Wednesday, December 13, 2017

Buying NHL Corsi

The statistic Corsi For % has become very popular in NHL circles as a means of measuring player effectiveness while on the ice. It's a form of "shot attempt +/-". CF = even strength shots attempts for while a player is on the ice. CA = even strength shot attempts against while a player is on the ice. CF% = CF/(CF+CA) where 50% means a player is on the ice for an equal number of shots attempts for and against.

As this statistic becomes more popular, we see more teams spending money to acquire players who succeed in this category. So how can we expect a player's CF% to change when he starts a new contract with a new team and how much does it cost? To answer this question, I've taken a sample of 435 players who played an entire season with one team, then started a new contract playing at least 1 whole season with a new team, playing at least 30 games with each. I'm not counting players traded mid season. 

The entire sample averaged just over 49% in the first year with their new team on a new contract, a number we would expect to see at exactly 50%, given that there's an equal number of CFs and CAs in every game. Every shot for is also a shot against for the other team. The sample of players going to a new team had an average CF% below what we expect from the entire population.

75% of players with a CF% greater than or equal to 50% will have a smaller CF% with their new team. Conversely 61% of players with a CF% less than 50% will put up a higher CF% with their new team. In other words, players at both ends of the spectrum trend back towards the mean.

*Players with a CF% over 54% will average 51.3% with their new team
*Players between 52% - 54% will average 49.8% with their new team
*Players between 50% - 52% will average 49.2% with their new team
*Players between 48% -50% will average 49.1% with their new team
*Players between 46% - 48% will average 47.5% with their new team
*Players under 46% will average 43.1% with their new team

We do see some very big swings in some individual players. David Clarkson went from 62% in New Jersey to 43% in Toronto. Jarome Iginla went from a 53% in Boston to a 42% in Colorado. Mike Ribeiro went from 45% in Washington to 54% in Phoenix. 

Buying Corsi

In terms of what it costs to buy Corsi on the free agent market, we can look at the expected Annual Average Salaries of players starting new contracts, expanding the sample to include players returning to the same team. There are 1775 players in the expanded sample (the same player can appear more than once, but different years different contracts), the average expected salary of the entire sample signing new contracts is $2.4M AAV.

*Players with a CF% over 54% earn an avg salary of $3.4M on their new contract
*Players between 52% - 54% earn an avg salary of $2.9M on their new contract
*Players between 50% - 52% earn an avg salary of $2.4M on their new contract
*Players between 48% -50% earn an avg salary of $2.3M on their new contract
*Players between 46% - 48% earn an avg salary of $2.2M on their new contract
*Players under 46% earn an avg salary of $1.5M on their new contract

If we treat all other statistics as being equal, players with a CF%>54% will get roughly a $1M premium. Those under 46% will get a similar sized deduction. Players between 52% - 54% get a $500K bonus, those between 46% and 48% get a $200K deduction. Everything from 52% - 48% does not significantly affect salary.

No comments:

Post a Comment