MLB playoffs are a crapshoot. You can construct a team that consistently performs well over the 162 game regular season but once you reach the postseason it often comes down to luck and which teams are hot at the right time. Billy Beane has been very unlucky there.
I'm not sure I agree with this. At least in comparison to other pro sports, the MLB playoffs seem to be the most representative of actual talent. The early rounds are 5-game series, and the best teams are playing the worst teams, so you would expect a low sample size is needed to determine the "best" team. In the later rounds, when teams are much more closely matched in talent, the series last 7-games. I haven't done the math on this (would anyone care to?), but I surmise that the statistical power of a 7-game series, spread over 10-14 days, is far higher than that of say, NFL playoffs, where the sample size is 1. This is not even considering the fact that each game has 200+ separate pitches from which to derive success or failure, as opposed to ~100 plays per NFL game.
Baseball has 162 games, which is twice as many as the NBA and NHL, and ten times as many as the NFL. It also has more independent events (pitches) per game than any of those sports. Of all the major sports, it seems baseball is the "fairest" in terms of representing talent, by the simple fact that it requires orders of magnitude more successes of independent events to win the World Series.
There are good reasons why the A's don't do as well as other teams in the playoffs. Mostly, it's that "playoff" baseball is different than regular season baseball. The strike zones expand, the scores drop and the breaks between innings increase to allow for more commercials. There's also more rest/travel days between games. All of this magnifies the value of top pitchers since it allows them to pitch a greater percentage of games, pitch longer in those games and pitch more effectively in those games. Meanwhile, the value of the pitchers at the bottom of the rotation is minimized or eliminated.
One of the reasons the A's have won so much in the regular season is that they make up for the lack of high-priced stars by being solid, top to bottom. As evidenced by this season, when two of their starters went down for the entire year, they had guys in the minors who were almost as good and the drop off was minimal. But whenever one of their guys starts to look really good, they trade him to replenish the minor league stock. It happened with Hudson, Mulder, Haren, Gonzalez and Cahill. This incredible depth of similar players works really well in the regular season and makes the A's pretty much injury-proof. But it works against them in the playoffs when 1 stud pitcher can basically dominate them as we've seen with Justin Verlander.
They've been eliminated twice by him in games where he basically pounded the outside corner on a pitch that isn't a strike but the umpire was giving the entire night, albeit to both pitchers. But since the A's don't have a pitcher that can hit that spot with near-100mph heat reliably, the Tigers were able to take advantage of a few location misses. When the schedule means that you're facing Verlander for 40% of a series and the extended breaks between innings make it easy for Verlander to pitch the entire game, it all adds up to the A's getting beat and, I think, explains why Beane went out and got Samardzija to pair against Verlander or whoever that #1 guy they face is thinking that both of them can pound that outside-the-zone strike with upper 90s stuff equally well.
Also, the OPS-minded strategy doesn't work as well against #1 guys as it does against #4 and #5 guys who you're not going to face much in a playoff series. And with lower scores, the teams that manufacture 1-2 runs in a game do better in the playoffs than the teams that know that sacrificing will lead to fewer runs over the course of a season.
It should be pointed out that not a SINGLE sabermetrician has shown the "facts" in this post to be true. BPro used a similar concept called "secret sauce" that had many of these ideas in them, and it was routinely debunked years after it was published.
The A's are almost certainly the victim of variance, the likes of which apparently even Hacker News readers can't grasp. For them to have the record they do in the playoffs given their expected talent level is something like a 17% shot on chance alone.
Ever see something happen that was an 83% underdog? No?
"For them to have the record they do in the playoffs given their expected talent level is something like a 17% shot on chance alone."
Not an expert on the subject, but baseball doesn't have and explicit chance component, so please explain what you mean when you say "on chance".
I assume, embedded in your definition, will be statistics about the same teams playing each other or something. However, it seems like it would be hard to apply these statistics to refute someone who is saying both that "the A's are different" and "championship games are different".
For instance, if Michael Jordan wins 100 one-on-one games of basketball against me, I could come out an say "Yes, but if we played on the International Space Station, then I'd win. ISS games are different.". To refute that you'd need a body of evidence of basketball games on ISS. Let's say you have 5000 games of history on the ISS, where the players' success rates are almost identical to those on Earth's surface. I could still say "But I'm different. Everyone else tries to play ISS basketball like Earth basketball, but I have these cool tricks.". Controlling for both of those things seems to require having lots of data about me playing on the ISS.
None of that means the post to which you responded is right. But if the claims "the A's are different" and "the championships are different" both seem plausible; and if the A's record in championships is poor compared with the regular season; then you should dial back the confidence a bit.
>This is not even considering the fact that each game has 200+ separate pitches from which to derive success or failure, as opposed to ~100 plays per NFL game.
NFL plays are VASTLY more multivariate than any single pitch in MLB. This is comparing apples to garbage dumps, at best.
I don't know about that. #4 and #5 starters are very important for a team's season-wide performance. But in the playoffs, these players see very little action.
