The NBA uses a variation of what is called a Borda Count voting system to select its MVPs, where voters (100 American and Canadian media members who aren’t affiliated with teams) are asked to rank their top MVP candidates, and then those candidates are given points based on where they are ranked on each ballot. The player with the most total points from all the ballots wins.
While most discussion about voting systems focuses on political elections, non-political elections like the NBA’s MVP vote are just as affected by the voting systems they use, and Borda Count systems are fantastic for non-political elections like the NBA’s MVP vote.
In political elections, different voters often want opposite things, and voting systems try to consolidate this kind of disagreement. Elections like the MVP vote are different; they try to consolidate agreement about what things are good and disagreement about how much each of these good things matter. Everyone agrees that playing 80 games at an MVP level is more impressive than playing 70 games at an MVP level, and everyone agrees that having a true shooting percentage of 70% is better than having a true shooting percentage of 65%, but not everyone agrees how much games played matters compared to scoring efficiency. One voter might care a lot about VORP, another might not care at all about, but no voter is going to penalize a player because he has a high VORP.
I’ve looked at non-political elections where voters agree about what qualities are good for a candidate to have and only disagree about how much different things matter and found that Borda Count is extremely good at finding the winner that best matches what voters as a whole care about most. In general, Borda Count’s strengths and weaknesses as a voting system work extremely well for picking MVPs. Its weaknesses mostly don’t matter in the context of MVP voting and are much more relevant for political elections, while its strengths match what you need MVP voting to do.
|Extremely good at picking a winner most people either like or don’t mind.||Very susceptible to manipulation in situations where you can recruit additional candidates to run.|
|The winner is likely to be the least controversial option. (Though that doesn’t mean the winner won’t be controversial at all!)||Factional/controversial candidates rarely win. Don’t use it to try to elect a diverse city council!|
|If there’s broad agreement among your voters, Borda Count will pick the consensus candidate virtually every time.||Asks voters to rank many candidates and won’t work quite as well when voters don’t rank all the candidates.|
Borda counts stand out from other voting systems, even the Ranked Choice (Instant Runoff Voting) system used for the Oscars at being extremely good at picking broadly agreeable winners, almost no matter what. (It can be hard to manufacture a realistic hypothetical scenario where Borda Count doesn’t pick a broadly agreeable winner.)
This also means that the major threat to Borda Count picking the right MVP is systematic bias among the MVP voters. One-off idiots like whoever who put Derrick Rose as their top MVP pick for the 2020-21 season get headlines, but they don’t put the MVP award at risk, because the points assigned by the other voters will easily swamp out their silliness. However, because Borda Count is so good at finding consensus among voters, it is also very good at capturing whatever biases are common among the voters.
Weirdo MVP voters with ridiculous ballots might not put the MVP award in jeopardy, but shared biases in how the voters perceive the candidates can send the MVP to the guy.
No other voting system can fix this problem; there isn’t a voting system that can magically see through the bias of its voters and pick the One True MVP. A voting system can only take what its voters give it. But that doesn’t mean that there aren’t things the NBA and the MVP voters can do to mitigate the risk that the bias of the times prevents deserving MVP winners from winning the award.
Including more voters, especially voters who are less susceptible to “groupthink”, can help. Voters who aren’t influenced by the same narratives that others are help neutralize the biases of other voters. Having voters rank more than 5 candidates can further protect the MVP from a handful of weirdo voters, but Borda Count is so good at inoculating the results from weirdos or extremely factional candidates that this probably wouldn’t change much, if anything.
Another way to reduce the likelihood that voters’ biases prevent a player from ever receiving an MVP he deserves is the much maligned voter fatigue, where voters tend to get tired of voting for the same players over and over and eagerly rally behind whatever real competitor arises. Which brings me to…
The stupid and smart arguments for taking into account past MVP wins when choosing a new MVP
Trivia about how voters behaved in the past (“Only these all-time greats won multiple MVPs in a row!”), coincidences about the order in which all-time greats won MVPs and championships (“X player didn’t get his Yth MVP until he won a championship!”), should not matter when choosing an MVP winner, especially because these historical facts are sometimes the result of screwups from MVP voters. Who has or hasn’t won three MVPs in a row has absolutely nothing to do with whether Jokic deserves MVP this year.
Most of the arguments for taking into account the history of the MVP award that I’ve seen are in this “stupid” category of arguments, which demand that the precedent of how MVPs were awarded in the past should override merit of this year’s play. However, I think there’s a smart reason to take into account who has won MVPs before: we don’t know what unfair biases we have that affect the MVP vote. Putting some emphasis on spreading out MVPs among top players reduces the risk that one player will win several MVPs while another misses out only for us to later come to believe that the MVP-earning player was worse than the MVP-less one.
If there’s an extremely close MVP race between a candidate who has won an MVP before and a candidate who hasn’t, I see the two possibilities as:
- The candidate who had won a previous MVP wins another. Best case scenario: He deserved all of his MVPs, and the voters did a great job. Worst case scenario: He actually wasn’t the most deserving player in any of his seasons, and another more deserving player may never the MVP he deserves.
- The candidate who didn’t win any previous MVPs wins his first one. Best case scenario: He deserved to win the MVP, and the voters did a great job. Worst case scenario: The player who had already won the MVP should have won another.
The second option seems much better to me. When the competition is very close, as it is this year, the risk of awarding one less-deserving player one year seems far less bad than the risk of showering one player with multiple MVPs only for us to later believe that other players were more deserving.
For instance, there was a time when a player who was great at taking and making contested midrange shots could easily get overrated by MVP voters compared to a player with a more efficient but less aesthetically exciting game. Advanced stats have shifted the narratives surrounding players to the extent that limitations of today’s advanced stats are likely culprits for misplaced MVP awards. Advanced stats aren’t in their infancy, but they’re certainly not adults yet, and the degree to which they overrate or underrate different players is probably something we’ll be figuring out over the next decade.
We know that we’re going to have better ways to assess players in the future and that we may misjudge players’ values in the moment, especially when they’re playing differently than players did in the past. The big 3 MVP contenders (Giannis, Embiid, and especially Jokic) all play significantly differently from past players and especially each other. Direct comparisons are really difficult: they don’t just have different strengths and weaknesses, they play in completely different ways, meaning that the differences between them are especially susceptible to systematic biases in the voters: aesthetic preferences and which statistics voters overvalue or ignore are going to be particularly important when none of these three players is anything like a “poor man’s version” of the others.
Maybe in the future we’ll come to believe that Jokic or Giannis were the most deserving of the 2023 MVP, but we’re just as likely to think that Embiid should have won when he hasn’t. With such a close race, I think the tie should go to the (past) loser; Jokic or Giannis missing out on a third MVP would be far less tragic than Embiid not getting an MVP when he’s having at least an equally good season as the other two and, for all we know, we may someday come to think of him as the best of the three.
There’s no perfect solution to make sure the MVP never goes to the wrong player or that everyone who should win an MVP trophy does. But the voting system the NBA uses is about as good of a choice as possible, and a little voter fatigue might actually help ensure that more deserving players receive MVP awards.
2 thoughts on “The NBA MVP voting system is actually good. Voter fatigue might make it better.”
[…] Western political voting. I won’t go into too much detail (the ever-educational Tim Schmitz does so here, if you’re so inclined) other than to say that the Borda system is designed to eliminate an […]
[…] range in how these offenses performed. At the top, you have the 76ers, whose past and (probably) future MVPs led them to the third best offensive rating. At the bottom, you have the third-worst offense […]