tag:blogger.com,1999:blog-8810508161008877312012-01-29T10:31:46.946-08:00A blog devoted to the research of predicting the outcomes of National Football League games.Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.comBlogger1521100tag:blogger.com,1999:blog-881050816100887731.post-80707796198691246102008-02-11T21:38:00.000-08:002008-02-11T22:43:22.882-08:00

For the next couple weeks, I'm going to be auditing the various statistics and systems I use to see what works, what doesn't. First up, the win total projections based on plugging in preseason stats into a linear regression system trained on regular season stats and win totals. The offensive stats used were those for projected starters only, and the defensive stats were total defensive stats.

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com9tag:blogger.com,1999:blog-881050816100887731.post-74291301409708593832008-02-03T22:10:00.000-08:002008-02-03T22:24:11.391-08:00

.nobrtable br { display: none }Home TeamAway TeamP(Home Team Won) (%)Final Score MarginNENYG44.5644-3Despite the home team bias of the system given the neutral site, the Giants came out with the higher win probability. Their 2 fumbles (both recovered) and the one interception weren't enough to cancel out the Brady's fumble and the Patriots' poor performance in general.As I pointed out in my

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com1tag:blogger.com,1999:blog-881050816100887731.post-59552931449243324912008-01-29T21:02:00.000-08:002008-01-29T22:11:55.895-08:00

Stats used in the preview are unadjusted for opponent and cover the regular-season only.Patriots Run Offense vs. Giants Run DefensePats' Run Off: 14th in league, 4.0998 yards per carry, 1.6663% VOLA (Value Over League Average)Giants' Run Def: 9th, 5.7191 yards per carry, 3.8020% VOLAAdvantage: Giants, worth 0.4905 pointsThe interesting thing about the Patriots this season has been their seemingly

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com1tag:blogger.com,1999:blog-881050816100887731.post-87426239282470093922008-01-20T21:49:00.000-08:002008-01-20T21:59:48.802-08:00

These probabilities of victory are based on the box score stats, training on 1996-2006. Inputs include rushing and passing efficiency by yards per play, third down efficiency, sack rates, turnover rates and penalty yards. The difference between probability and result can stem from recovered fumbles, where the ball was turned over, special teams, among other things..nobrtable br { display: none

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-52862052430806240912008-01-14T20:14:00.000-08:002008-01-14T20:15:28.963-08:00

.nobrtable br { display: none }Predictions based on opponent-adjusted statsGamePredicted Final Score MarginP(Home Team Wins)SD @ NE7.314872.2577NYG @ GB4.143664.2149Looks like another Pats-Packers Super Bowl coming up.

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com1tag:blogger.com,1999:blog-881050816100887731.post-41137486564187414752008-01-14T01:27:00.000-08:002008-01-14T20:11:26.830-08:00

These probabilities of victory are based on the box score stats, training on 1996-2006. Inputs include rushing and passing efficiency by yards per play, third down efficiency, sack rates, turnover rates and penalty yards. The difference between probability and result can stem from recovered fumbles, where the ball was turned over, special teams, among other things.As you can see, all of the

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-14188599151682561212008-01-07T20:40:00.000-08:002008-01-07T21:03:02.244-08:00

.nobrtable br { display: none }Predictions based on opponent-adjusted statsGamePredicted Final Score MarginP(Home Team Wins)JAX @ NE5.210565.7171SD @ IND7.574774.7456NYG @ DAL6.428770.6450SEA @ GB4.250064.4811Last week, my predictions and the system's predictions went 2 for 4.Let's take a look at the matchups I didn't see coming.NYG @ DALRush Off: NYG - 4.5940 yards per carry (3rd), DAL - 4.1799

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-31899032495990950792008-01-06T22:00:00.000-08:002008-01-06T22:04:10.865-08:00

.nobrtable br { display: none }Home TeamAway TeamP(Home Team Won) (%)Final Score MarginPITJAX71.8961-2SEAWAS77.948521TBNYG10.7964-10SDTEN90.368911Teams that "should have" won but didn'tPittsburgh Steelers (vs. Jaguars), Win Prob = 71.8961% Pittsburgh was more efficient passing (6.2 yards per pass vs. 4.2) and on 3rd downs. Despite the fact that Pittsburgh threw 3 picks and gave up a fumble, they

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-10475837531418894712008-01-04T18:28:00.000-08:002008-01-04T18:56:09.516-08:00

After the article I posted on Wednesday, Doug Walters over at the Football Professor blog and I decided to do another podcast about our expectations for the 2007 playoffs. Listen to it here or on Doug's blog.

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-59795378073103553902008-01-02T18:12:00.001-08:002008-01-03T22:20:59.500-08:00

As my prediction system is going to be picking all the home teams from this point out, I thought I'd go through each round and pick games using the system's predictions, reasoning based on past research, and maybe a smidgen of personal bias. (No, the tattoo is German for "The Brady, The.")WILD CARD ROUND.nobrtable br { display: none }Predictions based on unadjusted statsGamePredicted Final Score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com3tag:blogger.com,1999:blog-881050816100887731.post-50315840383493950092008-01-02T18:12:00.000-08:002008-01-02T18:15:09.750-08:00

.nobrtable br { display: none }Predictions based on opponent-adjusted statsGamePredicted Final Score MarginP(Home Team Wins)JAX @ PIT3.649362.5294TEN @ SD4.585064.7420NYG @ TB4.375165.1118WAS @ SEA0.293451.9924.nobrtable br { display: none }Predictions based on unadjusted statsGamePredicted Final Score MarginP(Home Team Wins)JAX @ PIT4.552865.8528TEN @ SD5.198966.2033NYG @ TB4.850767.0733WAS @

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-85492391724238599552008-01-02T18:07:00.000-08:002008-01-02T18:08:16.004-08:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The power rankings use logistic regression to predict the probability of winning. The score you see along with the rankings is the expected win total over a 16-game season

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-26924942625729686762008-01-02T18:06:00.001-08:002008-01-02T18:06:59.883-08:00

All GamesHome Team Win %: 56.6667%Avg. Final Score Margin: 2.7000InterconferenceHome Team Win %: 65.0000%Avg. Final Score Margin: 5.6000Interdivision, IntraconferenceHome Team Win %: 63.0435%Avg. Final Score Margin: 3.2935IntradivsionHome Team Win %: 44.3182%Avg. Final Score Margin: 0.1023

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-87886323304316833922008-01-02T18:03:00.000-08:002008-01-02T18:06:15.505-08:00

These are the final statistical rankings of 2007. The stats included are what I use in my predictions..nobrtable br { display: none }Rush offense, Unadjusted (League Avg = 4.0326)RankTeamVOLA (%)Actual

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-25901169010547855342007-12-31T09:26:00.000-08:002007-12-31T09:43:10.085-08:00

The following table lists the probabilities that the home team would win each game given the rushing averages, passing averages (includes sack yards lost and sacks as pass attempts), sack rates, third down conversion rates, penalty yards, and interception and fumble rates (by number of completions and rushing plays) for both teams. A logistic regression model trained on 1996-2006 data computes

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com2tag:blogger.com,1999:blog-881050816100887731.post-63211658881614076962007-12-25T13:12:00.002-08:002008-01-02T18:13:12.307-08:00

.nobrtable br { display: none }Predictions based on unadjusted statsGamePredicted Final Score MarginP(Home Team Wins)CIN @ MIA-0.117650.1338KC @ NYJ3.067461.8436PIT @ BAL-2.806442.2222SF @ CLE7.636873.4936JAX @ HOU-0.554747.5887TEN @ IND8.826177.2495MIN @ DEN0.852948.2096SD @ OAK-1.698440.4813NE @ NYG-2.116042.3982BUF @ PHI6.513870.5099DAL @ WAS-1.107746.4259NO @ CHI-0.146949.6856DET @

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-82673607896686183562007-12-25T13:12:00.001-08:002007-12-25T13:19:40.439-08:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The power rankings use logistic regression to predict the probability of winning. The score you see along with the rankings is the expected win total over a 16-game season

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-65445943469285457232007-12-25T13:12:00.000-08:002007-12-25T13:16:05.642-08:00

This is the data I use to make my predictions and power rankings. Below are the tables for the raw stats and the values over league average for all of the inputs I use. Rush and pass stats are yards per play. See for yourself where your team's strengths and weaknesses are..nobrtable br { display: none }Rush offense, Unadjusted (League Avg = 4.0623)RankTeamVOLA (%)Actual

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-36608495006433387312007-12-25T13:09:00.000-08:002007-12-25T13:23:03.126-08:00

One of the things I've learned in my football research is that the value of home-field advantage varies pretty widely from year to year, especially with interconference games. Because prediction systems tend to have a big bias towards the home team, this can have a large impact on accuracy. So throughout the season, I'll be keeping track of home team winning percentage and average final score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-20634500100332165752007-12-24T09:47:00.000-08:002007-12-25T13:14:37.533-08:00

The following table lists the probabilities that the home team would win each game given the rushing averages, passing averages (includes sack yards lost and sacks as pass attempts), sack rates, third down conversion rates, penalty yards, and interception and fumble rates (by number of completions and rushing plays) for both teams. A logistic regression model trained on 1996-2006 data computes

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-14323415393697119662007-12-18T17:56:00.004-08:002007-12-18T18:05:57.036-08:00

The predictions are based on yards per play on rushing and passing plays (sacks and sack yards included), sack rates, third down conversion rates, penalty yards per game, and fumble and interception rates. The models are trained on the 1996-2006 seasons. Linear regression predicts the final score margin (home team points - away team points). Logistic regression estimates the probability that

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-75322693735534463132007-12-18T17:56:00.003-08:002007-12-18T18:03:09.335-08:00

One of the things I've learned in my football research is that the value of home-field advantage varies pretty widely from year to year, especially with interconference games. Because prediction systems tend to have a big bias towards the home team, this can have a large impact on accuracy. So throughout the season, I'll be keeping track of home team winning percentage and average final score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-65105482084833712922007-12-18T17:56:00.002-08:002007-12-18T18:02:00.117-08:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The power rankings use logistic regression to predict the probability of winning. The score you see along with the rankings is the expected win total over a 16-game season

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-28695161548469960352007-12-18T17:56:00.001-08:002007-12-18T18:00:28.781-08:00

Using the projected win probabilities from the logistic regression power rankings, I estimated which games each team would win on the rest of their respective schedules. And then I summed the projected wins with the actual wins so far to get their projected win totals. The teams are sorted by division..nobrtable br { display: none }Projected Final StandingsTeamProjected Win TotalAFC

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-77002329379093051032007-12-18T17:56:00.000-08:002007-12-18T17:59:32.134-08:00

This is the data I use to make my predictions and power rankings. Below are the tables for the raw stats and the values over league average for all of the inputs I use. Rush and pass stats are yards per play. See for yourself where your team's strengths and weaknesses are..nobrtable br { display: none }Rush offense, Unadjusted (League Avg = 4.0400)RankTeamVOLA (%)Actual

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-89700673150729686502007-12-17T20:36:00.000-08:002007-12-18T17:57:32.847-08:00

Night games added...The following table lists the probabilities that the home team would win each game given the rushing averages, passing averages (includes sack yards lost and sacks as pass attempts), sack rates, third down conversion rates, penalty yards, and interception and fumble rates (by number of completions and rushing plays) for both teams. A logistic regression model trained on 1996-

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-45657853393703742602007-12-16T23:02:00.000-08:002007-12-16T23:55:14.027-08:00

See, this is what I've been saying about the Dolphins all season. With all of the injuries they've sustained and the ridiculous ways they've lost 6 games by field goals, the luck was eventually going to swing the other way. The other team would sustain the injuries, miss field goals, and make the bad decisions that would lead to the Dolphins' first victory. Even with the two out-of-bounds

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-18585715654739599242007-12-12T19:17:00.000-08:002007-12-12T19:59:47.904-08:00

The first regular feature I post every week after the games is the expected win probabilities for each game given the box score stats (i.e. in-game only performance of both teams). These probabilities are the result of a logistic regression model trained on the 1996-2006 seasons.P(team X wins | box score)=99.9% means that team X completely dominated the game statistically. It is easier to

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-23343997434361710692007-12-11T18:08:00.000-08:002007-12-11T18:38:20.091-08:00

This comes from an e-mail conversation I had at work with the guys I watch the games with. One of them posed the question of how many undefeated teams can there be in the same season. Below is an informal proof answering the question.An undefeated team will have won all of their divisional games. So the maximum possible number is ≤8.Each team in conference A will play all the teams in one of

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com3tag:blogger.com,1999:blog-881050816100887731.post-50329666104165924302007-12-11T18:00:00.000-08:002007-12-11T17:49:07.031-08:00

Download podcast here (you'll have to enter a CAPTCHA code then wait about a minute)Also available at the top of the Football Professor blogBeing a "network," we here at the blog like to network with other people to share ideas and find new ones. This is one of the fruits of that effort. The interview touches on what predictions of mine did/didn't work out this season so far and predictions for

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com2tag:blogger.com,1999:blog-881050816100887731.post-37353853367567145812007-12-11T17:37:00.004-08:002007-12-11T17:45:28.329-08:00

Using the projected win probabilities from the logistic regression power rankings, I estimated which games each team would win on the rest of their respective schedules. And then I summed the projected wins with the actual wins so far to get their projected win totals. The teams are sorted by division..nobrtable br { display: none }Projected Final StandingsTeamProjected Win TotalAFC

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-81731938389346317382007-12-11T17:37:00.003-08:002007-12-11T17:45:49.584-08:00

One of the things I've learned in my football research is that the value of home-field advantage varies pretty widely from year to year, especially with interconference games. Because prediction systems tend to have a big bias towards the home team, this can have a large impact on accuracy. So throughout the season, I'll be keeping track of home team winning percentage and average final score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-54408180184321107652007-12-11T17:37:00.002-08:002007-12-11T17:47:20.321-08:00

The predictions are based on yards per play on rushing and passing plays (sacks and sack yards included), sack rates, third down conversion rates, penalty yards per game, and fumble and interception rates. The models are trained on the 1996-2006 seasons. Linear regression predicts the final score margin (home team points - away team points). Logistic regression estimates the probability that

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-3833905356457961082007-12-11T17:37:00.001-08:002007-12-11T17:44:35.723-08:00

This is the data I use to make my predictions and power rankings. Below are the tables for the raw stats and the values over league average for all of the inputs I use. Rush and pass stats are yards per play. See for yourself where your team's strengths and weaknesses are..nobrtable br { display: none }Rush offense, Unadjusted (League Avg = 4.0288)RankTeamVOLA (%)Actual

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-67778063217220236842007-12-11T17:37:00.000-08:002007-12-11T17:39:25.562-08:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The power rankings use logistic regression to predict the probability of winning. The score you see along with the rankings is the expected win total over a 16-game season

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-77287579844110374562007-12-09T18:36:00.000-08:002007-12-09T18:40:01.382-08:00

The following table lists the probabilities that the home team would win each game given the rushing averages, passing averages (includes sack yards lost and sacks as pass attempts), sack rates, third down conversion rates, penalty yards, and interception and fumble rates (by number of completions and rushing plays) for both teams. A logistic regression model trained on 1996-2006 data computes

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com2tag:blogger.com,1999:blog-881050816100887731.post-5044177771666079392007-12-04T20:00:00.000-08:002007-12-04T23:32:38.318-08:00

The predictions are based on yards per play on rushing and passing plays (sacks and sack yards included), sack rates, third down conversion rates, penalty yards per game, and fumble and interception rates. The models are trained on the 1996-2006 seasons. Linear regression predicts the final score margin (home team points - away team points). Logistic regression estimates the probability that

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-42140677317052058592007-12-04T18:07:00.004-08:002007-12-04T23:29:48.926-08:00

One of the things I've learned in my football research is that the value of home-field advantage varies pretty widely from year to year, especially with interconference games. Because prediction systems tend to have a big bias towards the home team, this can have a large impact on accuracy. So throughout the season, I'll be keeping track of home team winning percentage and average final score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-73792722274756168292007-12-04T18:07:00.003-08:002007-12-04T23:28:34.420-08:00

Using the projected win probabilities from the logistic regression power rankings, I estimated which games each team would win on the rest of their respective schedules. And then I summed the projected wins with the actual wins so far to get their projected win totals. The teams are sorted by division..nobrtable br { display: none }Projected Final StandingsTeamProjected Win TotalAFC

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-21301502892497165302007-12-04T18:07:00.002-08:002007-12-04T23:27:48.068-08:00

This is the data I use to make my predictions and power rankings. Below are the tables for the raw stats and the values over league average for all of the inputs I use. Rush and pass stats are yards per play. See for yourself where your team's strengths and weaknesses are..nobrtable br { display: none }Rush offense, Unadjusted (League Avg = 4.0288)RankTeamVOLA (%)Actual

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-16136566245864121902007-12-04T18:07:00.001-08:002007-12-04T23:26:28.598-08:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The power rankings use logistic regression to predict the probability of winning. The score you see along with the rankings is the expected win total over a 16-game season

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-86962574921132899842007-12-02T21:35:00.000-08:002007-12-04T18:07:09.020-08:00

Night games added, tables updatedThe following table lists the probabilities that the home team would win each game given the rushing averages, passing averages (includes sack yards lost and sacks as pass attempts), sack rates, third down conversion rates, penalty yards, and interception and fumble rates (by number of completions and rushing plays) for both teams. A logistic regression model

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-22934501318237818792007-11-26T21:33:00.000-08:002007-11-27T19:09:38.529-08:00

The predictions are based on yards per play on rushing and passing plays (sacks and sack yards included), sack rates, third down conversion rates, penalty yards per game, and fumble and interception rates. The models are trained on the 1996-2006 seasons. Linear regression predicts the final score margin (home team points - away team points). Logistic regression estimates the probability that

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com2tag:blogger.com,1999:blog-881050816100887731.post-51268867358660968662007-11-26T21:25:00.001-08:002007-11-27T19:03:00.252-08:00

One of the things I've learned in my football research is that the value of home-field advantage varies pretty widely from year to year, especially with interconference games. Because prediction systems tend to have a big bias towards the home team, this can have a large impact on accuracy. So throughout the season, I'll be keeping track of home team winning percentage and average final score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-43282834190234352062007-11-26T21:24:00.001-08:002007-11-27T19:05:22.151-08:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The power rankings use logistic regression to predict the probability of winning. The score you see along with the rankings is the expected win total over a 16-game season

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-24347237981317144242007-11-26T21:22:00.001-08:002007-11-27T18:54:42.480-08:00

Using the projected win probabilities from the logistic regression power rankings, I estimated which games each team would win on the rest of their respective schedules. And then I summed the projected wins with the actual wins so far to get their projected win totals. The teams are sorted by division..nobrtable br { display: none }Projected Final StandingsTeamProjected Win TotalAFC

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-14498155868409182002007-11-26T21:20:00.000-08:002007-11-27T18:54:19.894-08:00

This is the data I use to make my predictions and power rankings. Below are the tables for the raw stats and the values over league average for all of the inputs I use. Rush and pass stats are yards per play. See for yourself where your team's strengths and weaknesses are..nobrtable br { display: none }Rush offense, Unadjusted (League Avg = 4.0045)RankTeamVOLA (%)Actual

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-37797682663749494672007-11-25T20:45:00.000-08:002007-11-27T19:57:41.912-08:00

The following table lists the probabilities that the home team would win each game given the rushing averages, passing averages (includes sack yards lost and sacks as pass attempts), sack rates, third down conversion rates, penalty yards, and interception and fumble rates (by number of completions and rushing plays) for both teams. A logistic regression model trained on 1996-2006 data computes

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-8540346110843723222007-11-20T21:35:00.000-08:002007-11-20T21:54:04.628-08:00

The predictions are based on yards per play on rushing and passing plays (sacks and sack yards included), sack rates, third down conversion rates, penalty yards per game, and fumble and interception rates. The models are trained on the 1996-2006 seasons. Linear regression predicts the final score margin (home team points - away team points). Logistic regression estimates the probability that

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-45505353761274994322007-11-20T21:24:00.000-08:002007-11-26T21:33:45.993-08:00

One of the things I've learned in my football research is that the value of home-field advantage varies pretty widely from year to year, especially with interconference games. Because prediction systems tend to have a big bias towards the home team, this can have a large impact on accuracy. So throughout the season, I'll be keeping track of home team winning percentage and average final score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-16261296373127020872007-11-20T21:19:00.002-08:002007-11-26T21:32:51.267-08:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The power rankings use logistic regression to predict the probability of winning. The score you see along with the rankings is the expected win total over a 16-game season

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-81703951251209136692007-11-20T21:19:00.001-08:002007-11-26T21:31:20.070-08:00

Using the projected win probabilities from the logistic regression power rankings, I estimated which games each team would win on the rest of their respective schedules. And then I summed the projected wins with the actual wins so far to get their projected win totals. The teams are sorted by division..nobrtable br { display: none }Projected Final StandingsTeamProjected Win TotalAFC

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-80663865167590718642007-11-20T21:19:00.000-08:002007-11-26T21:30:54.863-08:00

This is the data I use to make my predictions and power rankings. Below are the tables for the raw stats and the values over league average for all of the inputs I use. Rush and pass stats are yards per play. See for yourself where your team's strengths and weaknesses are..nobrtable br { display: none }Rush offense, Unadjusted (League Avg = 4.0188)RankTeamVOLA (%)Actual

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-641456235215505462007-11-19T00:47:00.000-08:002007-11-20T21:37:13.196-08:00

Night games added and tables updated. With the Thanksgiving holiday coming up, I'm going to scale back the articles I normally post. I'll just update this article and post predictions for week 12. The rest of the updates will be retroactively added over the weekend. Happy Turkey Day!The following table lists the probabilities that the home team would win each game given the rushing averages,

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-83443807228570909232007-11-13T22:11:00.000-08:002007-11-14T19:50:39.013-08:00

Tables labeled properly now.In Part I, I looked at the year-to-year correlations of wins to answer the titular question. In short, the team of this year is not very much like the team of last year. The draft, injuries, aging, free agency, development of players, and plain old luck mean there's always going to be a lot of change from year to year. The team of four years ago is (on average)

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com2tag:blogger.com,1999:blog-881050816100887731.post-75816138146248390482007-11-13T19:29:00.000-08:002007-11-13T19:57:36.526-08:00

The predictions are based on yards per play on rushing and passing plays (sacks and sack yards included), sack rates, third down conversion rates, penalty yards per game, and fumble and interception rates. The models are trained on the 1996-2006 seasons. Linear regression predicts the final score margin (home team points - away team points). Logistic regression estimates the probability that

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-42450465795964646222007-11-13T19:28:00.000-08:002007-11-13T19:54:12.925-08:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The power rankings use logistic regression to predict the probability of winning. The score you see along with the rankings is the expected win total over a 16-game season

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-48156168054860454382007-11-13T19:24:00.000-08:002007-11-13T19:28:46.019-08:00

One of the things I've learned in my football research is that the value of home-field advantage varies pretty widely from year to year, especially with interconference games. Because prediction systems tend to have a big bias towards the home team, this can have a large impact on accuracy. So throughout the season, I'll be keeping track of home team winning percentage and average final score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-10766273677945536932007-11-13T19:22:00.000-08:002007-11-13T19:23:55.010-08:00

Using the projected win probabilities from the logistic regression power rankings, I estimated which games each team would win on the rest of their respective schedules. And then I summed the projected wins with the actual wins so far to get their projected win totals. The teams are sorted by division..nobrtable br { display: none }Projected Final StandingsTeamProjected Win TotalAFC

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-34166831528396365312007-11-13T19:20:00.000-08:002007-11-13T19:22:34.036-08:00

This is the data I use to make my predictions and power rankings. Below are the tables for the raw stats and the values over league average for all of the inputs I use. Rush and pass stats are yards per play. See for yourself where your team's strengths and weaknesses are..nobrtable br { display: none }Rush offense, Unadjusted (League Avg = 4.0347)RankTeamVOLA (%)Actual

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-88977907291503852282007-11-11T22:02:00.000-08:002007-11-13T19:18:45.843-08:00

Tables updated for all games and teams.The following table lists the probabilities that the home team would win each game given the rushing averages, passing averages (includes sack yards lost and sacks as pass attempts), sack rates, third down conversion rates, penalty yards, and interception and fumble rates (by number of completions and rushing plays) for both teams. A logistic regression

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com5tag:blogger.com,1999:blog-881050816100887731.post-69365669422275070522007-11-06T21:58:00.001-08:002007-11-07T00:04:51.204-08:00

The commentary in this article is based largely on the stats found here and the expected wins method found here.After spending over an hour writing it, I realized these picks aren't particularly brave and pretty much follow current win totals. But the picks are based on the stats, and I'm sure half of them are going to turn out wrong as these things are wont to do. In short, though, my surprise

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-14959541733944541442007-11-06T20:04:00.000-08:002007-11-13T19:30:54.172-08:00

Quick Summary of AccuracyLAST WEEKLinear regression, opponent adjusted: 11 out of 14Logistic regression, opponent adjusted: 11 out of 14Linear regression, unadjusted: 11 out of 14Logistic regression, unadjusted: 11 out of 14YEAR SO FARLinear regression, opponent adjusted: 68.3673%Logistic regression, opponent adjusted: 67.3469%Linear regression, unadjusted: 64.2857%Logistic regression, unadjusted

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-29249411764896466462007-11-06T19:56:00.000-08:002007-11-06T20:04:30.583-08:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The power rankings use logistic regression to predict the probability of winning. The score you see along with the rankings is the expected win total over a 16-game season

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-70352270007645508162007-11-06T19:52:00.000-08:002007-11-06T19:56:21.931-08:00

One of the things I've learned in my football research is that the value of home-field advantage varies pretty widely from year to year, especially with interconference games. Because prediction systems tend to have a big bias towards the home team, this can have a large impact on accuracy. So throughout the season, I'll be keeping track of home team winning percentage and average final score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com1tag:blogger.com,1999:blog-881050816100887731.post-67869290085517004202007-11-06T19:51:00.000-08:002007-11-13T19:24:03.186-08:00

Using the projected win probabilities from the logistic regression power rankings, I estimated which games each team would win on the rest of their respective schedules. And then I summed the projected wins with the actual wins so far to get their projected win totals. The teams are sorted by division..nobrtable br { display: none }Projected Final StandingsTeamProjected Win TotalAFC

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-90654817866081867902007-11-06T19:49:00.000-08:002007-11-06T19:51:51.806-08:00

This is the data I use to make my predictions and power rankings. Below are the tables for the raw stats and the values over league average for all of the inputs I use. Rush and pass stats are yards per play. See for yourself where your team's strengths and weaknesses are..nobrtable br { display: none }Rush offense, Unadjusted (League Avg = 4.0894)RankTeamVOLA (%)Actual

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com1tag:blogger.com,1999:blog-881050816100887731.post-83032426622992717482007-11-04T19:15:00.000-08:002007-11-06T19:26:56.474-08:00

Night games added. Tables updated.The following table lists the probabilities that the home team would win each game given the rushing averages, passing averages (includes sack yards lost and sacks as pass attempts), sack rates, third down conversion rates, penalty yards, and interception and fumble rates (by number of completions and rushing plays) for both teams. A logistic regression model

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-44033475430381069942007-10-30T18:27:00.002-07:002007-10-30T18:39:52.446-07:00

The predictions are based on yards per play on rushing and passing plays (sacks and sack yards included), sack rates, third down conversion rates, penalty yards per game, and fumble and interception rates. The models are trained on the 1996-2006 seasons. Linear regression predicts the final score margin (home team points - away team points). Logistic regression estimates the probability that

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com3tag:blogger.com,1999:blog-881050816100887731.post-21632750501994873532007-10-30T18:27:00.001-07:002007-10-30T18:33:03.750-07:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The power rankings use logistic regression to predict the probability of winning. The score you see along with the rankings is the expected win total over a 16-game season

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-26849261974992754332007-10-30T18:27:00.000-07:002007-10-30T18:31:12.879-07:00

One of the things I've learned in my football research is that the value of home-field advantage varies pretty widely from year to year, especially with interconference games. Because prediction systems tend to have a big bias towards the home team, this can have a large impact on accuracy. So throughout the season, I'll be keeping track of home team winning percentage and average final score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-17882950058477838512007-10-30T18:26:00.000-07:002007-10-30T18:29:16.476-07:00

Using the projected win probabilities from the logistic regression power rankings, I estimated which games each team would win on the rest of their respective schedules. And then I summed the projected wins with the actual wins so far to get their projected win totals. The teams are sorted by division..nobrtable br { display: none }Projected Final StandingsTeamProjected Win TotalAFC

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-27125323813064001622007-10-30T18:25:00.000-07:002007-10-30T18:26:33.758-07:00

This is the data I use to make my predictions and power rankings. Below are the tables for the raw stats and the values over league average for all of the inputs I use. Rush and pass stats are yards per play. See for yourself where your team's strengths and weaknesses are..nobrtable br { display: none }Rush offense, Unadjusted (League Avg = 4.0915)RankTeamVOLA (%)Actual

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-27417704006484070662007-10-29T20:47:00.000-07:002007-10-29T23:36:21.531-07:00

(Credit goes to Aaron Hermann of accuscore.com for bringing the central fact of the article to my attention.)As previously discussed here, averages aren't very good for predictions because players and teams perform above and below averages. That's how you get the averages. "We" here at the blog use a system that classifies games as home wins or losses based on the stats from the game, which is

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-77474524844610639092007-10-28T21:25:00.000-07:002007-10-30T18:25:33.245-07:00

The following table lists the probabilities that the home team would win each game given the rushing averages, passing averages (includes sack yards lost and sacks as pass attempts), sack rates, third down conversion rates, penalty yards, and interception and fumble rates (by number of completions and rushing plays) for both teams. A logistic regression model trained on 1996-2006 data computes

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com1tag:blogger.com,1999:blog-881050816100887731.post-76203952069184463252007-10-23T21:36:00.000-07:002007-11-13T21:37:37.365-08:00

.nobrtable br { display: none }Predictions based on opponent-adjusted statsGamePredicted Final Score MarginP(Home Team Wins)NYG @ MIA0.705352.9249WAS @ NE5.573468.5039BUF @ NYJ2.414659.6878PIT @ CIN1.514451.4854OAK @ TEN10.903780.3202GB @ DEN2.987959.8897HOU @ SD5.185865.5187DET @ CHI1.696351.1043PHI @ MIN2.207459.1040IND @ CAR-4.935433.7619JAX @ TB3.514762.9880CLE @ STL-0.242447.8315NO @ SF-

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-72399399851406787072007-10-23T21:00:00.000-07:002007-10-23T19:10:20.557-07:00

Night games, commentary, and expected wins table added.The following table lists the probabilities that the home team would win each game given the rushing averages, passing averages (includes sack yards lost and sacks as pass attempts), sack rates, third down conversion rates, penalty yards, and interception and fumble rates (by number of completions and rushing plays) for both teams. A

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-21453223993638703612007-10-23T19:18:00.000-07:002007-10-23T19:31:29.012-07:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The power rankings use logistic regression to predict the probability of winning. The score you see along with the rankings is the expected win total over a 16-game season

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-27114349911060696152007-10-23T19:16:00.000-07:002007-10-23T19:18:11.166-07:00

One of the things I've learned in my football research is that the value of home-field advantage varies pretty widely from year to year, especially with interconference games. Because prediction systems tend to have a big bias towards the home team, this can have a large impact on accuracy. So throughout the season, I'll be keeping track of home team winning percentage and average final score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-8427469944785723302007-10-23T19:14:00.000-07:002007-10-23T19:16:30.854-07:00

Using the projected win probabilities from the logistic regression power rankings, I estimated which games each team would win on the rest of their respective schedules. And then I summed the projected wins with the actual wins so far to get their projected win totals. The teams are sorted by division. For the record, New England's only projected loss is on the road to Indianapolis. If you

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-74310721324869829552007-10-23T19:12:00.000-07:002007-10-23T19:14:08.925-07:00

This is the data I use to make my predictions and power rankings. Below are the tables for the raw stats and the values over league average for all of the inputs I use. Rush and pass stats are yards per play. See for yourself where your team's strengths and weaknesses are..nobrtable br { display: none }Rush offense, Unadjusted (League Avg = 4.1061)RankTeamVOLA (%)Actual

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-55566900164347626302007-10-16T18:45:00.000-07:002007-10-18T19:53:57.150-07:00

The predictions are based on yards per play on rushing and passing plays (sacks and sack yards included), sack rates, third down conversion rates, penalty yards per game, and fumble and interception rates. The models are trained on the 1996-2006 seasons. Linear regression predicts the final score margin (home team points - away team points). Logistic regression estimates the probability that

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com4tag:blogger.com,1999:blog-881050816100887731.post-35768021169205745202007-10-16T18:35:00.000-07:002007-10-16T18:44:38.341-07:00

Sunday night game and commentary (you'll be real surprised) addedMonday night game added (another surprise) and expected wins table addedThe following table lists the probabilities that the home team would win each game given the rushing averages, passing averages (includes sack yards lost and sacks as pass attempts), sack rates, third down conversion rates, penalty yards, and interception and

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com4tag:blogger.com,1999:blog-881050816100887731.post-11363825778325439242007-10-16T18:32:00.000-07:002007-10-16T18:36:00.506-07:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The power rankings use logistic regression to predict the probability of winning. The score you see along with the rankings is the expected win total over a 16-game season

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-51362108764269526012007-10-16T18:29:00.000-07:002007-10-16T18:32:33.828-07:00

Using the projected win probabilities from the logistic regression power rankings, I estimated which games each team would win on the rest of their respective schedules. And then I summed the projected wins with the actual wins so far to get their projected win totals. The teams are sorted by division. For the record, New England's only projected loss is on the road to Indianapolis. If you

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-14209005495630139712007-10-16T18:28:00.000-07:002007-10-23T19:17:05.850-07:00

One of the things I've learned in my football research is that the value of home-field advantage varies pretty widely from year to year, especially with interconference games. Because prediction systems tend to have a big bias towards the home team, this can have a large impact on accuracy. So throughout the season, I'll be keeping track of home team winning percentage and average final score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-29373210727428713182007-10-16T18:27:00.000-07:002007-10-16T18:28:31.626-07:00

This is the data I use to make my predictions and power rankings. Below are the tables for the raw stats and the values over league average for all of the inputs I use. Rush and pass stats are yards per play. See for yourself where your team's strengths and weaknesses are..nobrtable br { display: none }Rush offense, Unadjusted (League Avg = 4.1160)RankTeamVOLA (%)Actual

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-50539615827188958442007-10-15T19:56:00.000-07:002007-10-15T20:29:42.047-07:00

In 2005, Washington and Tampa Bay were playoff teams. In 2006, they were bad teams in a bad conference. This year, both teams have bounced back and look like strong playoff contenders.In 2005, New Orleans was the second worst team in the league. In 2006, they made it to the NFC Championship game. In 2007, they look to be one of the worst teams in the league again.When predicting how an

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com2tag:blogger.com,1999:blog-881050816100887731.post-72584151443766048012007-10-10T17:44:00.000-07:002007-10-14T21:25:56.330-07:00

Edit: I decided to remove the Hail Mary interception that Favre threw at the end of the Bears-Packers game from the box score. I have posted the resulting changes for this article, but as it does not significantly affect this week's predictions, I am not going back and fixing this week's other articles. Starting with Week 6, however, the stats will reflect the removed play. I have also done

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com1tag:blogger.com,1999:blog-881050816100887731.post-31267942319552100272007-10-09T09:02:00.000-07:002007-10-09T19:18:53.810-07:00

The predictions are based on yards per play on rushing and passing plays (sacks and sack yards included), sack rates, third down conversion rates, penalty yards per game, and fumble and interception rates. The models are trained on the 1996-2006 seasons. Linear regression predicts the final score margin (home team points - away team points). Logistic regression estimates the probability that

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-27403489616224150172007-10-09T09:00:00.000-07:002007-10-09T19:23:44.344-07:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The first power rankings use linear regression to predict the final score margins of each such game. The score you see along with the rankings is the sum of all of those

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-32967818367444172322007-10-09T08:59:00.000-07:002007-10-09T09:09:29.848-07:00

Using the projected win probabilities from the logistic regression power rankings, I estimated which games each team would win on the rest of their respective schedules. And then I summed the projected wins with the actual wins so far to get their projected win totals. The teams are sorted by division.Using opponent-adjusted stats.nobrtable br { display: none }Projected Final

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-72680954826008183412007-10-09T08:58:00.000-07:002007-10-09T09:07:45.278-07:00

One of the things I've learned in my football research is that the value of home-field advantage varies pretty widely from year to year, especially with interconference games. Because prediction systems tend to have a big bias towards the home team, this can have a large impact on accuracy. So throughout the season, I'll be keeping track of home team winning percentage and average final score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com2tag:blogger.com,1999:blog-881050816100887731.post-65923311875972884412007-10-09T08:57:00.000-07:002007-10-09T09:14:01.702-07:00

This is the data I use to make my predictions and power rankings. Below are the tables for the raw stats and the values over league average for all of the inputs I use. Rush and pass stats are yards per play. See for yourself where your team's strengths and weaknesses are. (Football Outsiders says that teams that perform significantly worse on 3rd down than 1st and 2nd will improve the

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-41212262621481517032007-10-02T09:09:00.000-07:002007-10-02T11:36:43.128-07:00

The predictions are based on yards per play on rushing and passing plays (sacks and sack yards included), sack rates, third down conversion rates, penalty yards per game, and fumble and interception rates. The models are trained on the 1996-2006 seasons. Linear regression predicts the final score margin (home team points - away team points). Logistic regression estimates the probability that

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com3tag:blogger.com,1999:blog-881050816100887731.post-41327733581969288682007-10-02T09:07:00.000-07:002007-10-02T09:10:40.669-07:00

The power rankings for this site were developed in this article. The idea behind them is simple: if each team played one home game and one away game against every other team in the league, who would be the most successful?The first power rankings use linear regression to predict the final score margins of each such game. The score you see along with the rankings is the sum of all of those

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com2tag:blogger.com,1999:blog-881050816100887731.post-49590719702720815102007-10-02T09:03:00.000-07:002007-10-04T11:22:34.033-07:00

This is the data I use to make my predictions and power rankings. Below are the tables for the raw stats and the values over league average for all of the inputs I use. Rush and pass stats are yards per play. See for yourself where your team's strengths and weaknesses are. If third down conversion rates correlate with third down DVOA at all, then that might be good news for the 2008 Dolphins

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-16701328450683080532007-10-02T08:57:00.001-07:002007-10-02T09:03:11.149-07:00

Using the projected win probabilities from the logistic regression power rankings, I estimated which games each team would win on the rest of their respective schedules. And then I summed the projected wins with the actual wins so far to get their projected win totals. The teams are sorted by division.Using opponent-adjusted stats.nobrtable br { display: none }Projected Final

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-46823570724656615382007-10-02T08:57:00.000-07:002007-10-02T08:58:55.119-07:00

One of the things I've learned in my football research is that the value of home-field advantage varies pretty widely from year to year, especially with interconference games. Because prediction systems tend to have a big bias towards the home team, this can have a large impact on accuracy. So throughout the season, I'll be keeping track of home team winning percentage and average final score

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0tag:blogger.com,1999:blog-881050816100887731.post-19729191875037386612007-09-30T21:46:00.000-07:002007-10-02T08:57:24.826-07:00

The following table lists the probabilities that the home team would win each game given the rushing averages, passing averages (includes sack yards lost and sacks as pass attempts), sack rates, third down conversion rates, penalty yards, and interception and fumble rates (by number of completions and rushing plays) for both teams. A logistic regression model trained on 1996-2006 data computes

Derekhttp://www.blogger.com/profile/17941314072950152029noreply@blogger.com0