clock menu more-arrow no yes

Filed under:

Clutchiness

Our very own Dan is becoming quite a hit with the stats crowd.  Since the advent of his website Clutchiness he has been featured on such sites as The Hard Ball Times.

What is exactly Clutchiness?  It turns out that isn't an easy thing to describe though it is an easy concept to grasp.  I was looking at his website on Saturday trying to find a simple way to present his project and frankly, I couldn't think of a way.  So I emailed Dan and asked how he would describe it and it turns out that Dan wasn't much better at parsing it down than I was.

But here are some excerpts from his email:

Now, based on linear weights, which evaluate all forms of production in terms of their contribution to run production, we can estimate how many wins a player with X OBP and Y SLG would contribute.

The difference between what a player has actually contributed, as measured by WPA, and what we would expect from his OBP and SLG, is entirely the product of his performance in higher leverage (more important) situations.  This is clutchiness.

---

... but it's hard to keep short since it's based on three relatively unknown concepts (leverage, win probability, linear weights).  I suppose to get really simple, one might say that clutchiness is the difference in opinion that someone who only watches games live and someone who only reads box scores would have on each player.  One can only go by the feelings and emotions that players cause (a fantastic description of WPA is that it's a numerical representation of the feelings you have watching a game); the other only has cumulative numbers that exist only in black and white.

Dan, if my cutting and pasting has completely misrepresented your project, please let me know and I will adjust accordingly.

What it comes down to is, the timing of every hit in a game doesn't carry equal weight given what is expected from the player.  A two out single in the second with no one on and the batter being stranded at first isn't as important as a single with the bases loaded. This is probably too simple of an explanation.

Dan goes to great lengths to let his readers know that his numbers only tell what has happened in the past and are not a predictor of the future.  While this is true, it is not a predictor but it has the potential to be an indicator.  Example: the best indicator of whether you will go to happy hour this Friday is if you went to happy hour last Friday.

Anyway, Dan has himself a hit and I recommend checking it out.