Go to end of page
Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.
This move actually increased the Marlins’ win probability by 1%.
What a ridiculous article, full of the worst tendencies of modern sabermetrics: Blengino thinks that if you have quantified any sort of advantage, that advantage is real and persistent and should be treated as dead-solid correct.
If you think a change in win probability of 1% has any validity at all, you're nuts. This is like deciding one player is better than another based on a 0.1 difference in WAR.
In no other sport is a game strategy repeatedly undertaken that has a measurably negative effect on that team’s chances of winning.
4 - That 4th down play was for just two yards. And your Bayesian analysis doesn't take into account maybe the most important factor - Belicheck knew there was no way his defense was keeping Manning out of the endzone regardless of what yard line Manning started on.
And the correct answer is you never hit on 16, regardless of the situation, if you still want to remain married.
The pt is that this large sample size statistic that 4th/6 is 40% play overlooks that in this particular situation the defense will SELL OUT to stop the run. They have to; it's their only logical play.
No it's not. I'm going to agree with your general point but this is off. The win probability calculations are based on thousands and thousands and thousands of data points. Somewhere around a sample size of 8,000 or so, the standard error (under the binomial assumption, see below) falls below .5%. Given we're talking data compiled over every game for decades, the sample size underlying the estimate is quire large -- maybe not large enough to give us complete confidence in the difference but almost certainly larger than the sample sizes involved in (the vast majority of) player comparisons.
#4 But most of that is just why football, basketball, etc. are much harder to model than baseball. The interaction between offense and defense is critical and that simply isn't the case in baseball. There's very little the defense can do to shut down Stanton
So, yes, casual saber folks simplify this down to the simple yes/no -- which is understandable and is not necessarily a bad strategy when required to make a decision on insufficient information. Which is the situation that McClendon was in
I stand by my original statement. Baseball is far too complicated to be reduced like this. How can you have a sample size of 8000 when every single play, every single situation, is unique?
SOrry I messed up the Belichek thing, it was 4th and 2. But the pt. was the assumption in all the pro Belichek arguments was that the % of making this would remain at >50%, which would only be true if the defense was acting like it always would in those situations. But in this particular case the defense would sell out to stop the run as they had much less reason to stop then the pass then they would in a normal 4th/2.
You must be Registered and Logged In to post comments.
Login to Join (2 members)
Page rendered in 3.0258 seconds, 73 querie(s) executed