clock menu more-arrow no yes

Filed under:

The Devils Win and Loss Streaks Don’t Mean Anything

New, comments

The Devils have been a very streaky team and it’s been kind of a roller coaster. Unless you investigate the probability of course -- in which case this is to be expected.

Philadelphia Flyers v New Jersey Devils Photo by Jim McIsaac/Getty Images

Over the Devils 5-game winning streak, a lot of people finally started believing. Everyone’s emotions were riding high. We had average 4 goals a game over that span, Brian Boyle had 8 points in 5 games, we just came off two multi-goal victories and an intense shootout victory over the Rangers, and we stood at 3rd in the NHL in Pts% behind only Tampa and Vegas. We had at last come to terms with the idea that the Devils were for real.

Then we had a hilarious (except not really) string of 3 games, in each of which the Devils had a controversial, if not blatantly wrong call — Gerard looked at them here, with pictures. Next, a 1-goal loss to Dallas came in a game with bizarre and uncharacteristic mistakes from defenders. Then they lost to the Islanders in another game with a questionable call -- this time for goaltender interference, and legitimately lost to the Flyers -- though, it was a game in which percentages paid them no favors. Bada bing, bada boom — 6-game losing streak.

Then, all of the sudden, two wins! A big pair of divisional victories -- first against the Islanders, then a roller coaster of emotion against a team the Devils really struggled with recently — the Capitals.

But then ...

Another losing streak!!

@#$@(&%@#&@*#$#@(_#

I thought we got out of this!

4 games in a row which included Hakstol and the Flyers taking out our top line en route to a drubbing, and a Hall injury that rendered us inept offensively, with two 3-0 shutout losses, and even a game in which we hugely outperformed an excellent Bruins team in pace of play. Nothing was going well.

Until the All-Star break, where we came back with a win against a Sabres team that hadn’t let up a goal in 200 minutes, redeemed ourselves in a win against the Flyers, and showed the business to a white-hot, 2-time defending Cup Champ Penguins team. We’re back baby!!!

So we’ve finally gotten the ship righted for good this time right? RIGHT?

No.

Streaks Are Random

During the second losing streak, between the losses to the Bruins and Predators, I posted this tweet:

I saw a lot of people thinking that there were significant changes that we should be making to coaching style, or to who was on the roster, or to what lines we should play, etc. because they were tired of losing. That made me realize that not everyone thinks about these stretches the same way I do -- the way that I’d argue we all should.

Hockey is random. In 2014, Josh Weissbock capped our ability to predict the winners of hockey games at about 62%. This means that even a perfect computer algorithm with all the data we have right now would get predictions wrong over a third of the time. As a results, all teams. no matter how good or bad will lose games -- frequently multiple games.

There is a calculation that you can run to figure out the probability of a team getting a losing streak or winning streak of a certain length. This nightmarish looking beast of an equation is how you do it. I’ve loaded this equation into an interactive Desmos graph that you can play with to find the probability that a team with a losing percentage of p would go on a losing streak at least m games long at some point in the year (you can also change the game total which is currently set to 82). The Devils have a win percentage of about 53% right now. The probability of a team with that ratio having at a win streak of at least 5 games at some point in the 82 game season is 86.6% and the probability of that team having a 6-game losing streak is 41.5%. So the longest streak of each type we’ve had this year is not that unusual. It’s barely even unlikely.

What we should be doing is looking at underlying stats. As an example, that Corsi graph above looks remarkably stable since December, despite erratic results in the win/loss column. What’s going on here?

Streaks Don’t Tell You About The Team

This is a chart of the Devils 5v5 statistics during each of the streaks I mentioned.

The biggest thing that stands out to me, is how crazy low the PDO numbers are. Look at that most recent losing streak ... 1.94 shooting percentage and a 0.911 PDO??? I mean, come the hell on. You could take 200 shots and lose 5-4 at that pace. And I know I’ve already blamed the refs for the first losing streak, but that 0.953 PDO isn’t much better. In comparison, this most recent streak is buoyed by a 1.05 PDO due to really great goaltending.

So those percentages certainly seem to be highly correlated with the results. Let’s look at CF% for comparison. I’ve adjusted for score and venue (that’s what the SVA means) for that column to make sure wer’re not rewarding the losing streaks for their additional shots. But even with that, we can see that the Devils were almost two full percentage points better in the 6-game losing streak as they were in the 5-game winning streak. Furthermore, we can see that although the Devils are only 10-7-3 in their last 20, as mentioned in the tweet above, our foundational statistics are actually more encouraging than the were at the beginning of the season when we were 17-9-5 but had a CF% SVA over 4 whole points lower. That’s better, and more definitive news than anything regarding streaks

Concluding Thoughts

First of all, our trend is strong on a season-wide scale as a middling to back-end playoff team. Don’t let a few losses or wins in a row distract you from that. Second of all, streaks of this length are a product of randomness and luck more than performance. As an example, PDO numbers were the more definitive culprit of the streaks than Corsi.

Don’t cut Zajac because we lost 4 games in a row. Don’t sell future assets because we won 5 in a row. Just view the season on the whole and use context and perspective as often as possible. Bad bounces will happen, but it’s likely that good bounces will come as well. Hockey is a random sport, shooting percentage is fickle, and weird things happen. If you enjoy the roller coaster of emotion, then I’m not gonna stop you. But if you’re looking for reason to alleviate your stress, this article should show you just a little bit of why nothing that has happened here is weird.

Thanks as always for reading guys and gals. Comment below if you want to yell at me and stuff. Until next week, friends!