I added winning percentage to the league, team, and coach pages last night, and boy has that generated a lot of e-mail. Several people have told me that I should use the following formula to calculate winning percentage:
(a) points / (2 * games_played)
That works just fine through the 1998-99 season, as the league-wide winning percentage is always .500. But in 1999-00 we encounter the abomination known as overtime losses. Using the formula above, the league-wide winning percentage is no longer .500, but .525. When shootouts are introduced in 2005-06, the league-wide winning percentage jumps to .557. Does that seem right to you? Should the league-wide winning percentage really be .557? To me the answer is no. What I do to calculate winning percentage is the following:
(b) (wins + 0.5 * ties) / games_played
Using this formula, the league-wide winning percentage is always .500.
Let me give you an example. In 2007-08, the Columbus Blue Jackets won 34 games, lost 36 games in regulation, and lost 12 more games in overtime or in a shootout. Using formula (a), we get a winning percentage of .488, very close to .500. However, the reality is that the Jackets won 34 games and lost 48 games, so their winning percentage shouldn’t be close to .500. Using formula (b), we get a winning percentage of .415. Think about it: formula (a) says the Jackets won almost as much as they lost, while formula (b) says they lost a good bit more than they won. I think (b) measures what we really mean when we talk about winning percentage.
I’d like to know what you, the users, think. Please let me know if you agree or disagree with me, and, if you do disagree, explain why my line of thinking is wrong.
UPDATE: Taking a page from Solomon’s book, I decided to add both points percentage (formula (a) above) and winning percentage (formula (b) above) to the pages.
UPDATE II: I’m just going to display points percentage for now and see how things shake out over the next few months. Comments are being closed (just for this post).