Pebble Hunting: The Best Teams Should Make The World Series; A Defense of Bud Selig

For all the excitement of this postseason’s individual games, there is a fairly common sentiment out there that something sucks about a system so random that sub-par teams get to fluke their way to the World Series, thus stripping the season of its power to make sure the best teams are rewarded. Why play a long season and then reduce the championship to coin flips? Why continually expand postseason until every champ resembles Chris Moneymaker? Zachary Levine foretold this postseason in his epitaph for Bud Selig, written in August:

There was divisional play before Bud Selig showed up, but we never saw the meaning of the championship altered like we have in the last two-plus decades. If you look back on Selig's main innovations that concerned the season, each decision weighed the notion that the best team (or at least a great team) would be crowned champion against financial or other concerns. And each time, the meaningful champion lost.

Selig's tenure saw the expansion to six divisions, which can lead to subpar division winners, the addition of a wild card, the reintroduction of the five-game series, the unbalancing of the schedule and the invention of interleague play with different opponents. These all serve to create randomness and decrease the chances of better teams advancing. (The second wild card has effects in both directions on this.)

I’m not invested in convincing you of anything, so take everything I say here in good faith, not in the sort of argumentative “gotcha” style of writing that you hate and ignore. I’m also not going to dispute matters of taste. If you want a baseball season to end with the best team jumping up and down, and you put a higher priority on that than in having an extra couple weeks of playoff baseball and playoff commercials, your wishes are valid. I’m not going to bother with the argument that tournaments are fun; wanting there to be a point to all this baseball season stuff is a very legitimate desire. Watching bad teams win over good teams can be stupid and lousy. I won’t try to convince you that watching the Royals win instead of the superior Angels is somehow morally better. It’s not. There are no wrong positions here—except, in the spirit of humility, perhaps mine.

But what I will propose is this: The unfair system isn’t nearly as unfair as we think it is, and Bud Selig’s record of screwing over good teams isn’t so clear as we think it is.

An important thing to take into account is that these series are not coin flips, at all. This year it has looked like they are, because the worst team has won almost every series (and, indeed, almost every game). But that’s the anomaly. That’s a freak show that we’ll probably never see again. Over the course of even one game, significant talent differences emerge: PECOTA thought that the Angels were more than 70 percent likely to beat the Royals in a couple of the scheduled ALDS games, and over the course of five games those emergences multiply: The Angels were 78 percent likely to win that series.

In one sense, the fact that the Royals knocked out the Angels is Bud Selig’s fault—he put the Royals into the postseason; if he hadn’t, they’d have had a 0 percent chance of knocking out these Angels. But in another sense, the fact that the Royals knocked out the Angels is due to a very unlikely roll of the dice. The fact that the Royals were playing the Orioles, a team they were evenly matched up against, instead of the Tigers, a team that would have been strongly favored over them—I’d estimate a 75- to 80-percent favorite over the Royals in a seven-game series—was also due to a very unlikely roll of the dice: The Tigers were 70/30 to win the ALDS.

So, let’s accept that the Royals are probably the worst of the AL playoff teams, as judged by their performance over the entire season. Obviously, they’ve looked amazing this October, as teams that are winning do, but if we had done a survey on the final day of the season based on the season, we would have probably agreed that the Royals were the worst of their five. We might consider it an abomination that the worst of the five would make the World Series to represent the league. But how likely is this?

If we treat each series like a coin flip, dictated by the screw-logic limits of small samples, then their odds would go like this:

• .5*.5*.5 = 12.5 percent.

But if we’re a bit more sophisticated in our math than that, and acknowledge that these series weren’t coin flips at all, but (imperfect) measures of team strength, it’s more like this:

• P(WC win)*P(ALDS win)*(P(ALCS win over BAL)+P(ALCS win over DET)) or
• .46*.22*((.3*.5)+(.7*.25)) = 3.3 percent

The playoff system, then—excluding the World Series—is an effective enough judge of talent to shave the Royals’ chances by nearly three-quarters. This is not bad! A 3 percent chance of the fifth-best team winning the league’s pennant doesn’t seem like such an abomination.

Meanwhile, if you’re fairminded and want to see the Angels represent the league because you think they’re the league’s best team, the odds are something like:

• .78*(.7*.60+.3*.80) = 51 percent.

Note that I just used estimated odds for the Angels over the Orioles or Tigers, but point holds: The Angels were not just as likely to represent the AL as the Royals were, or even twice as likely (because they didn’t have to play in the Wild Card game), but 15 times as likely. This is a system that makes it 15 times more likely that the great team will advance over the merely good team. Also not bad!

Put one more way: The chances of getting the four worst teams in the LCS—or, at least, the two worst division winners and two Wild Cards—was about 1 in 140, according to PECOTA. Selig made it possible, sure, but we’re responding to an incredibly unlikely sequence that reflects nothing like the typical season.

So: Perfect? Not if your only goal is to get the Angels to represent the AL because you think they’re the best team. If that’s your only goal, then you might just have the best record in each league go to the World Series, like in the old days. That would double the Angels chances, and make it impossible for the Royals to get there. If that’s your goal, that’s cool with me.

But is that your goal? If you want the best team to make the World Series for each league, then what happens to the A’s, who had the best third-order winning percentage in the league—who, by our most sophisticated measure of team performance, were the best team in the league this year? They miss the playoffs entirely.

So here’s the fairness-based argument for the current playoff system, or for any playoff system that grants multiple division winners, and maybe a wild card, entry: It gives great teams a back door into the postseason, which (the odds suggest) does a pretty good job of sorting the great teams from the merely okay teams.

This back door turns out to be pretty important. It didn’t used to be, back in the 1950s, when there were only eight teams per league. Back then, even the system of choosing one team per league (based on record) did a good job assuring that the best team per league (based on third-order) would get to play for the World Series. But with each expansion the likelihood that the third-order “champ” would also have one of the best records in the league has shrunk. Here’s how often the best third-order team has had the best record in its league, or one of the two best records in the league, in each of the various expansion eras:

In a system where only two teams make the postseason, nearly half of the “best” teams would be left out. If we double the number of playoff teams to four, as in the system of the 1970s and 1980s, then more than a third of the best teams would be denied a chance to play for the crown. If you want the best teams in the World Series, well, don’t you want the best teams? Or, in other words: Pity the Cubs, who had the NL’s best third-order record in 1969 and might have finally won a title, but for the overly restrictive playoff entrypoints of the day.

Now, my submission of third-order winning percentages to decide who is the most deserving team might not be your jam. You might figure that the point is to win games, so just use wins over a huge sample to decide who to reward—either in a pre-1969 system or a pre-1994 system. Again, I’ll agree that it is a fine position. But I find it unsatisfying. For one thing, regular-season wins remains a somewhat arbitrary means of translating “good play” into rankings, just as the multi-stage playoffs are. “The point is to win games” in a 162-game season is no more absolute than “the point is to win in October.” Why 162 games? Why nine innings? Why four bases to produce a run, and why three outs? All these are decisions that somebody once made about the structure of the sport, which we accept partly because they’ve stood the test of time and partly because we’ve grown up with them, but there’s nothing God-ordained or immutable about them. Somebody made an arbitrary choice, wrote down the rules, the participants go along with them, the teams build their strategies around them, and we accept them. All those words apply also to a tournament-style championship process, with personal exemptions for the last clause.

But that’s a bit heavy. A more grounded argument is this: Even the “best record” way of deciding who is most deserving involves elements of randomness. Schedules aren’t balanced. Even if they were perfectly balanced (which seems neither possible nor optimal in an era of 30 teams), there would still be randomness: Maybe every team plays the Dodgers nine times, but what about the team that plays them during the month when Kershaw is injured? Or the team that plays the Red Sox after they’ve held a trade-deadline firesale? Or the team that plays the A’s before Jim Johnson has been replaced as the closer? Or the team that plays the Astros before the Super Two deadline? Or the team that plays the Nationals immediately after back-to-back doubleheaders? Or the team that plays the Angels when they’re just resting their regulars, having already clinched? Balance is always going to be an illusion at best, and randomness will always keep us from saying with any certainty that a 97-win team played better than a 96-win team.

So, given the fact that we’ll never get it right, we’re stuck with a system that gives us the wild card Royals and the wild card Giants. This goes to the classic criminal justice conundrum: Is it more unjust for an innocent man to be imprisoned, or for a guilty man to go free? Is it more unjust if a team that doesn’t “deserve” to go to the World Series does, or if a team that does is denied even the chance to win in October? For the more serious analogue—the innocent/guilty question—we normally say it’s more unjust for the innocent man to be imprisoned. For the less serious one—baseball—I’d argue the same principle applies. The fact that occasionally—very occasionally—we see a Giants/Royals World Series isn’t a feature (as opposed to a bug) of the system, but it is an inevitable result of a feature of the system. I’m perfectly content with the possibility, and I’m even content when it leads to the actuality.

Huge thanks to Rob McQuown for research assistance.