NHL Playoffs vs. March Madness: How much harder is it to win a best-of-seven compared to one game?

One thing that makes the NCAA tournament so fun is that each game counts. One loss and you’re out. That means a worse team getting a couple of lucky bounces can often beat a better team having an off-night. But in a best of seven series, that’s not supposed to be as true. With more games to play, the truly better team should come out on top, right?

My question is: How much of an effect does a best-of-seven series truly have?

For instance, take the NYR-PIT series, in which Vegas odds (right now) say The Rangers have a 65% chance of winning game one and the Penguins have a 35% chance of winning. If those odds don’t change through the series, what are Pittsburgh’s odds of winning it all?

To answer this, I wrote some code in R (hockey_sim.r), and ran tens of thousands of simulated playoff series. Here’s the answer:

hockey_sim

According to this graph, The impact of the best-of-seven compared to a single game gets bigger as the teams get further apart in ability. The New York Rangers have a 65% chance of winning any single game, but an 81% chance of winning the series. Pittsburgh is on the other end of that equation.

hockey

And as you might expect, the more games in the series, the more pronounced this effect gets.

hockey

Take-aways:

  1. Playing multiple games in a series has a serious effect on the likelihood of winning.
  2. The further the odds are from 50%, the bigger the effect gets.
  3. The more games you play, the bigger the effect gets.
  4. The magnitude of this effect is surprisingly small. Playing up to seven times as many games only improves New York’s odds by 15%