The most useless baseball statistic has to be win expectancy a.k.a. win probability.
This pops up in the box score for ongoing baseball games as a historical measure of how likely a team is to win at a given point in the game, but is spectacularly meaningless.
For instance, as of a couple minutes ago the Red Sox led the Rays 3-0 in the second inning, and, wow, had an 82.7% win probability!
Guess I can go to bed happy, knowing that there’s only a negligible 17.3% chance that the Red Sox starter/bullpen will implode. I don’t know why they even bother playing the rest of the game.
I have, many times; it depends on how close the fielder is, more or less. The rule isn’t applied with any sort of consistency, though, which is one more reason why errors and unearned runs are stupid statistics.
This is spectacularly silly, especially when combined with the rest of your post.
The fact that you can create an expectancy or probability of a win based on this sort of calculation does not mean, and was never intended to mean, that the game is a foregone conclusion, or that the game shouldn’t be played out. Nor was it intended to substitute for the actual watching of the game. To be honest, I have to ask, along the lines suggested by RickJay, whether this was some sort of irony or attempt at humor that was lost in the textual medium, because if it’s intended seriously, it misses the point by the width of Yankee Stadium.
I assume, based on your post, that you also believe that weather forecasts are completely useless? After all, when a weather forecaster predicts a maximum temperature, and provides a percentage probability of precipitation, they are using similar methodology to that employed by the baseball statisticians who calculate win expectancy. And, just like with baseball, the percentages are not guarantees: if they predict a 20% chance of rain, and it rains, this doesn’t mean that the model is useless or that they were “wrong” in any objective sense. It simply means that the weather patterns defied historical expectations.
There will be cases where teams win games after being down 3-0 in the second inning. In fact, if you look, you’ll probably find that it historically happens about one time in every six. This doesn’t mean that a 3-0 lead in the second is cause to pack up and go home, but it does mean that if your team comes back from a 3-0 deficit to win the game, you can be (even more) happy about it because you know that they’ve defied expectations and had a really good game. I’m not sure how any of this makes the stat useless, or why it should deprive anyone of any of their enjoyment in watching baseball.
Wait, so you’re saying it’s useful for you to know based on a 3-0 score in the second inning, that the team that’s ahead is a heavy favorite to win based on the history of 3-0 leads in the second inning? And that fans should feel deprived if this stat is not included in an online box score or game cast?
Is there an unintended irony that I’m missing?
mhendo’s weather forecast analogy would make a degree of sense if, for example, the weather forecast was entirely based on statistical records (“it’s rained on 34.7% of August 11ths between 1890 and 2019, so we’ll call for a 34.7% chance of rain on August 11th, 2020”). But weather forecasting is a wee bit more sophisticated than that. Similarly, for baseball “win probability” to have a degree of meaning it should take into account how the teams are playing, what the pitching staffs are capable of and so on. But as presently constituted, it does not.
*postscript: the Rays beat the Sox last night, 8-7. Good thing I didn’t bet any large sums on the outcome based on that 82.7% Boston win probability.
I’ve read about people trying to do that using Markov chains to sum up the possible outcomes through every event sequence possible, but I haven’t seen any released implementation of it yet.
Baseball Reference has a pretty good explanation of how they calculate Win Expectancy here:
I find the stat interesting, particularly in looking at the play-by-play after a game has been played. In last night’s Rays-Bosox game, a 2-rbi double in the top of the 7th increased the WE for Tampa Bay by 20%. Then a 2-rbi single in the bottom of the 8th decreased it by 18%. Perhaps not useful, but interesting nonetheless…
You might be surprised how often exactly what you scoff at is used as a tie breaker when the rest of the weather forecast is equivocal.
But to your larger point MLB has always, even from the 1950s, been a hotbed of borderline useless stats. SABRmetrics was all about the idea that traditional stats had negligible actual predictive power and real data science could do better.
Nowadays of course we have “MLB powered by AWS” and any number they can observe or compute is dutifully spit out by first the computer & secondly the play-by-play goon.
I’ve been horrified to hear them mostly talking about gambling odds now, and which teams are making the over/under. Mid game stat readouts to invite you to try to out-arbitrage the bookies while the game is in progress. This will not end well.
I would absolutely feel deprived of this, yes. I am quite curious as to what the win probability is at any given time. I don’t know about you but I’m not Commander Data, and I don’t actually know how often a team ahead 5-2 in the bottom of the sixth wins after the other team loads the bases with nobody out. I would assume they will usually win, but how often is that? 55% of the time? 75 percent? I don’t know. I find that stuff fascinating. I want as much information about baseball as I can possibly get.
Complaining that WP% currently doesn’t include every possible is… well, I don’t get it. WP% tells you SOMETHING. You know what other statistics don’t tell you everything? Here’s a partial list:
Batting average
Stolen bases
Wins
ERA
RBI
On base percentage
Fielding percentage
Saves
Holds
Assists
Putouts
Passed balls
Losses
WHIP
Home runs
Range factor
Slugging percentage
Caught stealing
Hat size
Like, EVERY stat leaves something out. We’re supposed to sift through all of them for the truth, aren’t we? Isn’t that part of the fun? I don’t know everything about baseball but I know this: what I know about baseball will not be enhanced with LESS information.
Not really. It measures team performance and assigns a mostly arbitrary tick to the pitcher. The fact that identical pitching performances could lead to either or a win or a loss means it’s meaningless.
Honestly, nothing more than an after-the-fact indication of how exciting a game or particular moment was. Which is far more information than a pitcher win.
Why do you care? It’s not actually information. It’s not an event. It has no bearing on the actual world. It’s not a result. It’s not an accomplishment. It means nothing. Just wait for the game to be played through and you’ll know who won. Why should anyone care what the probability of that result was at every point in time up to then? At no point does it ever represent any actual event during or actual influence on the game itself or it’s outcome.
Comprehend what? There’s nothing to comprehend. The sole thing it might reveal is whether you should place a bet at that very instant. Beyond that it has no meaning.
How a particular game or moment compares to the thousands of examples that preceded it in baseball history.
I guess if you don’t particularly care for the data visualization that is presented by WP% (example from Game 7 in 2016), you can be content with “Chapman (2-0)”. The former is, to me, far more interesting and insightful than the latter.
Does anyone actually know EXACTLY how WP% is calculated? If not, we/you have no idea what meaning (or absence of meaning) it may have.
if it’s a generic mindless calculation such as “Situation is: bottom of the 3rd, 2 out, none on, score now 3-1. Across all pro ball history since 1880-whenever what percentage of the time did the visitors go on to win?”
A calc like that has less predictive power than the Magic 8 ball™. Small issues like which team is stronger this season, who’s hurt, who’s pitching tonight, have infinitely more predictive power for this particular game right now than “all the history of baseball.”
This kind of thinking is the equivalent of astrology; that somehow the location of Jupiter compared to where your Mom was pushing makes all the difference. Nonsense. By the time it gets to be top of the 9th, visitors down 1-10, nobody needs no steenkin’ AWS to tell 'em the home team has almost certainly won. Back in the 3rd at 3-1 AWS and all the history of baseball has no more idea than your grandmother does.
I’m going to take a wild guess that if the Mets are playing the Pirates and DeGrom is handed a 3-0 lead in the second inning, it means something more to the final outcome than if, say, Toronto goes up 3-0 on the Yankees in the second inning with Jacob Waguespack pitching for the Blue Jays.* But the WP stat dismisses player factors and essentially treats both situations as the same, because, y’know, baseball history.
LSLGuy: “A calc like that has less predictive power than the Magic 8 ball™”
See the link in my previous post describing how WP is calculated, and that’s a fair description. The only factor not related to what miscellaneous teams have done in the past is “run environment”, and that has to do with ballparks and not players.
Of course it’s information. It tells you exactly what they say it does; the percentage of times a team wins given the score, inning, outs and men on base that presently exist. I find that extremely interesting. I care because I’m a baseball fan and the more I know about baseball, the happier I am. I wish I knew even more.
I honestly find your position inexplicable. How does it help to NOT know things? Why is ignorance good?
