The Diminishing (but Positive) Returns of Tanking

Before we get going, a heads up: this article is about a mathematical model that explains tanking. I’ll give you a preview. The way baseball is currently set up, it’s no surprise teams tank. If you want to fix that, the game’s competitive structure must change. I suggest a few ways to accomplish that change at the end of the article, but be warned: the bulk of this is a no-nonsense dissection of why teams keep tanking even as the returns go down. Personally, I think the game should make the changes I suggest, because the boom and bust cycle of team contention makes for a fraught fan experience. The way the game is set up now, however, it’s no surprise that teams do it.

The logic behind tanking is straightforward and solid. Being in the middle is the worst; there are no prizes for winning 85 games, flags fly forever, and so on and so forth. You’ve surely heard it enough times that you don’t need a repeat, but just for completeness’s sake, we’ll do it one more time.

By trading present concerns for future value, you make your team better in the future. As a byproduct of trading present concerns away, your major league roster gets bad — bad enough, hopefully, that you’ll move to the top of the draft. Additionally, with no pesky need to be competitive, you can use your major league roster to give borderline players extended tryouts. Hit on a few of them, and that’s even more tailwind for the future.

In essence, tanking is making a bet that taking a step back now will let you take two steps forward sometime in the future. Even if that isn’t the case, being quite bad for a while and then quite good for a while sure sounds better than being mediocre the whole time. Tanking works on both axes, which explains its continued appeal. Do you think your team will win something like 77 wins? Blow it up! 75 wins? Blow it up! 70 wins? You guessed it.

Tanking isn’t the only way to build a good roster. It’s not even the best way to build one. The best way is by building for the future and present at the same time. The Yankees and Dodgers, and to a lesser extent the Cardinals, Nationals, and Rays, are current exemplars of this model. If you start with a solid roster and manage player acquisition and development well, there’s no reason you need to take a step back, particularly if you can supplement unexpected holes with the money you bring in from all that winning you’re doing. Every team in baseball would prefer that model. If you’re starting with a 70-win husk, however, you can’t exactly start competing for division titles tomorrow.

Ten years ago, that was the end of the story. Zigging while the rest of baseball zags is a great way to get ahead, so long as zagging is a decent decision in a vacuum. As more and more teams tank, however, the equation starts to get murkier. Leaving aside the fact that it’s not great viewing, multiple tanking teams fight each other on many axes.

By selling off parts at the same time, they lower the return on those parts. By competing for spots in the draft, they make the expected draft returns of tanking lower. By both getting good sometime in the future, they lower the value of being good in the future — being the best team in the league is worth a heck of a lot more than being one of the two best teams in the league. The more teams that tank at once, the more these effects get magnified.

That’s all straightforward, but the quantitative value of these effects is less clear. I decided to construct an abstract framework to better understand how much worse tanking gets when multiple teams attempt it at the same time. I don’t claim that it’s a perfect representation of what happens when two teams try to bottom out at the same time, but putting some rough parameters around the situation feels like a step in the right direction.

First, consider this 10-team subset of a league. Why 10 teams? It gives our league enough differentiation between the best and worst without trying to simulate all of baseball. In the first year of our simulation, we know each team’s true talent level. If there were no luck in baseball, here’s how many games each team would win:

Of course, there is luck in baseball. Even if you know how good a team is, runs scored and runs allowed don’t always cleanly translate into wins, and talent doesn’t always translate cleanly into production. To simulate this, I bumped each team’s win totals randomly, with a standard deviation of nine games, a number that Pete Palmer derived in 2017. I then re-scaled teams’ winning percentage using the odds ratio method assuming they played each other team an equal number of times.

For example, if we ran this season an arbitrarily large (100,000 in this case) number of times, here’s how often each team would finish first in our ten-team league:

Heck, even the 71-win team wins sometimes on a long enough time horizon. That’s the nature of baseball — every so often, a team catches lightning in a bottle. This model isn’t complete — there are no Dodgers in this world. It’s also not deterministic; I’m using a Monte Carlo simulation, which is why the two 81-win teams don’t have equal outcomes. Run this again, and you’d get slightly different answers. For our purposes, however, it’ll be good enough.

Next, we need to simulate several seasons. Tanking is a many-year process, so just looking at a single year doesn’t make much sense. I decided to simulate 10 seasons, long enough to let a tank job come to fruition. After each season, I bumped each team’s true talent level up or down randomly, again in a rough simulation of real life. Teams aren’t static piles of talent over time. They get better or worse between years, as players develop or trades and free agency. I also included a small mean reversion component (5% of a team’s talent level relative to average) to account for the fact that the salary and draft structure of baseball exerts a normalizing pressure.

Let’s look at our 10-team group again. This time, we’ll look at the percentage of first-place finishes each team has over 10 years of play, again repeated 100,000 times:

Mean reversion matters a lot. The best teams are still the best teams, but regressing 5% of the way to the mean (a 91-win true talent team becomes a 90.5-win true talent team, for example) pushes everyone together over 10 years. The best teams are still the best teams, and the worst teams are still the worst, but time flattens things out.

Now we have our neutral-state world. Let’s introduce a tanking team to break out of the downside of being in last. Our last-place team is going to embark on a stylized tanking campaign. In year zero, they’re going to drop 20 wins. That’s exaggerated, but we’re going for broad strokes. That step back will last four years. In year five, they’ll gain 35 wins. Again, the actual transition would be more of a glide path, but in the name of expediency, we’re making it a single step higher.

Tanking slightly increases the total number of games the team wins, and more importantly, it increases their cumulative championship odds significantly. Why? It avoids being run-of-the-mill bad; being a terrible team for a while doesn’t affect your championship equity very much, because as we learned, the 71-win team almost never wins in the normal course of things. Leapfrogging upwards at the end, on the other hand, puts you right in the mix at the top of the table. The tanking team is living large now:

You read that right: the team that started bad and got worse projects to win the fourth-most league crowns over a 10-year period, even after completely punting the first four years. Don’t read too much into the specific numbers, because the values I’ve assigned to the tank and also to year-to-year variation aren’t gospel. This, though, is what drew teams to a tank-and-rebuild strategy in the first place. It saves money. It increases cumulative championship probability significantly. What’s not to like?

Next, let’s add a second tanking team. Now our 74-win team is also going to be tanking. This does two things: first, it slightly decreases the returns to tanking. In our model, I simply lopped a win off of the tanking reward at the end of the rainbow for both squads. Next, those two teams will now be competing for titles. That cuts into the number of titles they should expect to win when the returns come in. Observe:

The top teams continue to lose championship equity to the two tankers, but now they’re cannibalizing each other as well. When one team tanked, that team gained 7.6 percentage points of championship equity (10.6% – 3%). When two teams tanked, they gained an average of 6.1 percentage points. Let’s toss a third tanker in the mix, while simultaneously lowering the rewards by another win:

Now we’re looking at only 4.7 percentage points of average gain for tanking. You’ll notice, though, that each team is still “right” to tank, if the objective is simply to maximize the number of championships you’ll win. Each team is improving that number despite the diminishing returns available. Finding 20 dollars on the street isn’t as good as finding 50 dollars on the street, but it’s still better than nothing.

Let’s keep it going! I kept adding a tanking team and reducing the returns to tanking until I found an equilibrium. Here’s what happens to teams’ win probability as more and more teams tank:

It doesn’t make sense for the eighth team to tank; they’d actually lose 2.7 percentage points of equity by giving up on the early, relatively non-competitive years. Don’t take this as some immutable statement of fact that 70% of the league should try to tank. I could monkey with the initial assumptions and form a completely new grid. For example, here’s how it looks if tanking only adds a net 10 wins in year five and if the returns decrease more quickly as more teams tank:

Now the fourth team to tank gets barely anything, while the fifth team shouldn’t tank regardless of what its competitors are doing. You can toy with the exact constants in the model as much as you’d like — you could make a league where every team except the best one should be tanking, or where only one team should, with enough tinkering. If you want to get fancy, you could create staggered tanking over time: when the first wave of tanking teams gets good, teams that avoided it in year one may want to change course.

Regardless of what your specific outcome grid looks like, two things are clear. First, more teams tanking lowers the efficacy of each individual team’s strategy. Second, even if tanking gets less effective as more teams do it, it’s still a good plan for each of those teams. If you’re the second-worst team in your league and the worst team is tearing down, you’re stuck between a rock and a hard place. You’re not competing now, and their long-term plan will make it hard to compete in five years — there will be an extra good team in the league then, after all. What can you do other than tank alongside them and try to compete in five years?

This paradox explains why teams continue to tank. The returns really are worse when multiple teams do it; sometimes, you’re the Phillies and tanking doesn’t get you where you hoped to go. The more teams compete to occupy the same space, the harder it is to occupy that space.

Even knowing that, however, if your entire goal is to maximize the number of times you win the title (in this case, first place in our 10-team league), you should still tank. It’s unfortunate, certainly, that the actions of other teams hurt your expected returns, but the returns are still positive. You might not enjoy your rebuild as much as you’d hoped, but it still beats sitting in the middle, particularly if you can save your owner some money while doing it.

If you generally agree with the construction of my model but also dislike tanking, you might find this conclusion depressing. Within the framework I’ve laid out, I think that’s simply unavoidable. Being the worst competing team in the league at a given time just isn’t going to work out well, regardless of assumptions. If you want to fix it, you’ll have to change the framework.

What do I mean by that? The reason that tanking works is that baseball’s structure makes it somewhat easy to move wins across time. The long team-control window and draft structure mean that accumulating prospects is both doable and valuable; if you set up a core of cheap talent, you can supplement it with free agents and reach phenomenal competitive heights without breaking the bank.

To stop that from being the case, the compensation structure needs to change. I don’t mean that players need to get a bigger piece of the pie — that’s a reasonable point, but it’s not part of this discussion. Simply moving the distribution of money around — paying free agents less but young players more — would do a lot of good. Shortening years of team control, with the understanding that free agent paydays would come down commensurately, would also stop teams from blithely kicking wins forward at the expense of today’s team.

Changing the draft would help. The NBA switched to a draft lottery to make finishing last in the league less attractive. Any number of structures — a draft order determined by games won after being eliminated from playoff contention, a set order, giving teams a set bonus pool and abandoning a draft altogether — could make finishing at the bottom of the standings less valuable from a draft perspective. Heck, tying shares of the league’s television deal to a minimum competitive standard would stop tanking in its tracks.

This list isn’t meant to be exhaustive, and many of these decisions might not be feasible. The point of this conclusion, and of the article as a whole, is merely to emphasize this: if you want to stop teams from purposefully tearing down today to build for tomorrow, you can’t just hope that this is merely a phase, that too many teams tanking will make the entire act pointless. Being one of two tanking teams isn’t as good as being the only one, of course. But if you’re already bad, it’s likely still better than standing pat. Fixing that core fact will require a major change to the competitive structure of the sport.