Corsi has a lot of flaws. First of all, it’s not an accurate measure of possession. Corsi is just shot attempts, so it doesn’t actually measure how often a team has the puck on its stick, or its time in the offensive zone, or any other useful metric like that. Second, all shots aren’t created equal. Corsi treats a feeble wrister from the point with no traffic in front the same as a point-blank one-timer in front. Finally, it doesn’t take into account compete level or chemistry. I’m not sure why people try and use Corsi to evaluate teams.
A metric that EQUALLY weights all shots is not a good metric (I’m looking right at you Corsi and Fenwick). It first ignores our prior that says that shots-that-are-goals contains more information than non-goal-shots. Secondly, it’s not supported by empirical research.
WHEN CORSI ANALYSIS GOES WRONG
(June 2010)
Even strength when the game is tied Corsi numbers only tell a fraction of the story.
It doesn’t account for goaltending or power play or penalty kill or shooting ability or any number of other factors that influence who wins hockey games so using it as a tool for determining which teams or divisions are better is a pointless exercise because on the ice, all those other things matter.
The better tool to use in evaluating which teams or divisions are better is the much simpler and more universally understood statistic known as win-loss records. Win-loss aren’t perfect, but they don’t try to tell me that the Leafs have been better than Montreal since the lockout or that the northwest division is only marginally better than the southeast division.
INTRODUCTION TO CORSI ISSUES
(Nov 2011)
The problem with looking at things on an individual level is that Corsi is inherently a team gage and must be put into context to try to get individual ratings. A player on a good team will be more likely to have a good Corsi than a player on a bad team. A player who plays in offensive situations or against weak opposition will be more likely to have a good Corsi than one who doesn’t (and one can imagine several other reasons a player might have a good Corsi related to the context in which he plays).
In order to remove the bias of the situation a player plays, Corsi is usually only recorded in 5 on 5 situations. Thus power play and penalty kill situations are removed
With an even strength Corsi, I find the most useful corrections (in that they are the biggest) are the team the player plays on and the situation in which the player plays (as measured by zone starts – does he start more shirts in the offensive or defensive zones). These corrections make Corsi a more useful number to compare various players because they better include the context in which the player plays.
Desjardins estimates that about 40% of the game is captured by Corsi analysis. It measures the shots attempted at even strength. This is a strong measure of puck possession and puck position on the ice, but it does not measure a lot. It does not measure special teams play. It does not measure the ability to score goals; it only measures the ability to generate shots. Not all shots are equal. Some shots are better than others and have a better chance of becoming goals. Shot quality is not measured in Corsi analysis. Neither is a particular player’s ability to score (his finishing ability) in a given situation. It also does not include team’s ability to prevent quality shots or to make saves (generally this is goaltending). In order to turn a Corsi rating into a more useful number, this must be taken into account if it is possible. Corrections can be made for some of these effects. Some of these effects are not too dependent upon the individual player involved. The saves percentage while a player is on the ice is strongly dependent upon goaltending and not the individual player involved.
In order to go from shots to goals you merely multiply by the shooting percentage of the player involved. On the team level, as we are talking about all shots when a player is on the ice, shooting percentage is not an individual number; it is the number for all players on the ice when a player is on the ice. How much control does a player have on the shooting percentage of another player? If the player is a good set-up man who can get his teammates into good scoring opportunities, he has some control, but it is generally very limited. On the flip side saves percentage of a given team is not well controlled by any other player on the ice.
Even on an individual level, shooting percentage for a player is not a very repeatable number. There are often wide differences in shooting percentage for an individual player from year to year. These shooting percentage differences are often enough to explain unusually high or low scoring seasons. In fact Corsi is found to be a better, more sustainable number from year to year than points or shooting percentage. It is more consistent from year to year and hence a better measure of the individual contribution of a player. Corsi is an underlying number that does a good job of showing how well a player is playing and it is a persistent measure of his talent.
Corsi analysis is a strong measure of puck possession and puck position. With some context based adjustments it is possible to find individual player values. This number is a useful number for rating players, just as goals or points is. Just as goals or points, nobody in their right mind would claim this number is a “be all and end all” statistic to rank players. Much of the resistance to Corsi analysis comes from the fact that people do not understand it. They often expect it to be a “be all and end all” statistic and criticize it when it fails, as it is expected to. It gives useful information about how players are playing. With this information we can better assess players than without it.
8 REASONS I DON’T LIKE CORSI/FENWICK
(June 2012)
· Look at the list of players with the top on-ice shooting percentage over the past 5 seasons and compare it to the list of players with the top corsi for per 20 minutes of ice time and you’ll find that the shooting percentage list is far more representative of top offensive players than the top corsi for list.
· Shooting percentage is a talent and is sustainable and three year shooting percentage is as good a predictor of the following 2 seasons goal scoring rates as 3 year fenwick rates and 3 year goal rates are a far better predictor.
· one year GF20 is on average as good a predictor of the following seasons GF20 as FF20 is as a predictor of the following seasons FF20 so with even just one full season of data goal rates are as good a metric of offensive talent as fenwick rate is. Only when the sample size is less than one season (and for almost all NHL regulars we have at least a seasons worth of data) is fenwick rate a better metric for evaluating offensive talent.
· Although difficult to identify, I believe I have shown players can suppress opposition shooting percentage.
· Zone starts affect shots/corsi/fenwick stats significantly more than they affect goal stats thus the non-adjusted shot/corsi/fenwick data are less useful than the non-adjusted goal data.
· Although not specifically a beef with Corsi, much of the corsi analysis currently being done does not split out offensive corsi and defensive corsi but rather looks at them as a percentage or as a +/- differential. I believe this is a poor way of doing analysis because it really is useful to know whether a player is good because he produces a lot of offense or whether the player is good because he is great defensively. Plus, when evaluating a player offensively we need to consider the offensive capability of his team mates and the defensive capability of his opposition, not the overall ability of those players.
· I have a really hard time believing that 8 of the top 9 corsi % players over the past 5 seasons are Red Wing players because they are all really talented and had nothing to do with the system they play or some other non-individual talent factor.
DECLINING VALUE OF CORSI/FENWICK WITH LARGER SAMPLE SIZE
with more years of data, fenwick becomes significantly less important/valuable while goals and the percentages become more important/valuable.
With increased sample size, the fenwick stats abilitity to predict future fenwick stats diminishes, particularly for fenwick for and fenwick %. All the other stats generally get better with increased sample size, except for shooting percentage which has no predictive power of future shooting percentage.
The increased predictive nature of the goal and percentage stats with increased sample size makes perfect sense as the increased sample size will decrease the random variability of these stats but I have no definitive explanation as to why the fenwick stats can’t maintain their predictive ability with increased sample sizes.
Let’s take a look at how well each statistic correlates with regulation points using various sample sizes
Nothing too surprising there except maybe that team shooting percentage is so poorly correlated with winning because at the individual level it is clear that shooting percentages are highly correlated with goal scoring. It seems apparent from the table above that team save percentage is a significant factor in winning (or as my fellow Leaf fans can attest to, lack of save percentage is a significant factor in losing).
The final table I want to look at is how well a few of the stats are at predicting future regulation time point totals
Regardless of time frame used, past regulation time point totals are the best predictor of future regulation time point totals.
Single season FF% is slightly better at predicting following season regulation point totals but with 2 or more years of data GF% becomes a significantly better predictor as the predictive ability of GF% improves and FF% declines. This makes sense as we earlier observed that increasing sample size improves GF% predictability of future GF% while FF% gets worse and that GF% is more highly correlated with regulation point totals than FF%.
One thing that is clear from the above tables is that defense has been far more important to winning than offense.
Regardless of whether we look at GF60, FF60, or Sh% their level of importance trails their defensive counterpart (GA60, FA60 and Sv%), usually significantly. The defensive stats more highly correlate with winning and are more consistent from year to year. Defense and goaltending wins in the NHL.
What is interesting though is that this largely differs from what we see at the individual level. At the individual level there is much more variation in the offensive stats indicating individual players have more control over the offensive side of the game. This might suggest that team philosophies drive the defensive side of the game (i.e. how defensive minded the team is, the playing style, etc.) but the offensive side of the game is dominated more by the offensive skill level of the individual players
The last takeaway from this analysis is the declining predictive value of fenwick/corsi with increased sample size.
WHY CORSI POSSESSION SHOULD NOT BE END GAME
if teams have been doling out contracts based on, effectively, shooting percentage then it is perfectly reasonable to assume that shooting percentage talent is more evenly distributed across teams than corsi-talent is. Under these circumstances corsi would be highly correlated with winning percentage because that is where the differences lie between teams.
This doesn’t mean that corsi is the main factor in out scoring the opponent though and valuing corsi at the expense of shooting percentage will be a detriment to any General Manager.
Furthermore, if General Managers as a whole started paying primarily for corsi we will start to find that corsi talent becomes more evenly distributed across teams and thus shooting percentage would become much more highly correlated with winning (even after adjusting for luck).
Furthermore, paying players based on corsi would potentially lead to players altering their style of play to optimize their corsi statistics to the detriment of the ultimate goal, out scoring the opponent.
It is certainly possible in the current hockey universe in which players are paid more by shooting percentage than corsi that they play a style of game to optimize shooting percentage at the expense of winning so it is not unreasonable to see the flip side occur of corsi because a metric by which general managers dole out contracts.
Ultimately, the goal of any General Manager is to optimize his line up for out scoring the opposition, not out shooting percentage-ing them and not out corsi-ing them. Corsi or possession should never be considered the goal just as shooting percentage or any other identifiable skill shouldn’t be. The goal has been, is, and alwayss will be out score the opposition and it’s the General Managers job to find the right balance of all the identifiable skills, not just those that seemingly correlate with winning.
LIMITS EVALUATING PLAYERS AND TEAMS
Corsi is quite easy and not very advanced at all; it is simply adding up all shot attempts together, whether they be on net, misses, blocked or goals.
Corsi is sometimes shown as relative to the team, where the player’s Corsi value is subtracted by the team’s value when the player was not on the ice for the same games.
Corsi is a proxy for possession: the differential at even strength seemed to mirrors puck possession dominance.
Corsi can proxy scoring chances reasonably well, although once a larger sample is established Fenwick (Corsi without blocked shots) does hold the strongest relationship.
Corsi also aligns powerfully with puck possession and territorial advantage, which when combined with scoring chances are pretty important parts of the game.
These relationships are why Corsi predicts future goal differentials better than past differentials can. Out scoring your opponent the primary goal and leads to wins, which is why Corsi then also predicts future wins better than past goal differentials or winning percentage.
The numbers can also be split into particular situations. WOWY’s (with or without you) are commonly used to show how certain players drive possession by improving their linemates, while anchors do the opposite.
Limitations
Corsi is simply the number of shot attempts, so it seems intuitively right that corsi leads to shots which leads to goals. But it also seems intuitively obvious that you would have to factor in the quality of the chances, the goaltending and quality of your shooters.
Winning in puck possession and scoring chances is important and will lead to wins but does not encompass the full game.
The largest factors outside of possession and chances are luck (ie: bounces), special teams, and combination of goaltending and shot quality (probably in that order).
Gabriel Desjardins once estimated that about 75% of winning percentage is attributed to a combination of Corsi/Fenwick and luck; this means that by accounting for just those two factors, you can tell a great deal of what is going on… almost -but not quite- everything.
The other major limit to Corsi is the need for contextual nuances. Recently Tyler Dellow stated: a player’s Corsi% can’t be divorced from the context in which he plays. There are many different things that can affect a player’s Corsi; who they line up against (Quality of Competition), who they line up with (Quality of Linemate), which zone they are predominately deployed in (Offensive Zone Starts), and what line they are on (Time On Ice) are the major ones. There are likely other minor ones as well,
Someone being a good Corsi player is helping their team in specific areas of the game and tilting the ice, but this does not guarantee they are a better overall player than someone with a weaker Corsi differential.
So no, Corsi has never told you who the best player is in the league or team. It does tell you who is the best Corsi player though, which is informative and helpful but not everything.
PERSPECTIVE ON POSSESSION
Not all possession shares equal worth. The differences that exist between shot rates and shooting percentages while on the ice add or subtract importance to the minutes you play and in turn, the share of shot attempts you generate. At equal CF%, a first-line player’s minutes will hold more value than a fourth-liner’s due to the simple fact more goals are scored in those minutes.
A different way to look at possession is to examine the expected goal differential based on shooting percentages we can reasonably expect from the quality of the players on the ice. In other words, how rewarding are a player’s minutes at a set possession share?
I compared TOI% with on-ice shooting percentages and percentage of attempts on goal. The latter showed no correlation, along with defencemen’s ability to impact Sh%. Forwards, on the other hand, showed a reasonable relationship:
The correlation is nevertheless strong enough to make informed estimates and bring us closer to real GF% than straight Corsi by plugging F QoT TOI% and F QoC TOI% into the equation.
There has been a lot of progress of late in adjusting shot-based possession for context of usage such as zone starts, QoC and QoT. In particular, @SteveBurtch has been doing tremendous work developing a metric he calls dCorsi, or Delta Corsi. Using multivariate linear regression, he can derive expectations for players’ CF% based on contextual stats. The difference between a player’s actual Corsi and that expected of him based on the situation he plays in is dCorsi.
Not all Corsi is created equal: A first line player who dominates first line competition provides more rewarding minutes than a fourth line player who dominates fourth line competition.
Raw CF% and other shot-based possession metrics require significant context in order to provide reasonable assessments. There is no all-in-one stat in hockey I would trust to measure player quality on its own, and the best method involves looking at things in more ways than just one.
PROBLEMS USING ON-ICE STATS TO RATE PLAYERS
(Aug 18, 2015)
If some players are far more involved in goals and shots for than others, while some are far more prone to make mistakes creating goals against and shots against, their on-ice stats won’t provide a fair and accurate indication of their two-way play. In fact, the on-ice number — created by a single player but also by the 20-plus teammates who shared his ice-time — will hide the actual contribution of the individual in the overall performance of the team. This will be true whether you use goals plus-minus or scoring chances plus-minus, or Corsi% (shots-at-net plus-minus) to rate an individual player.
I’m going to suggest that the individual scoring chance number is a much stronger indicator of a player’s performance and value than the on-ice number. Indeed, the on-ice number, when used as the base for individual player evaluation, is so misleading that, in lawyer lingo, it’s more prejudicial than it is probative.
Simply put: if an on-ice number is the bedrock of your evaluation of an individual player, your work is on shaky ground.
So why are on-ice stats like Corsi% weak base indicators for individual puck possession?
On average, there have been 3.6 contributions to each goal for and (including mistakes by goalies) 2.8 mistakes on each goal against.
Year after year there are more contributions to goals than there are mistakes on goals against.
Why? Because of the nature of attacking hockey. When attackers are exploiting a defensive miscue it’s common for them to rapidly pass the puck, involving more attackers in the sequence to maximize the damage caused by that single miscue. A final quick pass to an open man maximizes the chance of getting a high percentage shot, and such passes tend to bring an extra player into the offensive sequence on many goals, as well as on many scoring chances and shots.
Key takeaway: 40 per cent false positives and negatives when applying on-ice numbers to individual players
The key takeaway from this study is that while five Roth-Irvin plus and minus marks are awarded for all 5-on-5 goals for and against, not all players deserve their plus and minus marks. Indeed, about 40 per cent of the time, Roth-Irvin plus and minus marks are assigned to players who don’t deserve them. This means that the on-ice number for the average player will have 40 per cent false positives and negatives, a high rate of noise.
For each position, centre, winger and defence, the percentage of false positives and negatives was about the same, around 40 percent on average.
While players at each position have the same general rate of false positives and negatives, forwards have a much higher rate of false negatives on defence and a lower rate of false positives on the attack.
Defencemen have a higher rate of false positives on the attack and a lower rate of false negatives on defence.
IS POSSESSION HOCKEY PROVIDING DIMINISHING RETURNS
there has been a smaller spread in team CF% talent which he believes is due to a higher percentage of teams focusing on puck possession hockey.
I want to see if we are seeing a decrease in importance of puck possession hockey and an increase in importance in shooting percentage.
In 200708 to 200910 an 1 point increase in CF% would be expected to translate into a 0.82 point increase in GF% and a -0.05 point decrease in Sh%. In recent years a 1 point increase in CF% can only be expected to increase GF% by 0.42 points and the corresponding drop in Sh% is 0.12.
It is still good to improve your Corsi but we are hitting the point of diminishing returns.
These observations could be a sign that hockey analytics is working. Several years ago analytics showed that puck possession hockey was undervalued by teams and teams are now giving puck possession its proper weight.
Of course it could also mean that teams are starting to give puck possession too much focus and any benefits they are seeing from a better puck possession game is getting cancelled out by worse shooting percentages (and possibly save percentages).
Teams really need to seek to improve puck possession with as minimal an impact on the percentages as possible. We have seen several teams improve Corsi in recent years only to see their shooting percentage drop significantly.
As I wrote before, Corsi and puck possession should not be the goal. Scoring and preventing goals should.
Comments