October 26, 2018

Why the Red Sox WILL (probably) win the World Series

Because they are up 2-0. Period. No real baseball analysis is needed. Don't believe me?  Read on....

If instead of the outcome of the games being decided by who scores more runs in nine innings, it was based on a simple coin flip, we can build a simple model to see what the probability of winning a Series at any given point would be.  And luckily it involves VERY simple math that is easy to understand. For example, the odds of flipping heads two times in a row are 1 in 4 (50% x 50% or 1 in 2 x 1 in 2). So, if a team is down 3-2 in a Series, the probability of them winning both games and taking the Series is 25%.

Now, we all know the games aren’t based on coin flips, but I believe using a coin flip model helps to get at least a rough idea of what can be expected.  I first mentioned this in 2004. Down 3-0 to the Yankees, the odds of the Red Sox coming back to win didn’t look good. Using a coin flip model, it was easy to see that the odds of winning 4 games in a row (or flipping a coin heads 4 times in a row), was 1 in 16 or about 6%. Now, this doesn’t sound very promising, but it’s not the 1 in a million that people made it sound like. (For comparison purposes, the odds of being dealt pocket aces are 1 in 221. Also not likely, but everybody has had this happen to them, several times if you play poker often enough). Sure, it had never happened in the history of baseball, but only 24 times previously had a team been up 3-0.  The Red Sox coming back and winning and making it 1 time in 25, is pretty consistent with what would happen if every game was a coin flip*. It should happen about 1 and half times out of every 25. Once is slightly less than expected, twice is slightly more.

Anyhow, without even needing to use any math, it’s obvious that any time the same number of heads and tails have been flipped, the probability of getting a certain number of heads before the same number of tails is the same as tails reaching first (In other words… if teams are tied 1-1, or 2-2, the odds of either team winning a 3 of 5 or 4 of 7 are the same).  Going through all the permutations of heads/tails out of 5 or 7 flips, you end up with this chart showing the probability of the leading or trailing team winning, depending on the current Series score:
The Red Sox are up 2-0 right now, based on the above they have an 81% probability of winning 4 games before the Dodgers do. I saw a graphic after Game 2 that said teams with a 2-0 lead have won 84% of the time.  This is pretty damn close. The above chart is a useful reference if you ever want a quick and dirty guesstimate of the probability of a team to win given their current situation. It’s not perfect, because it ignores some minor details (specifically, it ignores BASEBALL, and everything related to it), but a handy thing to keep (literally and figuratively) in your back pocket.

(*) Note: I’m NOT saying that the outcome of each game is the same as a coin flip. Obviously some teams are better than others, pitching matchups, home field, injuries etc., all need to be taken into account. But when you get to the World Series, more often than not both teams are at least closely matched up. Even a 100 win talent team is only going to beat a 90 win talent team about 55% of the time in a given game. If we assume that in the previous 24 times, on average, the leading team had a 55% probability of winning each game, the chances of the trailing team coming back are (0.45)4=4% which is one time in 25, matching EXACTLY what has happened.

October 02, 2018

Why the Red Sox won't win the World Series

(Greg M. Cooper/USA TODAY Sports)
No, it’s not because of their bullpen. Without going into a lot of deep analysis, simply put it’s because there are 10 teams in the playoffs. The Red Sox may have the best one, and may have the best probability of winning it all, but it’s still far from a sure thing, or even a likely thing. 

Read on….

The Atlanta Braves made the playoffs 14 seasons in a row from 1991 to 2005, but only won the World Series one time and were considered a disappointment because of it.  With eight teams making the playoffs, on average, a team will win once every eight times it gets in there (slightly more if they’re better than the average playoff team, slightly less if they’re worse). Fans and baseball analysts usually grossly overstate this “slightness”.  More on that in a minute. Anyways, the Braves would have been expected to win 1.75 times out of those 14. If they had won twice they would have actually exceeded the expected.

But, if they were always the best team in the league surely they should have won more often, you may think. But not necessarily. As I alluded to above, most people overestimate the advantage a better team has over an inferior one. Sure, the better team is more likely to win, but the advantage is typically small. It’s rare in a playoff series to have a team be expected to win with more than a 60% probability.

October 01, 2018

WAR - What's it good for?

It's George Steinbrenner! Wait,
maybe this is the other boss
A whole lot more than you may think, Mr. Shaughnessy!  

Near the end of the season, Dan Shaughnessy, who is somehow still employed by the Boston Globe, tweeted that the Red Sox bullpen was ranked first by WAR, and that tells him everything he needs to know about WAR. 

His implication is that this is a made up stat that only nerdy geeks can determine on their supercomputers and it’s not indicative of anything. I mean, everyone knows how bad the Boston bullpen is, and there’s near universal agreement that it’s our Achilles heel heading into the playoffs. So any stat that shows it as being good is obviously flawed.

Sigh. 🙄 Where to start?  First of all, as the second most famous #45 to our readers would say, this is #FakeNews. The bullpen is actually ninth in WAR according to Fangraphs. Ok, well #9 is still pretty good, and this bullpen obviously isn’t, so PIG’s point stands regardless. {Sidebar: I HATE the CHB moniker for Dan Shaughnessy. It originated with Jurassic Carl (whose nickname I DO love), but it is somewhat homophobic and has no place in today’s world (unlike dinosaurs….), so I prefer to call him Pigheaded Irrelevant Guy, or just PIG for short}. But, what PIG fails to take into account (or did, but intentionally ignored so that the conclusion would fit in with his agenda), is that this is a season long stat. And in the first half Boston did have a pretty lights out bullpen: Craig Kimbrel was an all-star with a 1.77 ERA at the break. Joe Kelly, Bobby Poyner, and Matt Barnes had ERAs of 1.73, 1.69 and 1.95 respectively at some point in June.  Carson Smith was a reliable 8th inning option until he had a season-ending injury. Hector Velazquez was 5-0 as a long reliever.

The pen HAS struggled lately, but it was ranked 3rd by WAR in the first half, and 23rd in the second half, for a season ending ranking of 9th. This easily passes the smell test to me. 

Anyhow, if WAR is flawed as an indicator of bullpen effectiveness, then what would be a good stat to use? I don’t think that even PIG is naïve enough to suggest wins and losses. As useless as they are for starters, they are that much worse for relievers. Saves? For the most part they really only measure what your closer can do. How about ERA? Although somewhat flawed, it is the traditional stat that may have the most relevancy. How does the Boston bullpen rank in ERA? How about NINTH! Hmmm…. If WAR is useless what does that tell you about ERA? 

No, WAR is not useless. In fact bullpen WAR has such a good correlation with bullpen ERA (because I am a nerdy geek with access to a supercomputer - or at least a 3 year old laptop and excel - I ran it and confirmed that bullpen WAR did have a very high correlation, much higher than ERA to wins or saves, see here:)
Nerd Results
that without digging deeper into the numbers, that alone should tell you that although it may not be perfect, it is still a reasonable surrogate for the effectiveness of a bullpen. 

End of rant. Let the postseason begin!