Context is always crucial.
I feel like I’ve already read three different post-mortems on the Toronto Maple Leafs already this postseason, each piece a different account of a central theme: the Toronto Maple Leafs got crushed in Game 1 because they were due for a loss or two. No way could the wins keep coming they way they were in the regular season. The real Leafs are the ones we saw get crushed 4-1 by the Bruins.
That part is true, I guess. The most confident series’ predictions I made were Chicago and Boston advancing, but I don’t think that a poor performance in Game 1 is indicative of how the rest of the series will flesh out. Yes, the shots were 40-20 for the Bruins. Yes, overall shot attempts were 71-33, which is absolutely insane. Yes, I had the scoring chances at 24-9 for Boston.
So by no measure were the Maple Leafs fit to compete in Game 1. They were simply out-matched, and they will likely be out-matched for the remainder of the series. But that doesn’t ever mean it’s over. The playoffs are the equalizer. No matter how many wins you got in the regular season, you need to win 16 of your next 28 games, and in the proper order, otherwise you’re going home.
The regression hounds were out in full force as the Leafs kept getting pinned in their own end. The final score was 4-1 but it could have been 5-1 on an odd disallowed goal. I think the Detroit Red Wings were up 3-1 against Anaheim when the SB Nation official Twitter feed sent this one out. I have to admit, I retweeted it because I laughed.
Weird. Instead of a hockey game NBCSN has decided to show a documentary on statistical regression?
— SB Nation NHL (@SBNationNHL) May 3, 2013
All-in-all though, and it was pointed out after the Ducks tied it up, that the joke about “regression” was realistically, made basing the results of a one half of a game stacked up against the 50 the Ducks have played. Realistically, 1 per cent of the season. Even good teams fall behind by two or three on almost regular occasions.
There’s a misconception about regression. Or, there might be. I have no idea what people might take away from my articles when I suggest a team or players’ performance is unsustainable. Myself and other bloggers tend to be pretty aggressive with predictions and forecasts and it can take a while for a certain forecast to end up being vindicated, but let’s use the one above as an example. “Regression” isn’t something you’ll see in a night.
It may seem when we keep posting the PDO numbers of teams like Toronto or Anaheim that those teams are “due” to drop. But just because you’ve flipped a coin three times and it’s landed “heads” each time, doesn’t mean that there’s a greater chance of “tails” on the fourth flip. Strictly speaking, if the Leafs and Ducks were able to win with respectively high shooting and save percentages in the regular season, there’s no guarantee that will stop in the post-season. That’s not to say it *will*. I’ve hedged my bets in playoff pools and in certain series’ picking any team I can get favourable odds on. Those are Boston and Los Angeles. But I would never stick my life savings into the final four teams in each conference being the best two teams in Corsi Tied. I would likely bet all of my life savings*, in fact on the Conference Finals not involving Los Angeles and Chicago on one half and Montreal and Boston on the other.
But that doesn’t mean more modern analytics are useless. Just because something isn’t perfect doesn’t mean it doesn’t have value. The team with the higher seed will generally win 55% of the series’, the team with the most points (sometimes a lower seed) will win about 56% of the series and the team with the best Corsi Tied will win about 60% of playoff series’. (To get these numbers I backed up with my own research after mathematician Rob Pettapiece and I sent a few emails about the same topic)
I’ve been wondering over the last couple of weeks if the playoff run last year by the Los Angeles Kings was a total fluke. I had them beating the Vancouver Canucks in the first round, as did a few other stats guys, when hardly anybody else did. But nobody said that they’d get hot and win the first three of each series and lose just four games en route to the Stanley Cup.
The Kings are now down 0-2 in their series to St. Louis this season, a team they’d beaten on eight straight occasions. Each winning goal was a miscue on the part of Jonathan Quick, and it’s getting to the point where everything has to go so right to win a series. You need to be shooting better than you were in the regular season. You need to have spectacular goaltending, and, generally, you also need to be a good possession team. Not an awful lot of negative-possession squads make it to the Stanley Cup Finals, and a lot of good ones do make it far in the playoffs. It’s only happened twice in our five-year history of possession stats—both times Pittsburgh, although in 2009 they were an excellent group after firing Michel Therrien for Dan Bylsma but nothing was guaranteed for them either. They had to win two Game 7s along the way, and you won’t have success if you keep getting into a series of one-offs.
(Side note to that above point: the Boston Bruins have played in 10 playoff series’ in their recent run of playoff appearances. They’ve gone to seven Game 7s, so lots of long series. They’ve gone 3-4 in those games, but timed the three wins correctly. All of them came in the same year they won the Cup.)
Analytics became mainstream sometime around last season when there were angry fights across the Internet about how good the Minnesota Wild really were. Part of me wonders if the absolute capsizing of the Wild and subsequent dominance of the Los Angeles Kings were two false positives for predicting future outcomes. The Ducks and Leafs should have after 20 games, been the next to crash and burn, and the New Jersey Devils never should have lost so many games down the stretch.
So it always feels good when you write a piece and instantly the subject can’t catch a break anymore. I had that happen last December with Minnesota. On December 12 2011 I wrote an article about the Wild comparing them to Tim Tebow and his recent hot streak. They had just finished the end of a seven-game winning streak. They would lose eight straight and win just 14 times in their last 52 games. Another example would be James Mirtle’s article in the Globe and Mail on April 14 about Nazem Kadri’s high shooting rates. Kadri was a point-a-game player on April 14 when the article ran. Since then, he’s scored just three points in eight games including the playoff game.
But that’s generally not how it works, so if you’re getting into these , after more events have taken place and the luck slowly balances out on a larger scale, rather than a smaller one.
I did take the Red Wings in the series but not because I think the Ducks will “regress to the mean”. I took the Red Wings in the series because I think they’re a better team than the Ducks, and I’m not expecting over the final five games of the series (or more, if Anaheim advance) that we will see all the flaws of the Anaheim group exposed.
That’s not really how it works. These sorts of things work out eventually if the respective coach and management team allow the same thing to happen to a team. Anecdotally, I can suggest that the Ducks are getting out-shot perhaps less frequently than they were at the start of the season, which means they’re working to fix the problem so that they can potentially win a game or two when the percentages stop being there for them.
Ah but the Leafs look bad.
*I’m a pro blogger, so my life savings is basically any change I can find in the couch.