With the season approaching, here are some resources I use for research:
Football outsiders: http://www.footballoutsiders.com/: Good for summaries of performance, other statistics. Measures player value based on defense adjusted extra yards. Beware with the DVOA numbers year to year, because they are not really correlated with each other
http://www.ourlads.com/nfldepthcharts/: Good for NFL depth charts, more accurate than ESPN, whos lineups are not updated very often.
https://www.teamrankings.com/nfl/stats/: Good for team statistics, I use them for pace of play. Also has a good section for NCAA football statistics.
https://www.profootballfocus.com/: If you are looking at individual player performance, and to use past statistics as predictors, this is by far the most accurate in my experience. Ratings are relatively stable year to year, and are predictive of results (for example: High Rated QBs have a 60% correlation to higher completion %). I have tested profootball focus ratings, madden football ratings, and the DVOA from Football outsiders. DVOA was relative predictive in-sample, but year to year ratings for players have no correlation. Madden ratings had almost no correlation at all to actual performance data. They charge 30$ per year, but are well worth it in my opinion to building predictive models.
Football outsiders: http://www.footballoutsiders.com/: Good for summaries of performance, other statistics. Measures player value based on defense adjusted extra yards. Beware with the DVOA numbers year to year, because they are not really correlated with each other
http://www.ourlads.com/nfldepthcharts/: Good for NFL depth charts, more accurate than ESPN, whos lineups are not updated very often.
https://www.teamrankings.com/nfl/stats/: Good for team statistics, I use them for pace of play. Also has a good section for NCAA football statistics.
https://www.profootballfocus.com/: If you are looking at individual player performance, and to use past statistics as predictors, this is by far the most accurate in my experience. Ratings are relatively stable year to year, and are predictive of results (for example: High Rated QBs have a 60% correlation to higher completion %). I have tested profootball focus ratings, madden football ratings, and the DVOA from Football outsiders. DVOA was relative predictive in-sample, but year to year ratings for players have no correlation. Madden ratings had almost no correlation at all to actual performance data. They charge 30$ per year, but are well worth it in my opinion to building predictive models.
