Can MVP voting really tell us who the NBA's best player is?
If you're into analytics, you might dismiss votes as the opposite of salient analysis on determining who is best, but we also have to assume MVP voters take the important stats of an era into at least some consideration. If you assume one MVP vote can't tell you much, what if you went a little deeper? One way to measure performance is to look at MVP voting over multiple years. Although the best player might not win the MVP award every season, it is reasonable to assume that any player who has received significant MVP support for a stretch of seasons is a legitimate candidate for the "best player" label.
And we can measure MVP support by looking at something called award shares. An award share is simply award points won by the player divided by the maximum number of award points.
For example, last season LeBron James won 1,207 points in the MVP voting. Had he received every first-place vote -- which he should have, by the way -- James would have won 1,210 points, so his award share is 1,207 divided by 1,210, or 0.998.
Now that we have a way to quantify a player's voting support, we need to come up with a way to measure his established value in a given season. I decided to use the following formula:
● 0.4 times his MVP award share in season y, plus
● 0.3 times his MVP award share in season (y - 1), plus
● 0.2 times his MVP award share in season (y - 2), plus
● 0.1 times his MVP award share in season (y - 3)
In English: by using an accumulation of a player's MVP support over several years, you have a strong argument for a player as not only an MVP but the absolute best player in the league in that time frame. The MVP award was first handed out after the 1955-56 season, so let's start with 1959 (giving us four seasons of voting data) and use the method outlined above to determine the league's "best player" on a season-by-season basis, according to the award shares formula. The time span for which each was the league's best player also is noted.