tag:blogger.com,1999:blog-38600807.post4701056540397057512..comments2018-06-02T14:19:34.554-04:00Comments on Advanced Football Analytics (formerly Advanced NFL Stats): The Ellsberg Paradox and 4th DownUnknownnoreply@blogger.comBlogger8125tag:blogger.com,1999:blog-38600807.post-89391184294627258972014-03-04T07:01:35.969-05:002014-03-04T07:01:35.969-05:00I have to admit that i didnt look too much into ho...I have to admit that i didnt look too much into how these 4th down probabilities are actually calculated. i just can assume that they track all 4th down attempts and log down & distance and whether or not it was successful.<br /><br />so the numbers are skewed. for example, there was an article where it said that the chargers should have tried more 4th down conversions in wk17 and in the WC weekend. <br /><br />the example was a 4th-and-3 or so deep in your own territory. now, the chances to convert this arent that bad, when you look at it statistically. this is obvious because in which situations are teams going for it on 4th down in their own half? its late in the game, probably even under 2 minutes, one/no timeout left. the defense obviously would like to stop them, but is concerned with the game-winning long play that puts them in FG range/gives them the TD as well.<br /><br />just think about it. 4th-and-3 at your own 15, 1:20 left, FG wins it. lets say the offense brings 5 WRs in an empty set, what do you do as a defense? cover-0 and try to stop them right here but if one guy gets beat, the game is over? no, you probably sit back, cover deep and the sidelines first and maybe you can make a play on the draw/short pass.<br /><br />the chances of going for it on 4th-and-3 in the first quarter on your own 15 and the implications for the likelihood to win the game are a totally different animal. you cannot compare the two. this is why i think that coaches are generally right to punt as often as they do, since the statistics from converted 4th down attempts, especially in your own half, come from situations which arent transferrable to every single drive.Anonymoushttps://www.blogger.com/profile/07119263636491023531noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-80508224784424300502009-04-07T12:12:00.000-04:002009-04-07T12:12:00.000-04:00I meant to say there are only two colors when refe...I meant to say there are only two colors when referring to there being two ballsRyannoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-8890486093936574882009-04-07T12:11:00.000-04:002009-04-07T12:11:00.000-04:00Jack I think when he says "random" when referring ...Jack I think when he says "random" when referring to the random jar, he's talking about a random selection of 100 blue and red balls that were put into a jar. Because there are two balls, there is a 50/50 chance that each ball selected and put into the random jar will be red. In theory, if this random process is carried out 100 times with a 50/50 probability, there should be 50 red balls and 50 blue balls, or at least a close ratio in the random jar. I think that is the assumption that is being made here.Ryannoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-32716640022377581192009-01-08T16:46:00.000-05:002009-01-08T16:46:00.000-05:00The problem with your analysis is the unspecified ...The problem with your analysis is the unspecified assumption that both jars yield a 50/50 chance of picking a red bal. Your contention is "There is a 50/50 chance of winning $100 from either jar." But you don't state that the random jar has a 50/50 mix...which is critical to your analysis. If the random jar has 90 blue balls and 10 red balls, then the probability of chosing a red ball isn't 50/50 and obviously the non-random jar is your best bet. Without knowing the exact ratio of red/blue in the random jar, it is always "more reasonable" to choose the jar where the ratio is known...meaning no paradox. This is an age old problem in mathematical analysis...you must state your givens very precisely.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-60336435939240459592008-05-09T13:23:00.000-04:002008-05-09T13:23:00.000-04:00Canadian teams only have a 20-second play clock as...Canadian teams only have a 20-second play clock as well (vs. 40 for the NFL). They run more plays-per-game.JTapphttps://www.blogger.com/profile/17812390329612009756noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-13191448694749249722008-05-08T16:40:00.000-04:002008-05-08T16:40:00.000-04:00I would agree. A team with a 4-down mentality woul...I would agree. A team with a 4-down mentality would turn 3rd down into a 2nd down. There would be far more runs. <BR/><BR/>The CFL is a 3-down only league. From my vague memory of the Baltimore Stallions, the standard series was pass-pass-punt, with an occasional draw play sprinkled in to keep defenses constrained. It's as if they started with a 2nd and 10 on every series.<BR/><BR/>The fewer the downs, the more teams will tilt towards the pass. And the more downs allowed, the more attractive running becomes.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-52451754216689424052008-05-08T16:28:00.000-04:002008-05-08T16:28:00.000-04:00Also, looking at your previous work (The Passing P...Also, looking at your previous work (The Passing Paradox) it stands to reason that teams that go for it regularly on 4th down should be more risk-averse on 3rd down. This should greatly affect the team's playcalling mix.JTapphttps://www.blogger.com/profile/17812390329612009756noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-49061248570593755602008-05-05T12:56:00.000-04:002008-05-05T12:56:00.000-04:00Awesome analysis, I think you're on the money with...Awesome analysis, I think you're on the money with the maximin. Don Shula supposedly said that one day there will come a coach who just won't punt. We'll be waiting forever. <BR/><BR/>You could really publish a great follow-up to Romer on this.JTapphttps://www.blogger.com/profile/17812390329612009756noreply@blogger.com