naked cerebellum

History

Member for

3 years 34 weeks

Karma

Current value

Recent Comments

Date	Title	Body
3 years 1 week ago	This is great stuff Brian,	This is great stuff Brian, and you get an extra gold star because it came on a Friday afternoon. As for Gladwell, as annoyingly simplistic as he is at times, I must admit that he is thought-provoking, not in any profoundly original way, but in a manner that will make you to want to revisit and reanalyze a well-worn topic with more nuance and sophistication.
3 years 34 weeks ago	For the lack of subtlety, jeesh.	You want an example?  How about SuperBowl XXXVIII, when Carolina missed two 2-point conversions in the fourth quarter, the first of which was attempted w/ 12:39 left in the game.  The final score was 32-29.  Had Carolina kicked both extra points, they could have gone for two on a third 4Q TD they scored with one-minute left in the game.  I'm not sure why you are so adamant that I provide an example, because one example does not prove my point per se, not to mention that the example may not be analagous to the Michigan game.  Only a study using a decent sample size would be useful.  My question was not proferred as proof, but as anecdotal consideration.  Likewise, Brian's a priori assumptions of probabilities should not necessarily be taken at face value.  And that is my point: not that the 2-point attempt was a bad idea; rather, people's conceptions of likely final scores with a lot of time left in the game are in many cases flawed, and thus have a major effect on strategy.  In other words, it's a NOT no-brainer to me (even though I would have gone for two in the same situation.)
3 years 34 weeks ago	My problem with the 2-point	My problem with the 2-point conversion chart is that the statisticians who developed it should have included several pages of footnotes--not that anyone would bother to read them.  To arrive at those numbers, they are using estimates for a large number of variables, whose reliability is highly sensitive to time.  The smaller the score discrepancy and fewer the minutes left in the game, the more accurate those estimates are; the converse is also true.  To provide concrete figures for 30 minutes left in a game for example are silly.   No statistician in good faith should offer a definite figure (which creates the illusion of precision) without a glaring caveat emptor.   As an analogy, predicting the temperature and precipitation over the next 48 hours is fairly reliable; predicting an exact temperature (and not just a range) for a day a year from now is much less so.
3 years 34 weeks ago	Princess Bride & Inconceivable	Methinks you do not undertand the meaning of "no brainer".
3 years 34 weeks ago	Not so fast my buddy guy friend	Ask yourself when the last time was in college or the pros that a successful 2-pt conversion at the start of the 4th quarter or earlier resulted in a win based on that conversion.  Off the top of my head, I can't recall even one time.  On the other hand, it is illustrative that on any football weekend, one can point to a least one game where an early failure (meaning before the last 5 minutes of a game) of a 2-pt conversion resulted in a tie in regulation, loss or needlessly nerve-wracking finale. A really solid argument would incorporate a statistical analysis using a large number of games.  But who wants to do that?  My point is not that Michigan should not have gone for 2 points.  My point is that it is NOT a no-brainer.   Even teams whose offenses are held in check for most of the game often manage to move the ball late in the 4th quarter when the games are close.  Whether this is the result of prevent defenses or more aggressive, risk-taking strategies by the offenses is moot.  The fact that it happens on a regular basis changes the probabilities of the scenarios you listed above.  It may make an "obvious" 2-pt conversion with 15 minutes to go in the game much less certain.