RotoWire Partners

Draft Kit: The Corsi Line and QoC

Mike Wilson

Mike Wilson writes about fantasy sports for RotoWire.

Dan Waldner

Dan Waldner covers hockey for RotoWire, and has been involved in fantasy hockey pools for 15 years. He's a lifetime Toronto Maple Leafs fan, a passion his wife puts up with and his daughter is starting to emulate.

The Corsi Number and Quality of Competition: Looking past Goals and Assists

Ever since the release of Moneyball - first published in 2003, then released into the subsequent motion picture - the term sabermetrics has become common in sports vernacular. Ironically, it's not even limited to baseball anymore, where it was originally named after the Society for American Baseball Research (SABR), who developed it. Now, there's sabermetrics for every sport - making the term synonymous for advanced statistical analysis.

The structure of baseball is a statistician's dream. The pitcher throws the ball to the catcher, and the batter tries his best to hit the ball. The outcome of each pitch is fairly limited: the batter hits it - foul, out, single, double, triple, home run - or the batter does not - strike or ball. Once the play is complete, the pitcher receives the ball, and we go through the whole process all over again. This repetition occurs, literally, hundreds of times a game. At the end of a season, a player will likely have had 600 at-bats and, roughly, 2500 pitches. 2500 repetitions of roughly the same event, for every player, on which, a somewhat accurate picture can be painted using probabilities and observations.

Hockey is a much more difficult sport to model. Once the puck has been dropped, players scatter. There is no repetition in the baseball sense - players don't all take shots in the same location, with similar circumstances, with predictable outcomes. It's a fluid sport with real-time ebbs and flows. The same style of statistics that works so well in baseball becomes almost meaningless in a game where intangibles like grit and tenacity play so strongly in the outcome on the scoreboard.

Fortunately, the business of hockey sabermetrics has been developing steadily, albeit largely in the margins of score sheets. Coaches, GMs, and amateurs alike have been searching for various metrics to determine how effective a player is, purely using numbers. Finding a successful meter-stick is the holy grail of hockey statisticians; such a tool would have an extraordinary impact - poolies and bookies could determine the probability of successful season performance (barring injury), scouts could analyze a player without necessarily watching them, coaches could determine who best to play in what situation, and GMs, who best to acquire and who to pay the big bucks to keep.

While there's been no revolution as of yet, a number of prevalent metrics have been formulated that give insights beyond simply goals and assists.

Jim Corsi and his eponymous number

At some point during his tenure as goalie coach in Buffalo, Jim Corsi noticed a trend: his team did better when they outshot their opponents, and the more effective players on the Sabres got more shots on the opposing team's net than they let get shot on their own. In fact, it didn't even matter if the shots hit the net - it was simply the opportunity to possibly score that was important. As a result of this, Corsi started to record these opportunities, and developed a differential, expressed as a rate per 60 minutes of ice time. As this number started to become more known - and accepted as a useful and viable "Sabre"-metric, if you will - it would take his name: the Corsi Number.

The chart below shows the players with the top five Corsi Numbers in 2012-2013, who played most of the strike-shortened season:

Player GP G A +/- PIM PPP Shots Corsi
Justin Williams RW, LAK 48 11 22 15 22 6 142 29.8
Jake Muzzin D, LAK 45 7 9 16 35 7 77 27.6
Patrice Bergeron C, BOS 42 10 22 24 18 4 125 26.6
Anze Kopitar C, LAK 47 10 32 14 16 16 98 25.4
Tyler Seguin C, BOS 48 16 16 23 16 6 161 25.0

The interesting thing to notice is the similarity in the players that are listed. When extrapolated for a full 82-game season, these players are all in the mid-50s in scoring, with +25-+30 ratings, 30-40 PIMs, in the teens for PPP, and a healthy dose of shots. This can be directly linked to the fact that, while they are on the ice, they are the ones getting the preponderance of chances - not their opponents. Over the long term, that can only be a net benefit to the player's stats.

So, what does this tell us for fantasy hockey? Presuming your league is more than a point-aggregator pool, and measures all the stat categories listed above, high Corsi Number players are steals later on down the draft. These players provide consistency across most categories, and as long as you pick up a few responsible penalty minute players - perhaps adding some grit in the middle rounds, as we suggested in our earlier article - these players in the late rounds can really pad the trickier +/- category, while contributing all-around. While we are definitely not suggesting that the listed players above will drop to late rounds (most will go quite early), the following players will go very late (if at all), and their Corsi Numbers suggest they will be diamonds in the rough:

Player GP G A +/- PIM PPP Shots Corsi Y! Rank
Patrick Wiercioch D, OTT 42 5 14 9 39 10 81 19.2 198
Anton Stralman D, NYR 48 4 3 14 16 0 66 14.6 548
Carl Hagelin LW, NYR 48 10 14 10 18 1 132 11.7 207
Justin Abdelkader LW, DET 48 10 3 6 34 0 96 11.7 251
Viktor Stalberg LW, NAS 47 9 14 16 25 2 113 10.7 361

Quality of Competition

As more and more teams started to look at opportunity differential thanks to Jim Corsi and the Buffalo Sabres, the debate grew over how well the metric represented the quality of the player. For example, one of our favorite players, Jay McClement, was almost single-handedly responsible for the defensive improvement that the Toronto Maple Leafs experienced last season. He was on the ice for 176 minutes of PK - 1st in the NHL, and fully two-thirds of all PK time for the Leafs last season. Before he arrived, the Leafs were 28th in PK effectiveness; last year, they finished 2nd. While his season on paper was less than stellar (48GP 8G 9A +0 11PIM 48SOG), his impact on the team was undeniable, and the first to say so would be Coach Randy Carlyle.

So, what does Corsi have to say about Jay McClement? He finished last season with the worst Corsi Number in the NHL, -31.7. How is this possible?  When you think about it, it's quite reasonable, really. Low Corsi Numbers indicate more chances against than for, and for players who are in a shutdown role, the expectation is that their number will be low. But, as a compliment to that number, one should examine who that person is playing against - the Quality of Competition (QoC). And, if one should examine the Corsi Number of the opponents who that player plays against weighted by time on the ice, you'll end up with the Corsi QoC. A player who has a high Corsi QoC plays more often against players with high Corsi Numbers, and as such, is often in a shutdown role.

Here's a short list of the best Corsi QoC in the NHL for the 2012-2013 Season:

Player GP G A +/- PIM PPP Shots Corsi Number Corsi QoC
Nikolai Kulemin RW, TOR 48 7 16 -5 22 0 72 -22.4 3.4
Dion Phaneuf D, TOR 48 9 19 -4 65 16 88 -18.2 3.2
Nathan Gerbe C, BUF 42 5 5 -3 14 0 64 -17.8 2.9
Martin Hanzal C, PHX 39 11 12 2 24 7 93 3.9 2.6
Jay McClement C, TOR 48 8 9 0 11 0 48 -31.7 2.5

So, again, what does this new metric tell us about fantasy hockey? Two things really.

First, avoid players with high Corsi QoC numbers. These players play against the other team's best players, and unless your pool offers bizarre metrics like Blocked Shots or PK time, these players are just pits for your +/- category, and offer little tangible returns. Jay McClement is a wonderful, skilled, hard-working player to be admired, but from a fantasy GM's point of view, he's a player you should never own. Be especially mindful of players that appear to be gems in here - such as Phaneuf and Hanzal. Phaneuf's Corsi QoC is the second-highest in the league, which means that any mistake that he makes will likely get picked up by the best players on the other team. Anybody who remembers Game 4 of the Boston-Toronto series last year will know full well what we're talking about.

Second, if you can find a player with a high Corsi Number and a high Corsi QoC, one can expect to see a list of players that get a lot of opportunities against players that are likely not in shutdown roles. In other words, people who are statistically primed for success.

To illustrate, we tallied the table below of the percentages of 2012-2013 league leaders in both Corsi Number and Corsi QoC:

Player Corsi Number Corsi QoC Corsi % Corsi QoC % Average
Daniel Sedin LW, VAN 24.71 0.031 83.0% 0.9% 83.46%
Henrik Sedin C, VAN 22.84 0.183 76.7% 5.4% 79.43%
Alex Burrows RW, VAN 16.92 0.246 56.8% 7.3% 60.47%
Pavel Datsyuk C, DET 15.48 0.412 52.0% 12.2% 58.09%
Chris Kunitz LW, PIT 12.62 0.701 42.4% 20.7% 52.76%

NOTE: The Corsi % and Corsi QoC % are calculated as a percentage of the league leaders - 29.77 and 3.381, respectively, and we've eliminated all players from this list that have negative Corsi Numbers or Corsi QoC values.

Looking at these players, nobody should be surprised that these players will go early in many of your drafts. What is surprising, however, is that the Sedin line is so strongly represented in this table, given their weaker-than-expected year for them. This would indicate that, given a similar treatment this year, all members of this line should bounce back well.

At the end of the day, as you scan through your list of sleepers and keepers, franchise snipers, and gritty forwards, those GMs with the most tools at their disposal will be best equipped to make the right decisions, both at the draft, and later, when evaluating trade proposals. Keeping these numbers in mind will help you gain some additional insight into the player and their performance that the traditional player summary does not.