tag:blogger.com,1999:blog-38600807.post1540577984517641819..comments2014-04-16T09:45:31.066-04:00Comments on Advanced Football Analytics (formerly Advanced NFL Stats): Expected PointsBrian Burkenoreply@blogger.comBlogger18125tag:blogger.com,1999:blog-38600807.post-14209809843211887832013-09-30T17:43:47.931-04:002013-09-30T17:43:47.931-04:00Very interesting stuff. Do you have the data to br...Very interesting stuff. Do you have the data to break down expected points based on the down as well? Specifically, I'd love to know what's worth more: 1st and Goal from the 10 or 1st and 10 from the 11, 12, 13, etc. I have a suspicion that 1st and Goal from the 10 is worth fewer points, but it's obviously difficult to know for sure.gosieghttp://www.blogger.com/profile/10656799789460205512noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-49134833390502993482012-03-15T20:09:12.335-04:002012-03-15T20:09:12.335-04:00I'm down with the idea of expected points, but...I'm down with the idea of expected points, but I'm curious about using them to evaluate coaches' choices in game situations, especially 4th and short.<br /><br />Let's say I have a game that gives you an expected value of 10 times the money you bet (e.g., flip a coin; heads win 20x your bet, tails lose it all). Most people would be willing to bet $1 or $10, but most would not be willing to bet $1000, even though that gives them expected winnings of $10,000, which is 100x more than their $10 bet. The utility of winning $10,000 is influenced by the possibility of losing the $1000 bet. It would be interesting to do some modeling to see if you can determine the Utility of the Expected Value for NFL coaches. It's often modeled as f(x) = x^(1/2) or f(x) = ln(1+x).cassio598http://www.blogger.com/profile/13879621378274508671noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-15662953752704676252012-03-03T16:09:59.995-05:002012-03-03T16:09:59.995-05:00I like Chris's idea. I hope Brian can get arou...I like Chris's idea. I hope Brian can get around to it someday.Ben Moorehttp://www.blogger.com/profile/09665895222029834484noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-74150666552130866672011-12-06T12:01:34.230-05:002011-12-06T12:01:34.230-05:00Is there an analysis that breaks down the probabil...Is there an analysis that breaks down the probabilities of the next score being a TD, FG, (safety too for that matter) for both the offense and defense based on field position?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-2739040130428186362011-10-25T21:26:43.728-04:002011-10-25T21:26:43.728-04:00After watching the Ravens fiasco last night I'...After watching the Ravens fiasco last night I'm curious whether you've ever been able to have a look at the EP impact of various penalties? Sorry if it's posted already and I just missed it.<br /><br />Seems to me that most teams (and certainly announcers) don't have a good handle on just how costly some of these penalties are. <br /><br />Holding on 1st and 10 seems innocent enough but is really quite costly. (0.8 pts if I'm using your calculator correctly) Even worse if it happens on 3rd and 1.. Running into the kicker on 4th and 3 looks like couple points easily.<br /><br />Is there a hierarchy of horrible penalties?Chrisnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-76116055396204357632010-04-30T12:30:45.511-04:002010-04-30T12:30:45.511-04:00Is there anyway you could post exact numbers for t...Is there anyway you could post exact numbers for the expected point graph? It is hard to get accurate number just looking at the graph.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-5705354944719629982010-02-25T03:41:23.510-05:002010-02-25T03:41:23.510-05:00Brian, how does forcing a punt (via 3 and out), le...Brian, how does forcing a punt (via 3 and out), lets say at an offense's 25yd line, result in "better field position" for the other team's offense? I know it gets the ball back, however, how is it necessarily in better field position for the other team's offense? How about comparing that situation to lets say holding an offense at the 50 yr line by forcing a 3 and out. Does the other team's offense get better or worse field position? <br /><br />In the first situation the punt could result in a kick lands on the opposite 20yr line (no runback) and in the second situation, the punter hits the touchback and so its on the 20 also.<br /><br />Basically what I am trying to ask is forcing 3 and outs helps get the ball back, but does it necessarily get the ball back in a "good field" position? The next question should probably be does it matter where a defense forces a 3 and out?<br /><br />Also, why is the value of a TO worth more on 1st down compare to 2nd and 3rd down?<br /><br />Thanks Brian,<br /><br />JohnAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-7201967386705666472009-10-23T02:00:16.381-04:002009-10-23T02:00:16.381-04:00One of the reasons I love going for it on fourth d...One of the reasons I love going for it on fourth down from the one or two is partially explained by the above chart. You have to subtract the negative expectation from the 3 points. For example, if a possession 99 yards from the endzone has an expectation value of -1 points, then a field goal from the 1 is only worth 2 points. This is because if you just handed the ball over you have an expectation of 1 point (the other team's expectation is -1).<br /><br />This means that instead of the conventional wisdom, which might be that you need a 3/7 (43%) or better conversion rate to break even, you actually only need a 29% (2/7) conversion rate.<br /><br />This is an oversimplification, of course. It neglects the emotional aspect of the boost a goal line stand might give your opponent. But, particularly for teams with good defenses, going for it on 4th and goal from the 1 or 2 almost always makes sense.Happyhttp://www.bestofblog.net/nfl_picks_2009/noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-79758232130734290212009-09-21T16:56:28.008-04:002009-09-21T16:56:28.008-04:00Justin-Very interesting. I never considered the pr...Justin-Very interesting. I never considered the prospect of uneven numbers of possessions.<br /><br />This is kind of an old post, so I don't recall if the graph above makes the kick-off adjustment. I doubt it. The more recent '4th Down Study' post I just put up does feature a graph that definitely does has the adjustment.Brian Burkehttp://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-82971902112337154472009-09-21T16:39:13.285-04:002009-09-21T16:39:13.285-04:00I think instead of subtracting a point, you should...I think instead of subtracting a point, you should subtract half a point. No matter if you score or not, you're returning the ball at the end of your possession.<br /><br />Upon kicking off you know that from this moment forward each team will have either the same number of possession, or the opposing team will have one more. Thus when receiving a kick-off you should add 0.5 pts to the receiving team and subtract 0.5 pts from the kicking team.<br /><br />Whenever analyzing a point expectancy for a team, you should subtract 0.5 for the ensuing return of possession. The exception is if you have reason to believe that it's not 50/50 who will have the final possession (late in a game, for instance).<br /><br />Brian, do these point expectancy make any account for the expected field position of the opponent when the ball is returned? It seems like the answer is yes based on the 15 yd line being break-even, but it's hard to tell.Justinnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-33180703175297706982009-09-14T23:13:32.748-04:002009-09-14T23:13:32.748-04:00Also, the yard line of indifference is the 15. Tha...Also, the yard line of indifference is the 15. That's where the black line in the graph above intersects with zero points. A team would be indifferent to having the ball at their own 15 or their opponent having the ball at the opponent's own 15. There is no yard line where a team would be indifferent to a turnover.Brian Burkehttp://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-6768892885449333522009-09-14T23:09:40.864-04:002009-09-14T23:09:40.864-04:00Yes, it would cancel out. But keep in mind your co...Yes, it would cancel out. But keep in mind your condition: "and he scores." That's not a given. You might be the next to score (scoring twice in a row), so you can't assume alternating scores. Therefore, you have to account for the kickoff.<br /><br />Another way to think about it is if you consider a safety. A safety is 2 pts, but the scoring team gets the ball back and does not kick off. That obviously has positive value, and therefore kicking off to the opponent after a TD or FG has a similar negative value.Brian Burkehttp://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-18640372821089446152009-09-14T22:01:35.397-04:002009-09-14T22:01:35.397-04:00I am the guy who asked the question above on Janua...I am the guy who asked the question above on January 12 2009. I still don't quite understand. If you score and give possession to your Opponent (with a 1 pt expectancy) and he scores and gives you possession and the 1 pt expectancy back to you. Doesn't this all cancel out? So Romer would not be correct because any point expectancy you give to your opponent will be handed back to you.<br />ThxAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-4907040511990433342009-09-08T17:57:27.903-04:002009-09-08T17:57:27.903-04:00meeeeee,
the expected points for your team is 0 if...meeeeee,<br />the expected points for your team is 0 if you possess the ball at your own 15 yard line. however, you are not indifferent between your team and the opponent having the ball at your 15 yard line for their expected points would be approximately 4.5 points. This would clearly be a disadvantage for you. I think what Brian means is that you are indifferent between possessing the ball at your own 15 or the opponent possessing the ball at their own 15 because the expected points are equal at 0.<br /><br />Brian,<br />I like your work. You explain economic concepts very clearly. Thanks for making a cool site.Joe 25noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-11253638701166901032009-02-16T07:57:00.000-05:002009-02-16T07:57:00.000-05:00Brian,Nice work with this site. Clearly the expect...Brian,<BR/><BR/>Nice work with this site. <BR/><BR/>Clearly the expected point value distribution above would only apply to a 'pick-em' game with an average expected number of total points (say 43). Have you tried working on a model that takes into account the driving team's % probability of winning the game, and the expected number of points in the game, as additional factors in determining the expected points value of a drive?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-556765808556062932009-01-12T12:54:00.000-05:002009-01-12T12:54:00.000-05:00The chart above is a little different than Romer's...The chart above is a little different than Romer's. It should be about 0.7 points lower to account for the kickoff to the other team.<BR/><BR/>The line of indifference should be about the 15.Brian Burkehttp://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-28645213301555083192009-01-12T12:45:00.000-05:002009-01-12T12:45:00.000-05:00Just found the site and I love it. Question on the...Just found the site and I love it. Question on the above article - Romer says that a score is actually worth 1 point less than you think because the Opponent will take possession afterwards. Good point, but if the Opponent takes possession and scores then they will give possession back to the Original Team and if the Original Team scores they will give it back to the Opponent and so on and so on. My question is it you take the ball on the 70yd line and have a 1 pt expectancy(chart above), is this negated by the fact that you will be giving your opponent possession after you score? Or to ask it another way, At what yard line am I indifferent possesing the ball or having my opponent possess the ball?meeeeeeeeehttp://www.blogger.com/profile/12076636246070084911noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-73973505417855495522008-09-10T01:53:00.000-04:002008-09-10T01:53:00.000-04:00Have you ever considered trying to work other fact...Have you ever considered trying to work other factors other than field position in, such as down and distance, timeouts, time on clock?<BR/><BR/>Emperically, it would cut the data to thing, but we have set up our NES models to incorporate regressions to help value the relative effect (and 2nd order effects<BR/><BR/>-Mark mkamal@protrade.comMark Kamalnoreply@blogger.com