Learning R: Expanded playoffs

Last week, I promised that I would find a way to simulate the expanded postseason to figure out exactly how likely it was for a sub-.500 team to win the World Series. You know what they say, though: Promises are made to be broken. Or maybe that’s rules.

Anyway, I didn’t figure it out, but it wasn’t for any of the typical reasons for why I didn’t do something I said I would (laziness, depression, social anxiety). No, the reason why I didn’t simulate an entire postseason is good ole fashioned incompetence. This stuff is hard as it turns out.

I don’t have nothing to show for the week, however. I managed to simulate exactly one (1) postseason series, but I couldn’t figure out how to do an entire postseason in a way that would be easily repeatable.

After simulating a 60-game season (with some code taken from Jim Albert), I first had to figure out who was going to the playoffs. If the season makes it that far, the proposed playoff structure will include 16 teams. The top two teams from each division will earn a playoff spot, and two more Wild Card spots will be given to the two remaining teams with the best record in each league.

This is the season that I was working with. Note that talent levels were randomly assigned to each team, so that’s why the Mariners and Royals were among the best teams in the majors.

When Albert originally wrote this code, the expanded playoff structure was but a twinkle in Rob Manfred’s eye, so I had to edit the RMD file taken from GitHub to reflect the changes to the structure. Out of 300+ lines of code, I changed three, but hey, I needed to know which three to change.

While I was at it, I thought it would be worthwhile to re-make some of the graphs Albert originally made when he explored the relationship between talent and playoff chances in a 60-game season. If you haven’t read the piece, go check it out. I’ll wait.

Welcome back.

The most obvious change is that the wild card spots generally went to worse teams. Of course they did. Opening the playoffs to more teams means more of the rabble is let in. How much worse were they?

Originally, wild card teams were slightly above average. Below is the talent distribution under the normal playoff structure. A team with a talent level of 0 is roughly a .500 team. A team with a talent level of 0.2 is a .550 team, and a team with a -0.2 talent level is a .450 team.

Jim Albert

Here’s how that changes when the playoffs are expanded.

The peak of that wild card distribution is just to the right of 0, so the typical wild card team is slightly better than the average team but not by much.

Wild card spots generally go to Goldilocks teams. That is, teams that are good enough to remain above the scrum, but not great enough to rub shoulders with the elite. Previously, this Goldilocks zone was around a 0.25 talent level or a .550 to .570 winning percentage or thereabouts.

Jim Albert

Again, expanding the playoffs moves that curve to the left. It also gives the best teams a little more certainty.

Now, the Goldilocks zone is centered somewhere around 0.05. The average talent level of 4,000 wild card teams spread across 1,000 simulations was 0.05.

Naturally, fewer wins are needed to make it to the postseason. Before, a .500 team had about a 12 percent chance of making it to October.

Jim Albert

A team that wins as many games as it loses has better than a coin flip of making the postseason under the current structure.

Playoff chances rise almost linearly with each win. I don’t think anyone particularly likes the current structure, but it’s hard to deny that each game is crucially important (at least until you get to 37 wins). Expanding the playoffs certainly dilutes the talent level of the playoffs, but if the goal is to make the regular season more exciting, this achieves it with flying colors.

In those 1,000 simulations, wild card teams won anywhere between 25 and 39 games. The most common win total was 31 which would mean the typical wild card team still had to win more games than they lost. It was certainly possible for scrubs to sneak through, however.

I am somewhat surprised by the results. I thought that once more than half the teams got into the playoffs, the average wild card would be a sub-.500 team. I still think the threshold is too low, so when/if baseball gets back to a 162-game schedule, I hope expanded playoffs are thrown out with the extra innings runner on second. Still, if this season makes it to the end, the playoffs will be less of a farce than I thought.

Next week, I swear I’ll simulate a whole postseason. I know more or less what I need to do, I just need to bone up on how to write my own functions.

---

Kenny Kelly is the managing editor of Beyond the Box Score. You can follow him on Twitter @KennyKellyWords.