STATISTICAL ANALYSIS | Sometime in the early 80s, Bill James of “Moneyball” fame noted that a baseball team's true strength could largely be determined more accurately by looking at runs scored and runs allowed than by looking at wins and losses. To be more specific, he found that one can predict future win-loss records more accurately using only past runs scored and runs allowed, as opposed to using only past wins and losses. Here is the formula, known as the Pythagorean win expectation: Runs for^2 divided by (Runs for^2+ Runs against^2).
The same can be said for hockey using goals-for and goals-against.
For example, when the 2009-10 Washington Capitals scored 318 goals and gave up 233, using the Pythagorean formula we could expect them to win 53 games. They won 54, or 65.9 percent of their games.
|Capitals||Actual wins||GF||GA||Expected Wins||Difference|
We can also use this win expectation to figure out the likelihood that one team beats another in regulation. I will spare you the gory mathematical details, but it is based on a method called Log5, published by Bill James in the 1981 Baseball Abstract.
Using this, if Washington goes up against a team like the Florida Panthers, who have a .530 expected win percentage at home, we can determine that Washington's chance of winning that game is 36.4 percent. If a team is given a 30 to 40 percent chance of winning, it wins on average 35 percent of the time. If the team has a 60 percent or better chance, it wins 61 percent on average.
Here are Washington's expected win probabilities for the rest of the season:
|58||@||Tampa Bay Lightning||33.8%|
|62||@||Toronto Maple Leafs||38.1%|
|63||New York Islanders||74.6%|
|64||New Jersey Devils||64.0%|
|67||Tampa Bay Lightning||78.7%|
|69||Toronto Maple Leafs||59.5%|
|70||@||New York Islanders||43.9%|
|73||@||Detroit Red Wings||13.3%|
|80||@||Tampa Bay Lightning||33.8%|
|82||@||New York Rangers||21.9%|
This shows that there are likely 10 wins and 11 losses remaining. That would put the Capitals’ record at 38-34-5 (81 points) with five games as “toss-ups.”
These probabilities do not take into account the return of Nicklas Backstrom or Mike Green, nor any deadline deals that are made between now and Feb 27. However, if 92 standings points are needed to get in the playoffs, the margin of error is razor thin no matter what the future holds for the roster.
Follow Neil on Twitter: @ngreenberg