tag:blogger.com,1999:blog-38600807.post7279158903251141648..comments2018-06-02T14:19:34.554-04:00Comments on Advanced Football Analytics (formerly Advanced NFL Stats): Team Efficiency Rankings: Week 4Unknownnoreply@blogger.comBlogger15125tag:blogger.com,1999:blog-38600807.post-51926395504725176882013-10-05T09:29:42.964-04:002013-10-05T09:29:42.964-04:00As far as GWP I've come up with a way to get a...As far as GWP I've come up with a way to get a point spread using it.<br /><br />The big game this week would have Denver -4.5 over Dallas.<br /><br />The model is strongly suggesting the Cowboys cover the inflated line.<br /><br />That is assuming that a larger difference in WP would equate to a larger margin of victory in the game and not just a larger chance to win the game at ant margin of victory.mitchnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-28461753594504095712013-10-05T09:23:45.832-04:002013-10-05T09:23:45.832-04:00Trying to find sneaky teams like Gaints or Ravens ...Trying to find sneaky teams like Gaints or Ravens to win the SB in week 4 or 5 is not possible.<br /><br />Those teams won't become visible untill in the playoffs , not before.<br /><br />Don't you remember last season ? Prior to the start of the playoffs Brain wrote an article on the Ravens, go back and read it.<br /><br />He said the model was correct in regards to the Ravens not being as good as their 9-2 record and the Ravens 1-4 finish proved the worth of the model.<br /><br />The Ravens were the largest WP underdog against Denver of any playoff game.<br /><br />The model could not find that sneaky SB winner after 16 games, how will it do it after 4 or 5, even stating that the sneaky SB winner was not as good as it's record said it was.<br /><br />Your headed down the wrong path my friend.mitchnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-73299640912861055822013-10-05T06:00:49.762-04:002013-10-05T06:00:49.762-04:00The Pats are only a hair better than the Giants? ...The Pats are only a hair better than the Giants? Really? C'mon, this is wild, wacky stuff. I'm a believer in data, *crunched correctly*, being superiour to CW; but I think this model suffers from some weird glitches (I recall feeling that way last year too, though I can't remember specifics).SlackerInchttps://www.blogger.com/profile/08275358994906136088noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-9039365739769893282013-10-05T05:47:45.712-04:002013-10-05T05:47:45.712-04:00This comment has been removed by the author.SlackerInchttps://www.blogger.com/profile/08275358994906136088noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-12108753966966597242013-10-03T22:02:30.529-04:002013-10-03T22:02:30.529-04:00Mate is correct. Good teams will tend to have weak...Mate is correct. Good teams will tend to have weaker schedules because they can't play themselves. And vice versa.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-76379430480398290012013-10-03T10:25:26.470-04:002013-10-03T10:25:26.470-04:00Nate, you can check Brian's Thursday articles ...Nate, you can check Brian's Thursday articles for the game probabilities, but the equation is (H * (1 - A) * HF) / (H * (1 - A) * HF + (1 - H) * A * (1 - HF)) where H is the home team's GWP, A is the away team's GWP, and HF is the home field advantage factor (0.57 for football). If you want to know the away team's chances it's just 1 - (above equation). That might be hard to read/understand, but it is called the Log5 formula and you can google it for more information.<br /><br />And if you want to use these GWP to forecast the rest of the season, NFL-forecast.com is exactly the tool you want!Jameshttps://www.blogger.com/profile/01838293735141324662noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-53369574382509404722013-10-02T23:59:35.241-04:002013-10-02T23:59:35.241-04:00NFL-Forecast.com is now updated and up and running...NFL-Forecast.com is now updated and up and running for 2013Chrishttp://nfl-forecast.comnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-67216637249176742692013-10-02T18:15:45.711-04:002013-10-02T18:15:45.711-04:00Are your raw numbers accurate? I have much dif. OP...Are your raw numbers accurate? I have much dif. OPass * D Pass numbers:<br /><br /><br />from pro reference: I also ran my own from your formula : What have I missed?<br /><br />Mike<br />1 ARI 2013 4 2 2 0 0.500 5.5521<br />2 ATL 2013 4 1 3 0 0.250 6.6778<br />3 BAL 2013 4 2 2 0 0.500 5.1768<br />4 BUF 2013 4 2 2 0 0.500 5.1631<br />5 CAR 2013 3 1 2 0 0.333 4.6146<br />6 CHI 2013 4 3 1 0 0.750 5.9481<br />7 CIN 2013 4 2 2 0 0.500 6.1097<br />8 CLE 2013 4 2 2 0 0.500 4.2950<br />9 DAL 2013 4 2 2 0 0.500 5.5342<br />10 DEN 2013 4 4 0 0 1.000 8.7195<br />11 DET 2013 4 3 1 0 0.750 7.6101<br />12 GNB 2013 3 1 2 0 0.333 6.8409<br />13 HOU 2013 4 2 2 0 0.500 5.3989<br />14 IND 2013 4 3 1 0 0.750 5.9504<br />15 JAX 2013 4 0 4 0 0.000 3.5210<br />16 KAN 2013 4 4 0 0 1.000 5.4459<br />17 MIA 2013 4 3 1 0 0.750 5.3125<br />18 MIN 2013 4 1 3 0 0.250 6.2574<br />19 NOR 2013 4 4 0 0 1.000 7.1461<br />20 NWE 2013 4 4 0 0 1.000 5.6000<br />21 NYG 2013 4 0 4 0 0.000 5.7278<br />22 NYJ 2013 4 2 2 0 0.500 5.6689<br />23 OAK 2013 4 1 3 0 0.250 5.5231<br />24 PHI 2013 4 1 3 0 0.250 7.0511<br />25 PIT 2013 4 0 4 0 0.000 5.8475<br />26 SDG 2013 4 2 2 0 0.500 7.6419<br />27 SEA 2013 4 4 0 0 1.000 6.3932<br />28 SFO 2013 4 2 2 0 0.500 5.8740<br />29 STL 2013 4 1 3 0 0.250 4.8608<br />30 TAM 2013 4 0 4 0 0.000 4.3151<br />31 TEN 2013 4 3 1 0 0.750 5.5859<br />32 WAS 2013 4 1 3 0 0.250 6.0791Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-78947336153027653862013-10-02T14:39:43.376-04:002013-10-02T14:39:43.376-04:00It's not clear to me that the effect you descr...It's not clear to me that the effect you describe is amplified by playing division opponents twice. It's certainly true for W-L records, since a team like Denver could hand each opponent in its division 2 losses, but it's not clear to me that it applies to GWP. Assuming GWP is actually an objective measure of a teams strength I don't see why Peyton Manning moving from Indy to Denver would suddenly make the Raiders worse. At any rate even if I'm making a mental blunder here you could remove division games from the equation, although it would presumably be slightly more time consuming than simply grabbing the numbers from the Week 16 columns.<br /><br />Alternatively if you wanted to get really fancy you could correct for the above described affect. Simulate an artificial NFL by assigning 32 "teams" GWP's randomly in a manner that matches the distribution of GWP's of teams in the real NFL, then run them through a schedule to create Opp GWP's for them. Since the GWP's aren't generated by the model, any SoS effects won't affect them. Do this a few hundred times to get the correlation due to the "31 other teams are not the Broncos effect" down and then you can subtract it out from the actual data. Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-38902790985560466422013-10-02T13:52:24.487-04:002013-10-02T13:52:24.487-04:00Andrew,
There should actually be a negative corre...Andrew,<br /><br />There should actually be a negative correlation. The average team that's not Denver is weaker than the average team that's not Jacksonville, so you'd expect Denver's SoS to be slightly lower than Jacksonville's.<br /><br />The effect is small with 32 teams, but it's amplified by the fact that each team plays its division opponents twice.Natenoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-18762319394757486972013-10-02T13:45:57.854-04:002013-10-02T13:45:57.854-04:00What's the formula to calculate win probabilit...What's the formula to calculate win probability from the two GWP's?Natenoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-50303413140642988152013-10-02T13:28:35.966-04:002013-10-02T13:28:35.966-04:00Hey Brian, have you ever checked to see if there&#...Hey Brian, have you ever checked to see if there's any correlation between a teams GWP and their Opp GWP? Since I can't imagine any reason the quality of a team should, in general, depend on the quality of their opponents if it turns out there's no correlation that might be a neat little proof that your model is handeling strength of schedule correctly.<br /><br />On the other hand if there's a positive correlation, then maybe your model is somehow subtly overestimating the effect of SoS, while if there's a negative correlation perhaps it's underestimating it.Andrewnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-13681055604177003602013-10-02T09:54:22.189-04:002013-10-02T09:54:22.189-04:00The Broncos can't be 0% on offensive fumbling;...The Broncos can't be 0% on offensive fumbling; they've already fumbled 8 times and lost 5 of them. Last week, you had them at 4.1%dougalanleehttps://www.blogger.com/profile/12217255068523591534noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-70356226195276424542013-10-02T09:38:01.867-04:002013-10-02T09:38:01.867-04:00Interesting to see the eagles up so high, but I gu...Interesting to see the eagles up so high, but I guess they lost to good teams and didn't get any picks thrown their way. Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-69136375256421592652013-10-02T08:36:45.064-04:002013-10-02T08:36:45.064-04:00Anybody know what happened to NFL-forecast.com? It...Anybody know what happened to NFL-forecast.com? It had a simulator that determined playoff odds using these GWP and the NFL schedule, but it hasn't updated for this season.Jameshttps://www.blogger.com/profile/01838293735141324662noreply@blogger.com