × 0.984)
Bye
5 (↓4)Indianapolis
Colts
16-35.6848
(LW × 0.907)
lost by 14
vs Saints
6 (-)Minnesota
Vikings
13-55.5107
(LW × 1.021)
Bye
7 (-)New England
Patriots
10-75.2420
(LW × 0.999)
Bye
8 (-)San Diego
Chargers
13-45.1236
(LW × 1.001)
Bye
9 (-)Atlanta
Falcons
9-74.9684
(LW × 1.018)
Bye
10 (-)Green Bay
Packers
11-64.8731
(LW × 1.005)
Bye
11 (↑2)Philadelphia
Eagles
11-64.7361
(LW × 1.016)
Bye
12 (↓1)Baltimore
Ravens
10-84.6672
(LW × 0.991)
Bye
13 (↓1)Buffalo
Bills
6-104.5583
(LW × 0.974)
Bye
14 (-)Houston
Texans
9-74.4101
(LW × 0.986)
Bye
15 (-)Pittsburgh
Steelers
9-74.3174
(LW × 0.999)
Bye
16 (-)Arizona
Cardinals
11-74.0303
(LW × 1.011)
Bye
17 (-)Miami
Dolphins
7-93.7308
(LW × 1.002)
Bye
18 (↑2)New York
Giants
8-83.6756
(LW × 1.023)
Bye
19 (-)Cincinnati
Bengals
10-73.6120
(LW × 0.996)
Bye
20 (↓2)Tennessee
Titans
8-83.5815
(LW × 0.985)
Bye
21 (-)Denver
Broncos
8-83.4926
(LW × 1.001)
Bye
22 (-)San Francisco
49ers
8-83.3890
(LW × 0.997)
Bye
23 (↑2)Tampa Bay
Buccaneers
3-133.1991
(LW × 1.052)
Bye
24 (↓1)Cleveland
Browns
5-113.1891
(LW × 0.992)
Bye
25 (↓1)Chicago
Bears
7-93.1715
(LW × 1.007)
Bye
26 (-)Jacksonville
Jaguars
7-92.9492
(LW × 0.976)
Bye
27 (-)Kansas City
Chiefs
4-122.7491
(LW × 0.999)
Bye
28 (-)Washington
Redskins
4-122.6800
(LW × 1.024)
Bye
29 (-)Oakland
Raiders
5-112.3928
(LW × 1.000)
Bye
30 (-)Seattle
Seahawks
5-111.9869
(LW × 0.998)
Bye
31 (-)St. Louis
Rams
1-151.4323
(LW × 1.012)
Bye
32 (-)Detroit
Lions
2-141.4049
(LW × 1.007)
Bye

Comments

  • The Saints get a big boost from their 2-score win and retake the top spot.

  • The top 10 looks pretty reasonable... except for the Panthers at #2. They did have some good wins to end the season. But how that translates into a death-grip on the #2 spot, I can't say.

  • The Colts seemed to have a bit of a down year on offense, scoring 416 points in the regular season. Except for last year (when Manning had knee issues), you have to go back to 2001 to find a year with lower production.

    416 is only 9 points off their total for 2006, when they won Super Bowl XLI, but they faced a much weaker opponent then, too.

    There's already talk of the Colts as favorites to win the next Super Bowl. One thing to watch next season will be their point totals. The NFC is likely to send another high-scoring team to the Super Bowl next year, and if the Colts aren't on pace to score at least 430 points over the regular season, they could end up short again.

Epilogue

That's the end of another season, and the final power ranking until September. There will be some more season wrapup posts in the next week or so, and at least a couple historical data posts (I've got the regular season tie frequency post nearly ready).

But after next week, FSPI will be pretty quiet until August or so (you can check out what it was like last year by looking at the "Blog Archive" links on the left).

Once again, thanks for reading. See you in September.

Sunday, February 7, 2010

NFL Super Bowl Pick Results

2009 Navigation Links

Pick results for the Super Bowl.

Saints (SRS) vs. Colts (APR, Line)
The Saints really played a masterful game. They took chances when they needed to, kept the game within reach, and played to their strong passing game.

The Colts, on the other hand... of course, Manning's pick-6 comes up big. That looked to me like just an awful throw. But the real problem was clock management. They needed to have a quick score game plain against the Saints. Instead, their first two second-half drives used 11:48 off the clock and only got them 7 points.

Even if Manning hadn't thrown that interception, if the Colts had held on to the ball and tied the game, the Saints likely would have had two minutes to drive for a go-ahead score, with little or no time left for the Colts to respond.

It almost seems like the Colts were using the same "establish a lead and sit on the ball" strategy that they used against the Ravens and Jets. But that just doesn't work when your oppoent has a quarterback like Drew Brees.

Standings:

This Week Overall
APR  0-15-6
SRS  1-06-5
Line  0-18-3
Unanimous 0-03-1

The final power rankings to come, plus a few more season wrap-up posts.

Thursday, February 4, 2010

NFL Wrapup: APR Picks by team

RecordTeamsComments
14-1 Rams The Rams were not picked to beat the Lions in week 8.
13-2 Chargers
12-3 Lions, Colts, Buccaneers, Bears The Bears were a pretty bad team this year, especially for making it to 7 wins. APR had them ranked in the bottom 10 from week 9 to the end of the season.
11-4 Vikings APR never picked against the Vikings.
10-5 Seahawks, Saints, Redskins, Patriots, Packers, Giants, Eagles, Chiefs, Bengals I'm surprised APR did this well on the Patriots considering the gap between their home record (8-0) and road record (2-6).
9-6 Titans, Ravens, Falcons, Cowboys, Cardinals, Browns, Bills 16 teams picked 9-6 or worse; same as last year.
8-7 Raiders, Panthers, Jets, Broncos, 49ers The 49ers were another team with a big difference between their home record (6-2) and road record (2-6)
7-8 Texans, Jaguars, Dolphins How much does a crowd matter to home field advantage? In spite of their often empty stands, the Jaguars were another team did a lot better at home (5-3) than on the road (2-6).
4-11 Steelers How do you pick a team like this? Four of their wins came against playoff teams (Chargers, Vikings, Packers, and Ravens), while three of their losses came against some of the worst teams in the league (Chiefs, Browns, and Raiders).

Update: I should make clear that this is for regular season games only (weeks 2-17).

Wednesday, February 3, 2010

NFL Wrapup: Pythagorean Projection vs Reality

Last summer, I did a series of posts on Pythagorean Projection (Part I, Part II, Part III, Part IV). As a follow-up to Part III, I present the Projected Wins for 2009 versus Actual Wins.

Note that the format for this table is slightly different than what I used in Part II. Here, the projection was used as a baseline, so a positive difference means a team exceeded its projected wins; a negative difference means a team fell short of its projected wins.

TeamActualProj. DiffComment
Jets 9 9.2 -0.2 New HC, QB
Eagles 1111.3-0.3
Seahawks 5 5.4-0.4 New HC
Chiefs 4 4.4+0.4 New HC, QB
Raiders 5 4.6+0.4
Browns 5 4.4+0.6 New HC
Patriots 1010.6-0.6 Brady back
Falcons 9 9.7-0.7
Lions 2 2.8-0.8 New HC, QB
Jaguars 7 6.2+0.8
49ers 8 6.9+1.1
Rams 1 2.6-1.6 New HC, injuries
Bears 7 8.7-1.7 New QB
Texans 9 7.3+1.7
Broncos 8 6.2+1.8 New HC, QB
Bills 6 7.8-1.8
Dolphins 7 8.8-1.8 Injuries
Cardinals 10 8.0+2.0
Packers 11 8.9+2.1 New defense
Panthers 810.1-2.1 Jake Delhomme
Vikings 12 9.2+2.8 New QB
Steelers 911.8-2.8 Troy Polamalu hurt
Colts 1310.1+2.9 Manning healthy all year
Chargers 1310.1+2.9
Ravens 911.9-2.9
Redskins 4 7.0-3.0
Cowboys 11 7.9+3.1
Giants 811.3-3.3 Eli injured
Saints 13 9.5+3.5
Titans 812.1-4.1 No more Haynesworth
Buccaneers 3 9.0-6.0 New HC, QB
Bengals 10 3.2+6.8 Healthy Carson Palmer

Broken down into the range used in Part I, we have:

Range # of Teams % of 32 Classification
0≤δ≤ 2 17 53.1%  Reasonably Close
2<δ≤ 4 12 37.5%  Moderately Close
4<δ≤ 8 3 9.4%  Wrong
8 00.0%  Yikes!

(Note that in Part I, the difference was expressed in winning percentage; here it is expressed in games won in a 16-game season.)

Comments

  • The teams are sorted by the magnatude of difference; teams that came very close to their projected wins are at the top, teams further away come further down the table.

  • In 2008 (c.f. Part II), 8 teams (25%) finished at least 5 games away from their projected win total, which (as I said) seemed likely to be an outlier year. Here we see teams much closer: only two finished more than 5 games away from their projected wins total.

  • There's also a case to be made for more gradual coaching changes. The Cardinals 2-game improvement (which could've been 3, if they had had something to play for in week 17), can be attributed to continuing improvements under Coach Whisenhunt. While the Redskins 3-game fall-off might be attributed to Coach Zorn losing his squad.

  • Those two or three extra wins that Favre (presumably) brought to the Vikings might not seem like a big deal out of context, but (this year) they're the difference between the #2 seed and being the last wildcard (or maybe eliminated in favor of the Falcons or some other team)

Tuesday, February 2, 2010

NFL Wrapup: Significance of the first two games

Last September, I did a post on how the first two games of the season correlated to final standings. As a follow-up, here's how the 2009 season turned out:

TeamStanding Note
WW Colts 14- 2
TeamsSaints 13- 3
(9) Vikings 12- 4
Jets 9- 7
Ravens 9- 7
Falcons 9- 7
Broncos 8- 8
Giants 8- 8
49ers 8- 8avg wins: 10.0
WL Chargers 13- 3
TeamsEagles 11- 5
(7) Cowboys 11- 5
Packers 11- 5
Patriots 10- 6
Steelers 9- 7
Seahawks 5-11avg wins: 10.0
LW Bengals 10- 6
TeamsCardinals 10- 6
(7) Texans 9- 7
Bears 7- 9
Bills 6-10
Raiders 5-11
Redskins 4-12avg wins: 7.3
LL Panthers 8- 8
TeamsTitans 8- 8
(9) Dolphins 7- 9
Jaguars 7- 9
Browns 5-11
Chiefs 4-12
Buccaneers 3-13
Lions 2-14
Rams 1-15avg wins: 5.0

Comments:

  • I'm somewhat surprised at how well the first two games correlate to overall success. But it's good to see such strong evidence for one of FSPI's basic assumptions (i.e., that you can tell something about how good a team is from just one or two games).

  • It seems a little unusual that most teams fall into the 'WW' or 'LL' categories. I would have expected more of a Normal distribution, with most teams in the 'WL' and 'LW' categories.

  • As with the previous historical analysis, there is a significant difference between the 'WL' and the 'LW' groups, although unlike the historical data, this year it was the 'WL' teams that did noticeably better.