NHL Playoff Analysis

       Welcome to InsaneGambler.com. Ahhh, there's nothing like staying up all night and crunching some numbers. Waiting breathlessly, hoping against hope, that your research will produce a dynamite handicapping angle that will produce winners by the boatload. Sadly, more times than not, all the work ends up for naught. We're not sure if this analysis of all the seven-game NHL playoff series ever played will help anybody win some bets, but it has produced some interesting statistics. Read on, porcupine.

       First of all, we need to explain exactly what this analysis is based on. We could just give you the results, but that wouldn't be any fun. Plus, you wouldn't appreciate all the hard work that went into getting those results. Not to mention all the stat freaks who need every single thing accounted for or else their world will come to a blazing end.

       The first NHL playoff series was way, way, way back in 1918. There were series played before that for the Stanley Cup, but the NHL wasn't born until 1917. The first seven-game series wasn't played until 1939. Before that time, there were five-game series, three-game series, and even two-game series. In a two-game series, the winner was decided by total goals in the two games. Guess what? We don't care about all that. Since they only play seven-game series now, that's all this study will concern itself with. Just wondering, what is the plural of series? Is it also series, the same word?

       The 1939 and 1940 playoffs had three series in which the order of games was different then it is now, so we didn't count them. The 1939 finals were played in the format used today, the team with the home advantage playing games 1, 2, 5 and 7 at home. There were also some series later on in which the format had the better team at home for games 1, 2, 6 and 7. They got axed too. Anyway, when all is settled and done, we've come up with 550 seven-game series with the 1-2-5-7 home format. This includes the 2010 playoffs and since I'm still waiting for my paycheck, I haven't bothered to update it with the 2011 games. Or 2012. Or 2013. Or 2014.

        IG: Here we go again! Get on with it man. The average attention span of a web surfer is about 4.2 seconds. This isn't a college thesis you're submitting. Just give them the results. If there are stat geeks out there who want all the details, they'll e-mail you. And stop complaining about your pay! In the meantime, GOWI!

       Okay, boss. When all else fails, give them a table!

Results of 550 seven-game NHL playoff series
4-0  8515.5  18  3.3
4-110218.5  37  6.7
4-210218.5  7513.6
4-3  8014.5  51  9.3

        What's interesting here is that if the team with the home ice advantage wins the series, it's pretty much an even choice as to whether they'll win in 4, 5, 6, or 7 games. But if the road team wins, there's a definite bias to it being in 6 games. In the words of Tony Dunst, let's break it down:

If the team with the home ice advantage wins the series, here are the probabilities:
4-0 - 23.0%
4-1 - 27.6%
4-2 - 27.6%
4-3 - 21.7%

And if the road team wins:
4-0 -   9.9%
4-1 - 20.4%
4-2 - 41.4%
4-3 - 28.2%

       Of course, these stats are all dependent on how much better the team with the home advantage is than it's opponent. And here is some info you probably won't see elsewhere. IG's expert numbers cruncher figured out the regular-season winning percentage for all the teams involved in the 550 playoff series. Figuring winning percentages in most sports is easy but in hockey, how do you treat ties? Before shootouts began, let's say there was a team with a record of 43-27-10. Actually, there was a team with that record: the 1981-82 Boston Bruins. If we don't count the ties, their WP is .614, but if we treat each tie as 1/2 win and 1/2 loss, their WP drops to .600. For this study's purposes, that's what we did: 1/2 win, 1/2 loss.

       But what about those pesky shootouts? The ruination of statisticians everywhere. How do you treat those? Well, to be consistent, we just pretended the shootouts never happened. If a game is tied after the 5-minute overtime, it goes into IG's books as just that, a tie.

       So what did we get? In the 550 seven-game series studied, the teams with the home advantage had a .624 regular-season winning percentage. The opposing teams won at a .532 clip in the regular season. Using the The Magic Formula, we find that the better team should win games at a .593 rate. How accurate will that prediction be? Sorry, here comes another table.

Single-game win-loss records of team with home-ice advantage in 550 seven-game NHL playoff series
     1   359   19165.3
     2   360   19065.5
     3   282   26851.3
     4   300   24954.6
     5   284   16463.4
     6   153   15549.7
     7     80     5260.6
TOTAL 1818 126958.9

       Did you notice that there are only 549 game fours? What happened to number 550? In game 4 of the 1988 finals, Edmonton and Boston were tied 3-3 at the old Boston Garden when...the power went out. The game was called off, and another Game 4, which was really Game 5, was played in Edmonton, with the Oilers winning the game to sweep the series. We're not counting that as a Game 4, since it should have been played in Boston. For this table we're counting it as Game 5 because, well, because this is our study and we'll tinker with the numbers any way we want.

       All right, back to business. So the predicted winning percentage was .593 with the actual being .589. However, the percentages used in the formula only applied to the first four games. After that, many of the weak teams have been swept by the powerful ones, thus resulting in a slightly lower winning percentage for the favorite. If we take only the first four games, the WP is .592. So the magic formula's predicted WP was .593, the real WP was .592. Dang, that's pretty close, don't you think? [ Uh, oh. We just did some research to confirm the theory of weak teams being eliminated early and guess what? It's somewhat busted. Sort of.] Making things worse, we're going to have to make ANOTHER table to show these stats properly. Are you totally confused? Here we go:

Single-game predictions and results for 550 seven-game NHL playoff series
     1  62.4  53.2+6.165.465.3
     2  62.4  53.2+6.165.465.5
     3  62.4  53.2 -
     4  62.4  53.2 -
     5  63.0  54.1+
     6  63.2  54.4 -6.152.950.0
     7  62.5  54.2+6.164.660.6

I suppose you'd like to know what those abbreviations mean at the top of the table. RSWP stands for regular-season winning percentage. HIA is home-ice advantage. PWPCT is predicted winning percentage. AWPCT is actual winning percentage. GIGIG!.

Now let's do another type of breakdown. Specifically, what happens in game 5. Ugh, another table.

Results for Game 5 in 550 seven-game NHL playoff series
3-1102  5266.2
2-2133  7464.3
1-3  49  3757.0

IG: I hate to say it, but this is getting a bit complicated. The tables aren't all that easy to interpret. Please do a little more explaining as to what this whole thing means. Make it easier to understand! GIGIG!

All right, boss, I'll do my best. The table above breaks down 447 Game Fives in 550 NHL Playoff Series. Remember, there were 103 sweeps and in this table we're not counting the Bruins-Oilers game in 1989. Home teams that were up 3-1 were able to seal the deal 66.2 percent of the time, but when they trail in the series 1-3, their WP drops to 57.0. I'm sure the difference is simply that a team leading 3-1 has proven itself to be superior whereas a team down 1-3 is obviously not as good as its regular-season WP would indicate. More to come.... (This is a work in progress)

Unleaded IG

IG takes pride in having accurate and updated information. If you see something that needs to be corrected, added or deleted, please e-mail us at: error@insanegambler.com

IG = Always On Tilt