2013-14 Utica Comets: Analytical Year in Review

If you’ve been reading my weekly prospect reports here throughout the season, you noticed that I’ve been sprinkling in AHL #fancystats here and there whenever I could. Well, now that it’s over we’ve got ourselves a large enough sample size to deconstruct everything, and attempt to make sense of what exactly happened to the Utica Comets in 2013-14.

If you’re a regular reader of this blog, most of this stuff should make intuitive sense since it’s just an extension of what we use to analyze the Canucks on this platform. If not, don’t worry, because it’s all fairly simple and broken down into easy-to-read segments as we go along.

Just to shill myself — if you have an AHL blog, or an NHL blog, and want the equivalent data for your (farm) team let me know. You can also feel free to follow me @joshweissbock (my personal account), and @nuckprospects for the latest in Canucks prospects news and stats. 

For now make sure you’ve your snacks handy, settle in, and continue past the jump for an extensive report filled with all sorts of interesting little analytical nuggets. 

Cam Charron posted this image towards the end of April which was inspired by the guys at @JapersRink; it suggests that the success (or lack thereof) of an NHL’s team can be forecasted by the success of their AHL team. Well, in the last few years the Canucks’ farm team hasn’t had much success and more recently neither has Vancouver. So that checks out. 

We also have the Filipovic Rule, which suggests that the interest of a fanbase in their prospects has an inverse relationship with how well the parent club is currently performing. Once again, that checks out with the Canucks. With the parent club having their worst campaign in recent memory, a lot of attention shifted to how the kids were doing in Utica with the Comets.

While the Comets may’ve deceived some casual fans with their final record, there are some layers there that need to be taken into consideration. The made an admirable last-ditch effort for a playoff spot after a disastrous start to their inaugural season, and a lot of it had to do with some underlying numbers that we’d been harping on in the weekly reports right from the get-go.

The Comets ended the year 4 points out of a playoff spot, sitting at 3rd in their division, 10th in the conference and 19th in the league overall. The team posted the above photo for their own year in review, and in addition to it, have their own post on up called “Comets by the numbers“. The biggest thing to take away from that snazzy graphic is the points percentage by month; specifically the first month.  

They definitely improved as the year went on, that much we know. But *why* did they start winning more games as the year went along? What spurred their improved success?

If you’re a fan of this site, or hockey analytics, you probably have a healthy appreciation for the importance of puck possession. Teams that have the puck more often over the long-term are more likely to score, less likely to be scored against, and as a result, have a higher chance of winning games.  

In the NHL we measure possession with Fenwick or Corsi, but in the AHL we don’t have the luxury of detailed play-by-play sheets, which means we have to improvise instead. Thankfully we have smart people like Nick Emptage who looked into the correlation of Shot For% and Fenwick Close and found an r-value of .925. Looking at just the first two periods of each game helps further reduce the impact of score effects that are most common in the third period, when the game is more likely to be out of reach for one of the two teams.

So with all of that in mind, how did the Comets perform? At the end of the regular season the Comets were tied for 15th place in possession with the Hamilton Bulldogs at 49.55%. We can see, by evaluating their PDO, (Sh%+Sv%) they were quite “unlucky” this year with a 98.86%. Most of this has to do with the suppressed save percentage of their goaltenders (.9076), but it surely didn’t help that they only shot 8.10% as a team. If you’re looking to reference these numbers to those from across the league, you can do so right here.

The thing about possession in the AHL is that it can have a lot of wild swings in either direction depending on the moves the parent club makes over the course of the season. If they call up a key contributor, that could result in a pronounced dip. The same holds true in the other direction, if a player is sent down for whatever reason. The AHL is a farm league, and these sorts of moves occur all of the time.  

Understanding that possession is much more volatile given these circumstances, here’s a visual look at the peaks and valleys the Comets experienced themselves:

Near the beginning of the year the Comets weren’t all that good a possession team, and I’d wager to guess that part of this had to do with the team playing together for the first time. By the end of November, Cal O’Reilly joined the lineup, and there was a huge improvement seen around the 25-30 game mark.

From there the Comets were fairly steady until the end of the season, when they saw a number of key players (Corrado, Jensen) called up to Vancouver, while other core members were injured (Tommernes, Stuart, Ferriero).

Overall, it’s hard to take too much issue with this part of the equation, particularly for an expansion-level team that was behind the eight-ball to begin with. It would’ve been a non-issue altogether, if not for..

I should point out that we define a “Quality Start” for a goaltender as one in which he starts the game, and gives his team the best chance to win by posting a by posting greater than league average
in save percentage (.911 as per Raw Charge),
or saving greater than 88.5% of the shots he faced while allowing 2 or fewer goals. I was somewhat surprised
to see that Cannata actually had a higher percentage of QS than
Eriksson considering the latter’s superior overall numbers, but this could be due to a smaller workload.

Looking at the Really Bad Starts (RBS) — the opposite of a QS, where a
goalie posts < 85%, or between 85-87% and allowing 5+ goals —  Eriksson once again looked worse than his backup. 

Bail Outs (BOs) are an interesting concept, which we’re just starting to really look into; it’s the number of games in which your team wins despite the starting goalie posting a RBS. Unsurprisingly, given the talent in the Comets forwards, Utica only had
1 BO all year. That came on March 8th against the Binghamton Senators when the Comets were down 5-1 and ended up winning 6-5.

Let’s dive deeper into the numbers:

I included a couple of trends here to try and help paint a clearer picture of the overall performance. There’s the game-by-game Sv% (which as expected is the
most erratic), the rolling 10-game Sv%, the cumulative average and
the season average.

There’s three unique points I noticed right off the bat which I’d like to point out. The first
is Eriksson’s abysmal start to the year; this was a large part in why the Comets went
0-8-1-1 in their first 10 games, and I’d imagine that the adaption to North American hockey had a lot to do with it. 

About a third of the way into the season Eriksson started to settle into
his groove and figure this here thing out. He tightened up his play, and as a result
his save percentage started increasing steadily with each start. It increased all the way up to .916 at one point, which was pretty impressive given the early season struggles. Unfortunately he once again went through a lull in play as the year went along, and I wondered whether the large number of back-to-back starts he was being asked to handle were impairing his play by tiring him out. After all, he clearly wasn’t used to this large a workload, and his postseason save percentage in 2011-12 also dropped as his number of games played increased.

In reality, though, it’s quite possible we’re dealing with a sample size issue here; maybe he peaked when he was on a hot streak, and towards the end of the year the dip was just a regression back to his norm.

Finally, in the least fancy of all of my stats, I should note that he’s currently sitting at a .806 save %, 10.0 GAA in his NHL career based off of the one relief appearance he made during the 9 goals against game in Anaheim in the middle of the season. Either he’ll improve on those numbers, or we just wasted a whole lot of digital ink for no apparent reason..

We should discuss Cannata’s campaign, but it’s just not as exciting considering he was the backup this season. While I still believe he should’ve probably been in the ECHL, getting more starts to work on his craft, it’s hard to argue with having him eat up the occasional start for the Comets instead of Corbeil.

Cannata started off the year in net and similar to Eriksson he didn’t perform all that well; for both of these
goalies this may have been a result of the team in front of them
learning the system. After the team went through its growing pains, just like Eriksson, he improved his performance for the most part. By the end of the year, Cannata was riding a personal high. 

After the last game, Cannata told reporters that he started playing better when he stopped
focusing on when he was going to start games and rather started focusing on the things he could actually control himself. There may be some truth to this since he did improve over the
year, but it also sounds a tad-bit cliche. While Eriksson will be starting in net next year in Utica once again, Cannata is an
RFA this summer, which means it’ll be interesting to see how the Canucks handle things. Will they re-sign him to be the backup? Will they trade him for a pick? Or will they draft another goalie later in the draft, signalling that Cannata is more bottom-of-bin goaltending depth than anything else? 

I am including Corbeil in this section, but I can’t say anything of importance about
him. He started 2 games, posted a shutout in one, and came into a game in relief in another. We can’t draw any conclusions from this small of a
sample size. That being said, his ECHL work suggests there is much to
be desired; in six games in the CHL he posted a .831, followed up by a .881 and a .911 with Gwinnett and Wheeling. Ouch. 

We now turn our attention to the individual skaters to see how they themselves performed this past year. Above is the main table of your traditional statistics for determining and analyzing offense.  Included are all players who played any sort of significant amount of time with the Comets, but mostly focusing on the Canucks players. I cut a bunch of ECHLers who only played 2 games, since their sample size is not large enough, and let’s be honest you probably don’t care about them on this blog. But don’t you worry, Ludwig Blomstrand made the cut! The players with an (X) are players who are typically identified as the Canucks Prospects.

The stats that are included in this table are: Games Played (GP), Goals (G), Assists (A), Points (P), Points-Per-Game (PPG), Shots on Goal (SOG), Shots On Goal / Game (SOG/G), Shooting Percentage (Sh%), Individual Points Percentage (IPP) and NHL Equivalency Points.

If you have knowledge of analytics you know that shot-based data is always better than goal-based data given that players have more control over shots and that shots provide us with a larger sample size. For that reason my preference is to look at the players with the highest SOG/G. Nicklas Jensen led the Comets in this all season; he had the reputation of not being able to score but that was because he didn’t have any puck luck more than anything else, really. He held a 0% Sh% until January and after that it regressed to a much more normal 10.2%.

Shooting percentage, because of how volatile goals are, will regress for almost every single player over a large enough sample size. Some Comets players had terrible luck this year, especially on the back-end (most notably Biega, Tommernes and Sauve). One player who had a high Sh% that he won’t likely repeat is Pascal Pelletier (Ferriero as well). His 15.9% Sh% allowed him to earn 64 points, which was the second highest total in his career, only behind the 75 points he notched in 4 more games back in 2007-08 with Providence. His high Sh% is also reflected in his lower SOG/G and high IPP.

IPP is Individual Points Percentage and it’s the percentage of points you have compared to goals that have been scored while you were on the ice. Basically, it suggests how important you are to your teams offense.  Pelletier led the team in IPP, with Friesen right behind him, followed by O’Reilly and Ferriero.

Looking at some other Canucks prospects that we’ve yet to mention, Grenier in my opinion had the most unexpected season. He posted good numbers all over: a high SOG/G, a normal Sh%, and a respectable IPP. If not for an injury near the end of the year, his final numbers would look even more impressive. Kellan Lain on the other hand was quiet; he’s a bit trickier to analyze with goal-based statistics since goals simply don’t go in on either end of the ice with him out there. Yann Sauve, meanwhile, posted terrible numbers at both the AHL and NHL level, while Henrik Tommernes surely climbed up the Canucks defense depth charts with his strong campaign.  

Thankfully, the AHL does provide us with on-ice goal data which we can use to derive all kinds of analytics for. The problem with these is that goals suffer from a small sample size which give false interpretations of how good/bad a player truly is. We can look at a season’s worth of data to develop a better idea, but let’s be clear about something — they are not as good as shot-based data. For the AHL it’s all we have, which means that it’s better than nothing.

In the chart above we looked at every player’s Even Strength (ES) On-Ice rel Gf%. This is the goal version of Corsi rel, and allows us to see how the team performed with the player on versus off the ice, at even strength. 

From the leaders we can see Lain leading the pack but a lot of that has to do with the limited number of goals in total. This goes back to what I was saying earlier about how goals just don’t go in the net with him on the ice. This might be suggest he’s good defensively, but it may not necessarily mean that, and it’s not great that he’s not putting up any sort of meaningful points the other way. Andersson also showed off his defensive prowess by being way up the leaderboard.  

Cal O’Reilly, much like on the possession front, provided a much-needed shot in the arm for the Comets two months into the season and I wouldn’t be all that surprised to see him earn a look with the Canucks on their 4th line next year.

Friesen had very positive numbers, especially for a season where he started off with just 2 points in almost 40 games. Jensen was on the slight negative, which could be due to the lack of scoring he saw for half a season or it could be an area of weakness he needs to develop. For whatever it’s worth, John Tortorella praised his defensive play when he was up with the team. But then again, Tortorella was promptly fired, so what does he know?

At the bottom of the list is Sauve, which is not too surprising. Given that he is an RFA this summer I would be really surprised to see him re-signed. At this point he has probably worn out all of the shots a player can reasonably expect to get from one team. Corrado was also surprisingly at the bottom of the list, but honestly, that could very well be because of having been paired up with Sauve all year long.

We can also use the on-ice goal data to help us estimate time-on-ice for players (which really needs to start being provided by the AHL soon!). It might not be perfect, but it’s as good an approximation as we can hope for. The two sets of data seem to sync up for the Comets, with the top players typically being those in the top six: Pelletier, Ferriero, Biega, O’Reilly, Corrado, Sauve, Tommernes and Grenier. Andersson is actually fairly high up, but he just doesn’t have the special teams time to inflate his total ice time. We see down at the bottom of the list is Lain who is typically in the third line type role and right at the bottom are the 4th line plugs, players such as Mallet and ECHL call-ups Kaunisto and Kennedy. 

On the power play the typical first unit was composed of O’Reilly, Ferriero, Pelletier, Grenier and Tommernes. Jensen and Corrado did receive some PP time, but over the year they were not the primary unit players. O’Reilly, Biega, Corrado and Pelletier were the main penalty killers. 

Another great thing we can do with on-ice data is see which other players one is both playing with and against. We can use the ES rel Gf% from earlier, and take the average of them to determine a player’s Quality of Competition (how tough the players are they are facing) and Quality of Teammates (how strong their linemates are). We present the numbers above for each player along with the rank on the team and the combined rank.

Because QoC and QoT are based on on-ice events you need plenty of them for any real statistical analysis.  Players with low games played will have numbers that are a bit wonky, and as a result we won’t focus on them too much since they didn’t spend much time in the AHL. The higher the QoC, the tougher the opponents have been; the lower the QoT the harder production becomes for the player, given the worst quality of linemates. When you have a high QoC with a low QoT you have a difficult deployment while a player with low QoC and high QoT is probably sheltered (unfortunately, at this time we do not have zone start data). 

A reference point: for QoC the league average is 50% with a standard deviation of 2.9%, while for QoT the league average is 48.87% with a standard deviation of 7.12%. Players who have an N/A have not seen any on-ice data and we do not likely care about them as they probably have a low games played in the AHL.

Looking at the players most guys had a QoC within one standard deviation. Welsh had the toughest competition with Corrado and Cal O’Reilly just behind him. At the weak end was Patrick Kennedy which is no surprise as he was an ECHL 4th line filler while Peter Andersson was fairly weak too, probably explaining his high ES rel Gf%.  

In terms of Quality of Teammates, Ferriero had one of the weakest teammates all year which wouldn’t help his production (and he did come out with good offensive numbers, which was a testament to his strong play); Sauve and Corrado followed behind. The evidence is building up that Frank Corrado had one of the toughest deployments on the team, which is a good thing to keep in mind when evaluating his season as a while.  

The best way to put all this data together is with some usage charts!

Usage Charts are composed of four parts. The x-axis is the Quality of Competition (left signifies easy opponents and right means more difficult). The y-axis is Quality of Teammates; when you play with strong players it will bring you to the top but weak opponents drops you. The size of your circle is your ice time and the the darker blue your circle is the better your ES rel Gf% is. You would like to be a big blue circle, most preferably in the bottom right corner. 

Zack Hamill did well in his short stint with Utica. Once again, Cal O’Reilly did really well here. Kellan Lain also looks pretty good in his limited ice time. Jensen managed to stay about even in a tough deployment, while Friesen looks good with average opponents but strong teammates.

With regards to the defensemen, Andersson looks great against tough competition but he does have limited ice time and strong teammates. Tommernes is staying about even with more ice time, weak teammates, and average competition. Sauve looks like a complete tire fire while Corrado isn’t doing too bad with tougher competition, weak teammates and lots of ice time. It is no surprise Biega won defensemen of the year with his huge ice time, about average competition and teammates and a slight positive ES rel Gf%.

And just for fun, a few Comets played in Vancouver this year, so let’s see how they looked with an NHL Player Usage Chart:

Based on all of the data, it’s safe to assume that the early season struggles for the Utica Comets were a combination of a new team that hadn’t really played together before, a goalie that was just getting used to playing in North America, and a slightly below average possession team as whole. Combining that with their 0-8-1-1 start, it’s pretty remarkable that they came as close to the postseason as they did. They were in the mix until the very end, and had they gotten some luck at the beginning of the year or played at their true talent level it’s likely they would have been competing for the Calder Cup.

For Canucks prospects, there’s definitely some positive signs to pick out for a few players. Frankie Corrado didn’t do all too bad in his first full season as a pro, while both Andersson and Tommernes are also looking good as defensive prospects. On the other end, while Nicklas Jensen received all of the attention, it was actually Alex Grenier who was turning the most heads. Friesen had a good end to his year, and Kellan Lain was strong defensively in his limited role.

Come July 1st we will start seeing roster changes both for the prospects, Comets signed players and Canucks players on 2-way contracts. In terms of the veterans I would like to see Pelletier, O’Reilly, DeFazio and Stuart back. But the Comets will also be joined by an influx of Canucks prospects from the college ranks (Zalewski, Costello), Dane Fox will likely be in Utica, and Gaunce could be sent to the AHL if he doesn’t go to the juniors as an Over-Ager (while Shinkaruk is AHL-eligible as well if he doesn’t make the Canucks). 

I definitely learned a lot this year in running this project, from using python for web-scraping, developing AHL analytics and evaluating prospects. I will continue to run with this and will keep posting the prospect reports over the summer (as long as there is news). As I continue to figure out new ways to derive analytics for the AHL I will continue to publish them. Finally, I’d like to thank you all for providing me with an audience all year long. 

If you have any questions or comments feel free to send them my way!