Making Sense of Advanced Stats

Hockey has been a sport that has long been dominated by seeing how players look on the ice. If a player is flashy and can score beautiful goals, we often look at them as more valuable to a team. After all, a team can only be successful if it can score more goals than its opponents. However, not every player looks good every night. Some go through random spikes of goal scoring then remarkably go quiet for a number of games. Some look brutal one season, but return with a vengeance the next. How can we best understand this?

One way is to look at the underlying numbers of a player or team to see what is happening. Some numbers are easier to understand than others. For example, if a player is averaging one shot per game, you cannot expect reasonably that they will score on every one of those shots. If a player is playing upwards of 20 minutes a night, you can expect they will put up more points than a player playing less.

Here at The Win Column, we like to look a little deeper than that. We believe that underlying numbers on a player are the best indicator of sustained success, or sustained struggles. In each of our About Last Night articles posted after each Flames game, we look at a few different metrics that we will outline here.

5v5 Score and venue adjustments

The metrics we use to get a picture of game flow are generally at 5v5 SVA. This stat basically measures how players do at full strength (taking out power play and penalty kill) and adjusted for the score and venue. The theory behind this metric is that teams with a lead take less shots depending on how many goals they are up. This has been quite definitively shown as seen in the chart below from TSN.

This adjustment accounts for the score of the game and shows more accurately how teams are doing through a game. If a team is up, it adjusts their performance up by a factor. If they are down by a goal, it adjusts their performance down because they are likely pressing more to try to tie the game. If a team is up by a goal and their performance has been boosted by that factor and their Corsi for in that period is still below 50%, the opposition is pressing more heavily than a team in that position normally would. This gives you a clearer indication of the flow of the game than if you were looking at the numbers purely on a 5v5 basis.

This metric also adjusts for if a team is playing at home or away. The term home ice advantage is very real, and can be quantified. Whether it is because of the comfort of a home-cooked meal before the game, the crowd cheering the players on, or something else, teams at home tend to shoot more than when they are on the road. The math is explained in this article Micah Blake McCurdy (@ineffectivemath on Twitter), and is used to adjust the performance of each team relative to whether they play at home.

Corsi For Percentage (CF%)

Most simply, Corsi for counts the number of shots towards the net by a team. This includes shots on net, missed shots (including shots that hit the post or crossbar), and blocked shots. Each of these is counted as a Corsi event, and is calculated both as a raw number (number of total events by a team) as well as the percentage of events by each team in a game.

The reason people look at Corsi for as opposed to just shots on net is it more accurately shows the flow of the game. For example, if a team has heavy offensive presence and directs five shots on the net, but two bounce off the post, two are blocked, and one goes wide, none of those will count as a shot on net. If you only look at shots on goal, it does not show the other offensive chances at all, even though some were probably even more dangerous than the ones that did hit the net, and can be misleading.

As well, if the opposing team takes a number of shots on net, but they are all from outside and hit the goalie square in the crest, they are all counted as shots on goal, but none of them had any chance of going in the net. This skews the results and makes it seem like one team controlled the game, when in reality it was more mixed.

Corsi events help us to better understand the flow of the game.

SCoring Chances For Percentage (SCF%)

Scoring chances for is exactly what the name suggests. The number of real scoring chances a team has over the course of the game. This takes Corsi events and subtracts any shot from outside a team’s offensive zone.

In this example from WAR on Ice, shots from the yellow zone indicate low danger chances, shots from the pink zone indicate medium danger chances, and shots from the blue zone indicate high danger chances. In calculating SCF, shots from the yellow zone are assigned a score of one, pink assigned a score of two, and blue assigned a score of three. Rush chances and rebounds are assigned an additional point. A blocked shot lose one point. Scoring chances are counted as any event that is above a score of two.

For example, only rebounds or rush chances from the yellow zone that hit the net would be counted as scoring chances. Any unblocked shots from the pink zone, a blocked rebound or rush chance from the pink zone, or any shot from the blue zone would also count as scoring chances.

High Danger Chances for Percentage (HDCF%)

Just as scoring chances for is calculated as chances with a score of two, high danger chances for are calculated as chances with a score of three. This is an unblocked chance from within the blue zone above, a blocked rebound or rush chance in the blue zone, or an unblocked rebound or rush chance from the pink zone. These are typically the most difficult shots for a goalie to save or a team to defend and are most likely to go in.

Scratching the Surface

Hockey advanced stats go much, much further than the three metrics discussed above. In fact, these are some of the oldest metrics people use, and there are a lot more advanced ones that have become popular recently including expected goals and goals above replacement models just to name a couple.

Advanced stats don’t paint the whole picture, but neither does just using the eye-test. To properly analyze players and teams, a combination of both is usually needed. Hopefully this provides a little bit of insight into some of those basic advanced stats.

One thought on “Making Sense of Advanced Stats

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s