tag:blogger.com,1999:blog-38600807.post6778914424591400017..comments2023-11-05T04:16:44.937-05:00Comments on Advanced Football Analytics (formerly Advanced NFL Stats): Sunday's Numbers Have Been CrunchedUnknownnoreply@blogger.comBlogger9125tag:blogger.com,1999:blog-38600807.post-54960009034506094002013-10-04T22:49:53.099-04:002013-10-04T22:49:53.099-04:00Brian - that's awesome you found the problem, ...Brian - that's awesome you found the problem, that makes sense. And yes, I am using the ANS EP model, as much as possible anyway. Here's how I got my numbers: I downloaded the data from the ANS site, but noticed that there were some holes for weird down-and-distance situations like 1st-and-11 on the 12-yard line, etc. But, when I plugged those situations into the WP Calculator, I do get a result. So, I plugged in just about every situation that wasn't in the spreadsheet, and interpolated any others that I didn't get a result for (if I recall, there were some situations that didn't work on the WP Calc, not sure if that's the case now).<br />Long story short, I am using your numbers, but I can't say for sure that every single down-and-distance is the same. I would estimate that at least 95% are though.<br />It's unreasonable to expect that "my" numbers would exactly equal those on ANS, but it's also frustrating when some totals are dead on, and others are 20 points off. The error is most likely mine, but I would love to see at least one example of how you calc a game using play-by-play data and your EP values. That way I could at least see if the methodology is correct, and move on to correctly blunders and bad values, etc.<br />If you're wondering why I'm even bothering, I'm doing some analysis regarding penalties and special teams, and these EPA totals are not shown on your site, so I'm calc'ing them on my own.<br />Sorry for the long post, thanks!Anonymoushttps://www.blogger.com/profile/03108711283276399018noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-6190593792456072112013-10-04T20:10:05.678-04:002013-10-04T20:10:05.678-04:00Thanks, Tom. I found the problem. My code looks fo...Thanks, Tom. I found the problem. My code looks for a change in possession on the following play to ensure that it's a defensive TD. But when there was a penalty on the offense, or a penalty between plays, or some other reason for another line of data attributed to the offense, it assumed the TD was for the offense. <br /><br />Are you using the ANS EP model for your calculations?Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-10122640166369512162013-10-04T14:14:32.721-04:002013-10-04T14:14:32.721-04:00Brian - don't mean to go on about this, but I ...Brian - don't mean to go on about this, but I just want to follow up with the EPA issues. I haven't checked all my numbers, but there does indeed seem to be an issue with interceptions run back for a TD. Your team stats show San Diego with a Off EPA of 56.9, I've got 41.4. Sure enough, Rivers has thrown two pick sixes, the first in Week 1 against Houston, -6.50 EPA, and the second against Dallas in Week 4, -8.50. The total is 15.0 which would make our numbers coincide. There are a number of other teams that may have this issue, I just haven't looked into all of them. Again, I don't mean to be a pill, just trying to be a "concerned" member of the community--if I'm wrong, please call me out on it! Thanks.Anonymoushttps://www.blogger.com/profile/03108711283276399018noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-76768788820398211312013-10-01T20:02:07.644-04:002013-10-01T20:02:07.644-04:00Thomas-You nailed it. Corrections have been made i...Thomas-You nailed it. Corrections have been made in the main database, but it might be a bit before I fix it in the boxscore. Found a few other similar errors. Thanks, man.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-51465275232746773222013-10-01T17:43:48.998-04:002013-10-01T17:43:48.998-04:00@Brian - that might be it...he threw a pick six fo...@Brian - that might be it...he threw a pick six for -8.20 EPA. By my rough calc's, his EPA without counting the picks is around +25, counting the picks, it's +9. Anyway, thanks.Anonymoushttps://www.blogger.com/profile/03108711283276399018noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-29732063763936898122013-10-01T10:06:26.915-04:002013-10-01T10:06:26.915-04:00Hmmmm. Where are my Broncos, can't find my Br...Hmmmm. Where are my Broncos, can't find my Broncos, must be a mistake. Look in the "Good Offense" section. Good offense bad defense? Oh snap! They are off the charts for good offense. <br /><br />I Literalol'd when I saw how far to the right they were for "Good Offense" on that chart. I think their Defense isn't "Bad" though. Bad in comparison to what? Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-85485304479535549622013-10-01T07:03:13.170-04:002013-10-01T07:03:13.170-04:00Thomas--You might be right. Occasionally, the scri...Thomas--You might be right. Occasionally, the script that computes EPA mistakes a defense return for a TD as an offensive TD. I'll look into Eli's numbers.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-90790558581752642232013-10-01T00:45:30.413-04:002013-10-01T00:45:30.413-04:00Been looking at the "crunched" numbers.....Been looking at the "crunched" numbers...something seems off with the Eli Manning's numbers. With 9 picks, a 46.4 SR and 3.9 AYPA, I would have thought his EPA would have been lower than +5.1, maybe even in the negative. <br />I believe the problem is the first game of the season against Dallas. In the box score, Manning is shown to have an EPA of +21.6 which seems impossible considering he threw three picks for a total loss of -16.0 EPA.<br />Just curious, thanks guys! Anonymoushttps://www.blogger.com/profile/03108711283276399018noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-74220064336525030122013-09-30T21:58:37.170-04:002013-09-30T21:58:37.170-04:00Brian -
I'd love to see a sample play-by-play...Brian -<br /><br />I'd love to see a sample play-by-play layout of an entire game with EPA and WPA values so that we all could see how you arrive at your totals for offense, defense, etc. They do a similar thing at Pro Football Reference where they show the EPA tally as part of their play-by-play layout. It wouldn't be needed for every game, but I think a lot of us stat geeks would love to an example. Thanks!Anonymoushttps://www.blogger.com/profile/03108711283276399018noreply@blogger.com