Thursday, November 22, 2007

NFL Referee Statistics 2007 (after week 11, November 19, 2007)

Here are the updated NFL referee statistics after week 11 (including games through November 19, 2007):

We finally get rid of what we are confident are errors in the NFL's official gamebook. The NFL incorrectly listed "Barry Anderson" and "David Coleman" as referees in weeks 1 and 2, but we put those phantom names to rest now and reassign those games to the referees we believe really handled those games.

The strong correlation between the number of penalties per game and the rate the home team wins is still a striking positive of 0.525. Refs who call more penalties are more likely to have the home team win.

Possible relationship between the referee who calls the game and how many penalties there are and the rate the home team wins the game. One way to check on whether the referee has a relationship to how many penalties are called or the rate the home team wins is to compare those statistics in 2007 with the same statistics for those referees in 2006. Then to see if there is some correlation between the 2006 rates and the 2007 rates for those referees. After 8 weeks, a comparison of those referees with stats in 2006 and 2007 suggests there is a relation between the ref and penalties called or home team win-rate. The correlation for penalties called is a fairly positive 0.301 while the correlation for the home team's win rate is a very large positive 0.461!

Referees by penalties accepted per game (followed by home win rate):
At the top, with a 22-penalty game, Ron Winter takes over top spot from Walt:
Ron Winter (15.8, 56%)
Walt Anderson (15.1, 78%)
Tony Corrente (13.8, 67%)
Terry McAulay (13.5, 70%)

and at the bottom no major changes with Gerald's lead growing very slightly from 0.3 to 0.4:
Pete Morelli (10.1, 22%)
Bill Carollo (10.1, 50%)
Scott Green (9.3, 40%)
Gerald Austin (8.9, 60%)

Referees by home win-rate: if you believe the referee has some influence (perhaps an unintentional tendency) on whether the home team wins, then use this list. Here's the list, again with the penalties per game followed by the home team's win-rate. Walt Anderson takes over the top spot:
Walt Anderson (15.1, 78%)
Terry McAulay (13.5, 70%)
Tony Corrente (13.8, 67%)
Jerome Boger (11.3, 67%)
and at the bottom, Pete Morelli famously extended his lead in the crazy Cleveland Browns overtime victory while visiting the Baltimore Ravens:
Jeff Triplette (10.8, 44%)
Bill Leavy (12.1, 43%)
Scott Green (9.3, 40%)
Pete Morelli (10.1, 22%)

So, if you like the home team, hope for Walt Anderson, Terry McAulay, Tony Corrente, or Jerome Boger. If you like the visiting team, hope for Pete Morelli but be glad if you have Scott Green, Jeff Triplette, or Bill Leavy.

Here is the information, alphabetically, with total penalties, total yards, and home win-rate separated by spaces:
Anderson, Walt 15.1 117 78%
Austin, Gerald 8.9 82 60%
Boger, Jerome 11.3 86 67%
Carey, Mike 13.1 98 50%
Carollo, Bill 10.1 77 50%
Coleman, Walt 11.9 89 56%
Corrente, Tony 13.8 107 67%
Green, Scott 9.3 79 40%
Hochuli, Ed 12.2 93 50%
Leavy, Bill 12.1 96 43%
McAulay, Terry 13.5 108 70%
Morelli, Pete 10.1 80 22%
Nemmers, Larry 11.1 95 60%
Parry, John 12.6 100 60%
Steratore, Gene 12.3 98 60%
Triplette, Jeff 10.8 82 44%
Winter, Ron 15.8 117 56%

Saturday, November 17, 2007

Referee Pete Morelli better for visiting teams?

Is it possible that a referee in some way is better for visiting teams than other referees? By this, we mean that there may be some unintentional bias for visiting teams, not an intentional effort. After all, if anyone was intentionally trying to influence games, they would probably flip flop from week to week which side they were favoring and change their bets accordingly. A long-term trend is more likely to be an unintentional and unconscious bias.

Let's focus on Peter Morelli: is it possible that he favors visiting teams?

Well, visiting teams have done unusually well in games that he has covered. Let's focus over the 2006 and 2007 season (the 2007 season through week 10 or through November 12, 2007). Let's narrow the analysis only to the referees who have worked both in 2006 and 2007. We're taking out the referees who only worked one of the two seasons.

In 2006, Peter Morelli had the best winning percentage for visiting teams -- they won 67% of the time. The next-best was four referees where visiting teams won 53% of the games. So Peter Morelli was the best out of the 16 referees we are studying.

In 2007 (so far), Peter Morelli again has the best winning percentage for visiting teams -- they are winning 75% of his games! The next-best is Scott Green, where visiting teams are winning 67% of the games. (Strangely, in the 2007 season, the visiting teams went 0-2 to start the season and have since then gone a blistering 6-0 in Pete Morelli's games.)

Here are some examples of visiting teams doing unexpectedly well in his games:
2006 Wk 6: Carolina (8-8) beat Baltimore (13-3) 23-21
2006 Wk 10: NY Jets (10-6) beat NE Patriots (12-4) 17-14
2006 Wk 9: Miami (6-10) beat Chicago (13-3) 31-13
2006 Wk 15: SF (7-9) beat Seattle (9-7) 24-14
2007 -- no unusual examples, in each of the six victories, the visiting team has arguably been having a better season overall than the home teams. The one that you might argue was close might have been week 4 where Tampa Bay (now 5-4) beat the Carolina Panthers (now 4-5) by a score of 20-7, where Tampa Bay had one penalty for 5 yards and Carolina had 7 penalties for 65 yards. Maybe you would say sure in 6 of 8 games the visiting team was better, but maybe it means something that the 6 visiting teams won those games?

So, is it an irrelevant statistical oddity that Pete Morelli is best in 2006 and for now also best in 2007 out of the 16 referees who worked in 2006 and 2007? What do you think? Post your thoughts as a comment to this blog posting.

Update: now that the December 18 games are over, how about Peter Morelli happening to be at the Cleveland Browns and Baltimore Ravens game that ended 33-30 in overtime for another visiting team victory after Phil Dawson kicked a 51-yard field goal at the end of regulation, the referees initially called it no good, they conferred and corrected themselves, and they had to get the Ravens to come back out from the locker room to play overtime? Naturally, it was good that Peter Morelli and his crew got the call right, but how about the odds that the game was barely won by the visiting team, just after we pointed out that the visiting team had won 6 straight before that game and visiting teams did the best with him in the 2006 NFL season? Again, post your comments about whether this is an unimportant statistical oddity or might somehow mean something, perhaps an unconscious bias.

NFL referee statistics (after week 10, November 17, 2007)

Here are the updated NFL referee statistics after week 9 (including games through November 12, 2007):

The strong correlation between the number of penalties per game and the rate the home team wins is still a striking positive of 0.434. If you take out two strange entries (Barry Anderson and David Coleman in week one), the correlation is even higher at 0.612! Refs who call more penalties are more likely to have the home team win.

Possible relationship between the referee who calls the game and how many penalties there are and the rate the home team wins the game. One way to check on whether the referee has a relationship to how many penalties are called or the rate the home team wins is to compare those statistics in 2007 with the same statistics for those referees in 2006. Then to see if there is some correlation between the 2006 rates and the 2007 rates for those referees. After 8 weeks, a comparison of those referees with stats in 2006 and 2007 suggests there is a relation between the ref and penalties called or home team win-rate. The correlation for penalties called is a fairly positive 0.283 while the correlation for the home team's win rate is a very large positive 0.418!

Referees by penalties accepted per game (followed by home win rate):
At the top, Walt extends his lead over Ron Winter but not much change in the standings:
Walt Anderson (16.6, 71%)
Ron Winter (15.0, 63%)
Terry McAulay (14.2, 67%)
Tony Corrente (13.9, 75%)

and at the bottom no major changes with Gerald Austin's lead slightly slipping:
Bill Carollo (10.2, 56%)
Pete Morelli (9.6, 25%)
Scott Green (9.0, 33%)
Gerald Austin (8.7, 56%)

Referees by home win-rate: if you believe the referee has some influence (perhaps an unintentional tendency) on whether the home team wins, then use this list. Here's the list, again with the penalties per game followed by the home team's win-rate. Tony Corrente's lead slides a little bit:
Tony Corrente (13.9, 75%)
Walt Anderson (16.6, 71%)
Walt Coleman (11.6, 71%)
Terry McAulay (14.2, 67%)
and at the bottom, Pete Morelli was idle and Scott Green is eating into his lead:
Bill Leavy (12.1, 43%)
Jeff Triplette (10.4, 38%)
Scott Green (9.0, 33%)
Pete Morelli (9.6, 25%)

So, if you like the home team, hope for Tony Corrente, Walt Anderson, Walt Coleman, or Terry McAulay. If you like the visiting team, hope for Pete Morelli but be glad if you have Scott Green, Jeff Triplette, or Bill Leavy.

Here is the information, alphabetically, with total penalties, total yards, and home win-rate separated by spaces:
Anderson, Barry 10.0 50 100%
Anderson, Walt 16.6 124 71%
Austin, Gerald 8.7 78 56%
Boger, Jerome 11.9 90 63%
Carey, Mike 13.8 103 44%
Carollo, Bill 10.2 78 56%
Coleman, David 9.0 84 0%
Coleman, Walt 11.6 81 71%
Corrente, Tony 13.9 109 75%
Green, Scott 9.0 74 33%
Hochuli, Ed 12.4 97 56%
Leavy, Bill 12.1 96 43%
McAulay, Terry 14.2 115 67%
Morelli, Pete 9.6 75 25%
Nemmers, Larry 11.3 97 56%
Parry, John 13.1 105 56%
Steratore, Gene 12.6 98 56%
Triplette, Jeff 10.4 82 38%
Winter, Ron 15.0 109 63%

Saturday, November 10, 2007

John Parry's Patriots-Colts Game in Week 9

Let's take a look back with 9 weeks of the NFL season completed at how John Parry's referee crew called the Patriots and Colts game in week 9.

They had 146 penalty yards accepted against the visiting team. Out of 130 games so far, it was the most penalty yards accepted against the visiting team (1st/130). At 146 yards, it surpassed Terry McAulay's 143 yards in the Philadelphia-NY Giants game of week 4 and the 115 yards that Walt Anderson called in the Green Bay-Kansas City game of week 9.

They had 146 penalty yards accepted against one of the teams. Out of 130 games (260 teams playing games) so far, it was the most that any team had (1st/260):
John Parry, week 9, New England-Indianapolis, 146 yards against NE (NE won)
Larry Nemmers, week 3, Carolina-Atlanta, 135 yards against Atlanta (Carolina won)
Terry McAulay, week 4, Philadelphia-NY Giants, 132 yards against NY Giants (NY Giants won)

They had 171 combined penalty yards accepted. Out of 130 games, that was 4th most (4/130):
Gerald Austin, wk 5, Arizona-St Louis, 184 total yards
Tony Corrente, wk 9, Car-Ten, 181 total yards
Larry Nemmers, wk 3, Car-Atl, 180 total yards
John Parry, wk 9, NE-Ind, 171 total yards

85% of the penalty yardage was against the visiting team. Out of 130 games, that was 2nd most (2/130), amazingly surpassed by Gerald Austin, who in week 6 called only 2 penalties in the Philadelphia-NY Jets game, both against Philadelphia, so 100% of them were against the visiting team.

NFL Referee Statistics (week 9, November 5, 2007)

Here are the updated NFL referee statistics after week 9 (including games through November 5, 2007):

The strong correlation between the number of penalties per game and the rate the home team wins is still a striking positive of 0.425. If you take out two strange entries (Barry Anderson and David Coleman in week one), the correlation is even higher at 0.553! Refs who call more penalties are more likely to have the home team win.

Possible relationship between the referee who calls the game and how many penalties there are and the rate the home team wins the game. One way to check on whether the referee has a relationship to how many penalties are called or the rate the home team wins is to compare those statistics in 2007 with the same statistics for those referees in 2006. Then to see if there is some correlation between the 2006 rates and the 2007 rates for those referees. After 8 weeks, a comparison of those referees with stats in 2006 and 2007 suggests there is a relation between the ref and penalties called or home team win-rate. The correlation for penalties called is a fairly positive 0.286 while the correlation for the home team's win rate is a very large positive 0.481!

Referees by penalties accepted per game (followed by home win rate):
Walt Anderson takes the lead from Ron Winter! (Walt is 16.6, up 0.4 from last week while Ron is now 15.4, down 0.9 from last week's average!)
Walt Anderson (16.6, 71%)
Ron Winter (15.4, 57%)
Terry McAulay (14.1, 75%)
Mike Carey (13.9, 50%)

and at the bottom Gerald Austin takes over the bottom spot at 8.6 (down 0.4 from last week) while Peter Morelli falls into the middle of the pack at 9.6 (up 0.7 from last week):
Pete Morelli (9.6, 25%)
Jeff Triplette (9.3, 43%)
Scott Green (9.3, 38%)
Gerald Austin (8.6, 50%)

Referees by home win-rate: if you believe the referee has some influence (perhaps an unintentional tendency) on whether the home team wins, then use this list. Here's the list, again with the penalties per game followed by the home team's win-rate. Tony Corrente boosts his lead while Walt Anderson slips off the tie for the top.
Tony Corrente (13.6, 86%)
Terry McAulay (14.1, 75%)
Walt Anderson (16.6, 71%)
Jerome Boger (11.6, 71%)
and at the bottom, Pete Morelli extends his lead while Jeff Triplette falls back and Scott Green makes some headway toward the bottom of the list:
Bill Leavy (12.1, 43%)
Jeff Triplette (9.3, 43%)
Scott Green (9.3, 38%)
Pete Morelli (9.6, 25%)

So, if you like the home team, hope for Tony Corrente, Terry McAulay, Walt Anderson, or Jerome Boger. If you like the visiting team, hope for Pete Morelli but be glad if you have Scott Green, Jeff Triplette, or Bill Leavy.

Here is the information, alphabetically, with total penalties, total yards, and home win-rate separated by spaces:
Anderson, Walt 16.6 124 71%
Austin, Gerald 8.6 77 50%
Boger, Jerome 11.7 92 71%
Carey, Mike 13.9 104 50%
Carollo, Bill 10.4 81 63%
Coleman, Walt 10.3 72 67%
Corrente, Tony 13.6 113 86%
Green, Scott 9.3 70 38%
Hochuli, Ed 12.4 98 63%
Leavy, Bill 12.1 96 43%
McAulay, Terry 14.1 115 75%
Morelli, Pete 9.6 75 25%
Nemmers, Larry 11.0 98 50%
Parry, John 12.8 104 63%
Steratore, Gene 12.4 98 50%
Triplette, Jeff 9.3 71 43%
Winter, Ron 15.4 115 57%

Monday, November 05, 2007

Major Controversy Over Referees and the Patriots-Colts Game November 4

"Worst officiated game of the season so far"? "Perhaps the worst officiated game I've ever seen"? These are some of the blog postings so far about the Patriots-Colts game that John Parry's crew officiated on November 4.

One visitor asked whether there are any good articles on the referees in the mainstream press. Great question -- why don't sportswriters pay more attention to refs and how well they perform? It's an area that the fans need newspapers and ESPN commentators to talk about because the NFL and other sports leagues refuse to talk about how well referees do. Even worse, the NFL and NBA and other leagues actually fine owners and players who criticize referees (just ask Mark Cuban). If they are gagged, then it's even more important for journalists and commentators to talk about it!

There is also a responsible way to criticize the refs. Just because the refs blow a bunch of calls does not necessarily mean they are cheating or bribed -- maybe they are having a bad day or they have an unconscious tendency. That tendency could even be something unbiased like always favoring the offenses of both teams. It's like an umpire with a big strike zone as opposed to other umpires with smaller strike zones.

Post your comments on the Patriots-Colts game. Blogger even accepts anonymous comments.

What do you want to focus on? Various people are focusing on:
  • John Parry dropping the coin and saying it was not a coin-flip (Hank Gola of the NY Daily News wrote about this in an article)
  • Offensive interference call on Randy Moss for pushing off
  • No holding calls against the Colts until late in the game
  • No-call on the possible fumble on the 2 yard line
  • Bad call that the Patriots used a challenge to reverse
  • Defense interference call against Hobbs when he seemed to be tackled
  • Late flag on some calls
  • No-call on the defender guarding Kevin Faulk
Go ahead and post comments on the Patriots-Colts game. I'll try to focus another way on referees by updating my statistics to track the various referee crews' tendencies.

Sunday, November 04, 2007

New England Patriots at Indianapolis Colts Referee Preview (Week 9, November 4, 2007)

I'll focus on the New England Patriots game against the Indianapolis Colts today because there is so much interest in it.

Various web sites are reporting that first-year referee John Parry is scheduled to referee the game. Let's take a look at John Parry's season so far:

I'm taking the average points scored and average points allowed for the teams in his games so far then comparing it to the actual points scored and allowed in his games. This will be a rough adjustment for the teams that by chance were in the games he covered so far.

After week 8, the expected visiting team points would be 20.7 per game and the expected home team points would be 20.9 per game. The actual averages are 20.3 for visiting teams and 21.1 for home teams. So there is only a slight +0.6 points per game advantage to the home teams in the games John Parry has covered.

Using that rough measure, the Patriots would be expected to score 28 points and Indianapolis 23.9 points, so that's a 4.0 expected advantage to the visiting Patriots.

John Parry has not had games where the visiting team would have a 4 point expected advantage using that measure. Let's take a look at his seasons, sorted by expected home advantage:

Atlanta at Minnesota (week 1): home team had 3.7 point expected advantage and won by 21.
Cincinnati at KC (week 6): home team had 1.6 expected advantage and won by 7.
New York Jets at Buffalo (week 4): home team had 1.4 advantage and won by 3.
Houston at Carolina (week 2): home team had 1.1 advantage and lost by 13.
Cleveland at Oakland (week 3): home team had -0.6 advantage yet won by 2.
Tampa Bay at Detroit (week 7): home team had -2.4 advantage yet won by 7.
New Orleans at San Francisco (week 8): home team had -3.4 advantage and lost by 21.
[New England at Indianapolis in week 9: home team has -4 advantage and we'll see what happens]

So, even though in John Parry's game the home teams have gone 5-2, the home team either would be expected to win using the measure or just barely behind in 5 of his 7 games.

Let's turn to the over/under. Using the same measure to project the expected total points scored, the teams scored more than expected in 4 out of 7 games so far. Overall, teams have combined for 41.4 points per game while the estimated total points using their season totals so far would have projected 41.6 points per game. (For NE and Indianapolis today, the projection would be 51.9 total points.)

Let's turn to referee totals for the season so far:
12.6 penalties per game (6th our of 17 refs with at least two games this season)
94 penalty yards per game (9th of 17)
7.5 average yards per penalty (13 of 17)
52% of penalties are against the visiting team (12 of 17)
71% home team win rate (tied for 3 out of 17)

If you have suggestions for analyzing John Parry's season so far, let me know by posting a comment.

Saturday, November 03, 2007

NFL Referee Statistics (after week 8)

Here are the updated NFL referee statistics after week 8 (including games through October 29, 2007):

The strong correlation between the number of penalties per game and the rate the home team wins is still a striking positive of 0.496. If you take out two strange entries (Barry Anderson and David Coleman in week one), the correlation is even higher at 0.644! Refs who call more penalties are more likely to have the home team win.

Possible relationship between the referee who calls the game and how many penalties there are and the rate the home team wins the game. One way to check on whether the referee has a relationship to how many penalties are called or the rate the home team wins is to compare those statistics in 2007 with the same statistics for those referees in 2006. Then to see if there is some correlation between the 2006 rates and the 2007 rates for those referees. After 8 weeks, a comparison of those referees with stats in 2006 and 2007 suggests there is a relation between the ref and penalties called or home team win-rate. The correlation for penalties called is a positive 0.339 while the correlation for the home team's win rate is a fairly large positive 0.393!

Referees by penalties accepted per game (followed by home win rate):
Ron Winter (16.3, 67%)
Walt Anderson (16.2, 83%)
Terry McAulay (13.7, 71%)
Mike Carey (12.9, 43%)
and at the bottom are
Jeff Triplette (9.5, 33%)
Scott Green (9.4, 43%)
Gerald Austin (9.0, 43%)
Pete Morelli (8.9, 29%)

Referees by home win-rate: if you believe the referee has some influence (perhaps an unintentional tendency) on whether the home team wins, then use this list. Here's the list, again with the penalties per game followed by the home team's win-rate.
Walt Anderson (16.2, 83%)
Tony Corrente (12.5, 83%)
Terry McAulay (13.7, 71%)
John Parry (12.6, 71%)
and at the bottom:
Mike Carey, Gene Steratore, Bill Leavy, Scott Green, Gerald Austin (tied at 43% home win-rate)
Jeff Triplette (9.5, 33%)
Pete Morelli (8.9, 29%)

So, if you like the home team, hope for Walt Anderson, Tony Corrente, Terry McAulay, or John Parry. If you like the visiting team, hope for Jeff Triplette or Pete Morelli.

Here is the information, alphabetically, with total penalties, total yards, and home win-rate in parentheses:
Barry Anderson (10, 50, 100%)
Walt Anderson (16.2, 117, 83%)
Gerald Austin (9, 79, 43%)
Jerome Boger (11.3, 88, 67%)
Mike Carey (12.9, 98 43%)
Bill Carollo (10.6, 84, 57%)
David Coleman (9, 84, 0%)
Walt Coleman (10.3, 72, 67%)
Tony Corrente (12.5, 101, 83%)
Scott Green (9.4, 70, 43%)
Ed Hochuli (11.9, 93, 57%)
Bill Leavy (12.1, 96, 43%)
Terry McAulay (13.7, 112, 71%)
Pete Morelli (8.9, 72, 29%)
Larry Nemmers (11, 98, 50%)
John Parry (12.6, 94, 71%)
Gene Steratore (12.7, 102, 43%)
Jeff Triplette (9.5, 72, 33%)
Ron Winter (16.3, 119, 67%)