Possession is a performance metric that has been well tracked, documented and analyzed by those in the hockey analytics community. The simple premise is if you have the puck more, you are more likely to score and less likely to be scored against. Strong possession teams have shown to consistently do well over the long term. While the NHL has a plethora of data, other hockey leagues barely track a small percentage of what the NHL makes available.
Read on to see my attempt at a proxy for possession in the American Hockey League…
As most AHL teams are affiliated with the National Hockey League, the NHL will often place developing young prospects on their AHL farm teams. If we know which teams are strong possession teams, we can likely figure out which NHL teams have better prospects, and in the long run we can guess who has a bright future.
One of the main proxies for possession is Fenwick – the summation of shot attempts for both teams at even strength (less blocked shot), either represented as a differential or a percentage. Teams over 50% are strong possession teams, while teams under are usually struggling. Nicholas Emptage compared the Fenwick Close % to Shots For % of all NHL teams over the last five years and found an r correlation of 0.925. Over a large sample size Shots For % and Fenwick Close % become quite close.
To create a proxy for possession in the AHL, I looked at all games that have been played so far this year (up to 13 November), added up each teams Shots For and Shots Against and calculated their Shots For %. I executed this with a Python script so during the season I can continue to update these numbers.
This means there are some obvious limitations. These are currently based on a small sample size, since most teams are only 12 games into the AHL season. These numbers also include special teams (not just even strength) and don’t take account score effects. So this a very rough proxy for the possession numbers we usually use for the NHL.
|Team||Shots For||Shots Against||Possession|
|San Antonio Rampage||494||409||54.71%|
|Grand Rapids Griffins||414||379||52.21%|
|St. John’s IceCaps||461||450||50.60%|
|Bridgeport Sound Tigers||364||360||50.28%|
|Lake Erie Monsters||344||348||49.71%|
|Oklahoma City Barons||403||454||47.02%|
|Hartford Wolf Pack||370||425||46.54%|
While these numbers are very much likely to change we can see who has been controlling possession in the beginning of the AHL season. Some teams, such as Bridgepoort, likely have poor possession numbers, under a small sample size, due to being blown out after playing three games in 48 hours against a rested home team. Utica was a bit surprising as I expected them to be much worse given their starting results of 2-8-1-1 (but a 96% PDO won’t help).
I hope to update these numbers once every few weeks until I create a site where they are always updated (An AHL BehindTheNet if you would). Eventually I would also like to look at other leagues, including the SHL, SHL-2 and Finnish Liiga as they have some data that could be used for a proxy metric as well.