Who might regression strike?

We’ve danced around the R-word all of last season, and then all of this offseason: regression. In 2014-15, the Flames carried what many viewed to be an unsustainable shooting percentage. After all, they were shooting at 10.5% last season, when in 2013-14, they only shot at 9.2% – a difference of about 30 goals.

It’s substantial. And if the Flames’ possession woes continue, and their shooting percentage drops – after all, the league average has been 8.9% the past few seasons – then we may see a drop in goal scoring, and with that, probably wins.

However, just because team regression is expected doesn’t mean it’s going to affect everyone. 

Forwards

A small caveat: not everyone is included because, well, some of these guys have only played one season. Sam Bennett, Micheal Ferland, Johnny Gaudreau, and Josh Jooris aren’t included here, because there’s hardly anything to compare them to; they’ve only played one season (and not even, in Ferland’s case).

Comparatively, players like Paul Byron, Joe Colborne, and Sean Monahan – guys who really haven’t played much, but have more than one full season under their belts – are, even though we don’t have a lot to go off of. After all, for someone like Monahan, who has only played two seasons and has posted high shooting percentages both years, we don’t know if he’s just a really accurate shooter, or if he’s been lucky his first two years in the league.

forwardssh

Mikael Backlund, Brandon Bollig, and Michael Frolik have generally been below-average shooters, Mason Raymond average, and the rest, above average.

Bollig, Byron, and David Jones are the only Flames who had worse seasons last year than one would have expected from them based on their career averages, most notably Byron. Byron, however, has only played 138 NHL games, 57 – or 41% – of which came in 2014-15. He may score more than he did last season, but a drastic increase probably shouldn’t be expected.

Bollig, who has played 187 NHL games – 62 (33%) last season – is in a similar position. He scored one goal during the 2014-15 regular season; not much more than that should be expected.

Jones, on the other hand, has a resume of 387 games to draw upon. He wasn’t that far below his career average – just 1.1% – but if he regresses to his mean, combined with presumably finding a steady place in the lineup, his scoring could go up.

The news isn’t as great for everyone else. Even though Backlund and Frolik were worse than the league average in 2014-15, it’s been that way throughout their careers, and they actually had better 2014-15 seasons than may have otherwise been expected from them. This could be an indication of growth, but considering how both are in their mid-late 20s and have played hundreds of NHL games (Frolik significantly more so), it’s probably just who they are.

Colborne, Monahan, and Mason Raymond stayed about the same; Colborne and Monahan do not have as much NHL experience, but Raymond, over his 513 NHL games, likely is who he is.

Matt Stajan saw a small jump, and he could dip for 2015-16. Especially when you consider he’s likely to get a lot of defensive zone starts, and probably won’t get too many shots on net – he has never gotten 100 shots on net over a full season with Calgary – his scoring output may drop.

That leaves us with two huge spikes to talk about: Lance Bouma and Jiri Hudler, both of whom had career years in 2014-15. Bouma is only 25, and his next NHL game will be his 200th, so there is a chance his career average to date isn’t an accurate reflection; that said, for the time being, it’s reasonable to be wary if he can match an output similar to last season.

Hudler, on the other hand, is 31. Throughout his career, he’s always been a high accuracy shooter – but only once before 2014-15 has he ever been that accurate. He’ll still likely put up a lot of points, especially if he sticks with Gaudreau and Monahan, but a dip in his production wouldn’t at all be surprising.

Defence

Note the following graph has the NHL average shooting percentage at a different number than the one above; that’s because, well, defencemen tend to be less accurate shots than forwards. We’re looking at these guys on a different level.

defencesh

It isn’t surprising to see Deryk Engelland and Ladislav Smid be below-average shooters: they aren’t guys you rely on for offence.

Kris Russell, on the other hand, while also not expected to score a lot, does have some offensive instincts in him: and he had a down year in 2014-15. It took him all the way until Feb. 4, 2015 to score his first goal of the season. He’s a veteran of over 500 games, too, so it’s not as if the disparity between his 2014-15 season and career average is negligible. He very well could put up better numbers this upcoming season, particularly as he’s likely going to be placed in a role he’s more suited for.

Dougie Hamilton had a bit of a down year as well, but he only has three seasons in the league and 178 games, so he’s a little more difficult to get a proper read on. An improved season for him wouldn’t be surprising, though; especially now that he’s a part of a Flames defence that encourages its blueliners to jump up into the play.

Mark Giordano’s career season in 2014-15 wasn’t a fluke or a stroke of luck: he kept pretty much right in line with what his career averages are. Since getting more ice time, though, he’s started putting the puck on net more, resulting in increased offence for the Flames’ captain. He’s been consistent ever since being elevated to the top pairing and having another top pairing defenceman with which to work, and that trend should continue this season.

There are two major spikes to look at in terms of greater production, though: T.J. Brodie and Dennis Wideman, who both saw massive 2014-15 seasons compared to their careers to date. Brodie, 25, is entering just his fifth season in the NHL, but his offence has consistently improved with each passing year. It’s possible he was an abnormally high shooter in 2014-15 and won’t have nearly as good a 2015-16; it’s also possible he’s still growing as a player, and scored at a level we can expect for the future.

Wideman, however, is a completely different story. He’s 32 years old. He’s a veteran of over 700 NHL games. He had an excellent season last year, yes: but we know it’s not likely to repeat itself. He isn’t Brodie, still coming into his career and developing offensively. We know who he is by now, and he isn’t as accurate a shooter as last season would have you believe. He still puts the puck on net a lot, so he’ll likely continue to put up points from the blueline; that said, the chances of him having another 15 goal season aren’t particularly great.

    • beloch

      Yeah, those are the ones I’m expecting too. Perhaps Gaudreau might face a sophomore slump, but Treliving has improved this team considerably with another top 6 forward and top 2 defence man added.

  • MattyFranchise

    To say Regression assumes a Norm for each player and a return to the norm from a high level. I would rather think of it as normal variance, or chatter than a Norm. Perhaps if you plotted several individual players over their whole careers we could see whether there actually is a “Norm” or whether the numbers are (normally) all over the place?

  • The Fall

    Flames shooting percentage will go down, but their shot totals will likely go up.

    Moneyhands will ride a high shooting percentage for most of his career. He is a perfect fit for Bob’s system.

  • beloch

    Nitpick: That’s the wrong type of graph to show this data. The lines between points shows intermediate sh% values between players, but this is meaningless. You should use bars or points instead.

    Suggestion: You might want to include a goal or shot weighted graph, to give an idea of what regression would mean for the team. Bollig coming up to league average would have less impact than Hudler coming down to league average simply because Hudler shoots much more often.

    Otherwise, nice article!

      • beloch

        If every Flame had a shooting percentage bang on the league average last season, the Flames would have scored 36.6 fewer goals. The top line (Monahan, Hudler, Gaudreau) would have produced 40.1 fewer goals and the rest of the team would have produced 3.5 more.

        As I pointed out previously, the Flames’ high shooting percentage is driven almost entirely by their top line. Everybody outside of those three shot a combined 8.7%, which is below league average.

        The real question of whether the Flames sh% will cause significant regression this season boils down to whether or not the shooting percentages of Gaudreau, Hudler, and Monahan will regress significantly. Hudler shot above his career average last season, but not significantly so, and it also wasn’t the highest sh% he’s had in a full season either (he shot 19.7% over 81 games in 2011/2012). Monahan only has two seasons under his belt with shooting percentages of 15.7% and 16.2% respectively. Gaudreau only has one season of data to go on, but the eye test says he’s a sniper and will likely have an above average career shooting percentage.

        To be blunt, I don’t think the top line is going to regress to anywhere near league average this season. As a result, the Flames’ are likely going to have a high shooting percentage again this season, and it’s going to be driven by the top line. Get used to people saying the Flames shooting percentage is unsustainable, because it will likely remain above average.

        Statistics say the league’s average shooting percentage will not change much next season. It’s harder for individual teams to defy that average for long, but individuals frequently do so for their entire careers. Some players are more skilled than others (e.g. Hudler vs Bollig). If the entire Flames’ team was shooting above league average last season, regression would be highly probable. However, that’s not the case. It was mainly three highly skilled players.

        Some Flames players are going to regress. Bouma would be a top choice. Wideman shot below league average, but above his career average last season. He will likely get less ice-time (thanks to Dougie), shoot less, and score on fewer of his shots. His point total is going to go *way* down, but those goals will be compensated for (i.e. by Dougie).

        Bottom line, the Flames are going to score a lot of goals this season if they start shooting more, as we hope to see with the addition of positive possession players like Hamilton and Frolik and another year of development from the young rookies.

        • beloch

          It also helps that Hudler is a shot% driver (that is, he consistently has a high on-ice shot%). I think (hope) Monahan and Gaudreau increase their volume this season.

          • beloch

            It will be interesting to see how the lines shake out. If Hartley splits up the top line to form two scoring lines, one likely with Bennett, then Monahan and Gaudreau may play with someone less offensively gifted than Hudler. If that happens their shooting percentages may go down. However, one hopes that the overall output of the top six will improve this season. It almost has to with the addition of some combination of Frolik, Ferland and Bennett. People have heaped praise on the Flames’ blueline, but their top six is getting dangerous. It may finally hit home if the team’s possession improves and the top six gets fewer defensive zone starts as a result.

  • beloch

    Haven’t even played one regular season game and authors on this site can’t even stay away from the word “regression”.

    You damn well know if the flames keep winning this year all will be fine but if/when the flames do start to slip every fancy stats nazi is gonna be pouncing like a pack of wolves on everyone to say I told u so…in fact some (Ryan Lambert) are likely saying prayers every night hoping this occurs so they can tell us that they were right and how smart they are

    • MattyFranchise

      I want to add to this and say that there are writers here talking about regression as though it’s a bad thing. In math, regression to the mean is common. So under performing players have a chance to ‘regress to the mean’ or their league average SH%.

      Some will go up and some will go down. But over a long enough timeline they should ‘regress’ to their career %.

  • MattyFranchise

    The Flames may not do as well this year. Remember the 2011 Lightning? They overachieved by getting to the conference finals. Then they missed the playoffs the next two years, only to re-emerge as top contenders with more experience and good team building. It wouldn’t surprise me to see the Flames follow a similar path.

    But if they do worse this year, most of it will not be due to to regression to some mythical mean. One big hole in the current practice of statistical analysis is the idea that every player and every team reverts to the same mean. Perhaps the genius of a great GM is to find value players who can compensate for the team’s low possession or shooting percentage stats.

    Take a look at this video clip of Leo Messi, and tell me that some players don’t transcend conventional statistical analysis.