Basic corsi definition:
“Corsi” is fancy stat lingo for all shot attempts (on net, missed, or even blocked) minus any shot attempts against. Therefore X player’s corsi will be all the shot attempts they are on the ice for minus all the shot attempts against they are on the ice for.
Why corsi matters:
– The fancy stat people have determined that puck possession leads to winning and that corsi is an excellent proxy for possession;
– X player’s corsi shows how much X player’s team is in possession of the puck while X player is on the ice;
– Team corsi is simply the team’s total shot attempts for minus the shot attempts against.
Where to find it:
Extra Skater is probably the easiest way to get a quick corsi count for players. The website even tracks shot attempts live during games – select “Games“ at the top of the main page. You can also view corsi stats by “Players“ to view overall performance or by “Teams“(click on your team and scroll down for on-ice stats) to compare players on the same team. Note that generally, linemates will tend to have similar corsi numbers because they play on-ice at the same time. To correct for this see WOWY stats below.
Acronyms and meanings:
– CA = Corsi Against = total shot attempts directed against players teams net;
– CF = Corsi For = total shot attempts directed to opponents net;
– CF% = (Corsi For) / (Total Corsi) = percentage of Corsi for;
– CF% rel = The difference between a players corsi and that of the teams’ corsi while the player is not on the ice (a negative number means the team does better without the player on the ice).
Best time to use corsi:
– To compare team line and player performance in-game;
– To compare players (to do this more accurately see WOWY stats);
– To compare teams’ performances during a game. This is similar to looking at shots on net but purports to be a more accurate statistic in terms of indicating possession.
Correcting for accuracy:
1. Different play scenarios: Because hockey is played differently in a variety of scenarios (e.g. not aggressive with big lead in the 3rd period), the fancy stat types have created a number of corsi scenarios to try and keep this stat more accurate. For example, only counting corsi while playing even strength or when the score is close. Typical scenarios include 5v5, 5v5 close (when the score of the game is within a point in the 1st or 2nd periods), 5v5 tied.
2. Quality of other players on the ice: This stat generally levels the playing field a bit to better compare players to one another by taking into account the quality of opponent players on the ice. Not all sites show this but can be found at Behind the Net and the acronym is Corsi QoC or Corsi Rel QoC when relative corsi is the measure used.
3. With or Without You (WOWY): WOWY stats show corsi numbers for a player with and without each other player on his team (or opposing team). To view WOWY stats go to Hockey Analysis under the heading “Players” (choose the desired player, then scenario and year). This is a great way to see if one of the players on a line is really responsible for all the good corsi stats and who on his line is benefiting from this. Be careful to check the Time-On-Ice (TOI) numbers when looking at this to make sure you have a large enough sample size to compare the two scenarios. E.g. if a player has only played with another player for 5 minutes, this will not be an accurate measure of their corsi together.
4. Fenwick: Fenwick is the same thing as corsi without counting blocked shots. This tends to give slightly better correlations with winning then corsi. The acronyms for fenwick stats are FA, FF and FF% (fenwick against, for and for percentage respectively). Extra Skater tracks this nicely during games using a graph. This stat has been shown to predict making the playoffs but not necessarily playoff wins.
Predictions:
Corsi is unfortunately not a great way to do short-term forecasting in terms of how teams perform. A better measure would be using a combination of stats for your team such as PDO and fenwick numbers. Generally, corsi is a really great tool in removing bias and determining who was really outperforming who during a game as well as being used at about halfway through the season to predict (so longer term forecasting). See this article here for more details on when corsi starts to show accurate predictions, here for details about confusion using corsi as a predictor, and finally here for the best type of corsi scenario for predictions.