How can we measure a goalie’s rebound control? Examining Pekka Rinne and James Reimer.

Embed from Getty Images

Pekka Rinne is good at controlling his rebounds. I know this, because people on the internet have made their opinions abundantly clear. Scouts and fanalysts alike credit Rinne’s quick glove hand with helping him catch a significantly higher volume of shots than most other goalies, leaving few opportunities behind for lurking opponents to deposit into his net.

James Reimer is not good at controlling his rebounds. I know this, once again, because people on the internet have made their opinions abundantly clear. Reimer’s (supposed) inability to prevent the shots he’s saved from bouncing into dangerous areas is often cited as one of the main reasons for why he should be the #2 goalie behind Jonathan Bernier on the Leafs’ depth chart.

Goaltenders are notoriously difficult to analyze, however rebound control seems to be one aspect of the position that everyone feels educated enough to opine on. While it may be more difficult to analyze a goalie’s positioning or reflexes, rebound control is an area where the results are immediately visible even to the novice observer. For better or for worse, the average fan tends to believe they have a good handle on which goalies are adept at keeping the puck away from their opponents post-save, and which netminders might as well be playing for the other team.

In spite of the abundance of opinions on rebound control, most of the assessments that we see tend to be quite qualitative, focussing on observations over data-driven analyses. This is in spite of the fact that past efforts to quantify rebound control using data from the NHL play-by-play files have shown that there are elements of talent that can be measured in rebound control – Rob Pettapiece (now of the Toronto Maple Leafs) wrote a great piece at NHL Numbers looking at Rinne otherworldly skill in preventing rebounds, while I followed up his article with my own version of a rebound control statistic which came to similar conclusions. Both of these pieces are now relatively old, however, and I’ve had a few new ideas come to me in the meantime, and as such a more comprehensive look at rebound control.

There are 3 distinct ways that a goalie can display good rebound control. The first way is to prevent the opportunity for a rebound all together, by freezing the puck. We can measure this ability by calculating the percentage of saved shots that a goalie is able to freeze within 2 seconds (i.e. number of stoppages within <= 2 seconds following a save). We’ll call this a goalies Frozen Shot Percentage, and we’d expect, given the work that’s been done to date, for this to be a persistent talent year-to-year.

This is also the metric where we’d expect Pekka Rinne to do the best, and, if the rumours are true, James Reimer to do the worst. So what do the results look like?

Season Pekka Rinne James Reimer
Frozen Shot % Rank Frozen Shot % Rank
20102011 36.5% 1/36 34.6% 5/36
20112012 37.9% 1/41 27.2% 33/41
20122013 34.1% 2/24 32.0% 5/24
20132014 37.7% 32.8% 4/42
20142015 37.6% 2/44 32.8% 11/44

It appears as if the legends of the Finnish netminder’s glove hand are true – Rinne has finished in the top 2 in the league in 4 of the last 5 seasons (the ranks above are for goalies with at least 500 5v5 shots against, so he didn’t qualify in 20132014), while freezing at least 5% more shots than an average goalie in each year over that period (the league mean from 2009-2015 was 29%).

For James Reimer, however, the story is a bit different – Reimer’s puck freezing ability hasn’t been nearly as bad as many Leafs fans and analysts would have you believe, and he’s been above average in most of his seasons to date. Even including his forgettable 2011-2012 season, where he struggled with injuries including missing 19 games with what many have speculated was a concussion, Reimer places in the top 5 amongst all goalies since 2009, with an overall Frozen Shot Percentage of 34.0%.

Even better news for Reimer is that this remains a very persistent skill for goalies – the year to year correlation for goalies who saves at least 500 shots at 5v5 is 0.58, which is extremely high for a goaltending metric. Put more simply, the percent of shots that a goalie freezes may be the most repeatable talent that a goalie possesses which we can measure. A goalie’s Frozen Shot Percentage in one year is a fairly strong predictor of their Frozen Shot Percentage in the next year, meaning that when we see a goalie’s numbers in a given season, we can trust that they won’t likely too far off that mark in the future.

The second way a goalie can control their rebounds is by deflecting the puck into low (or no) danger zones, where a rebound opportunity for their opponents is unlikely. We can measure this in a way similar to Rob’s original metric, by looking at the percentage of shots that result in a goalie’s opponent getting a rebound shot attempt. The key difference here, however, is that we only want to look at the cases where the possibility of a rebound attempt exists, as we know that when a goalie freezes the puck the potential for a rebound attempt goes down to zero. This allows us to measure it is a distinct ability from freezing the puck, and therefore doesn’t penalize netminders whose playing style leads them to direct pucks rather than catch them. To more accurately measure a goalie’s ability to deflect the puck into low danger zones, we’ll calculate each player’s Adjusted Rebound Percentage, which is simply the total number of rebound shot attempts (shot attempts taken <= 2 seconds following a save) divided by the total number of saves minus the total number of frozen pucks.

Both Pekka Rinne and James Reimer appear to be fairly strong by these metrics as well – Rinne has been in the top 5 twice over the last 5 years, while placing just under the average mark of 5.7% in the other three seasons. Reimer has been similarly strong, although he’s trended more towards the middle of the pack recently, giving up rebounds on more than 6% of his non-frozen saves in two of the past three years.

Season Pekka Rinne James Reimer
Adj. Rebound % Rank Adj. Rebound % Rank
20102011 5.4% 22/42 3.8% 1/42
20112012 3.8% 4/44 4.4% 6/44
20122013 4.7% 3/25 7.0% 21/25
20132014 5.2% 5.5% 16/43
20142015 5.3% 17/46 6.3% 28/46

If the numbers look a bit more random from year-to-year here, it’s because they are. Adjusted Rebound percentage is a much less consistent metric for goaltenders than Frozen Shot Percentage. The season over season correlation is only 0.26, which isn’t entirely shocking – Adjusted Rebound Percentage will be impacted a lot more by outside factors, since each goalie’s total will be influenced in part by the play of the defenders in front of him.

The third and final aspect of good rebound control that we want to look at is a goalie’s ability to keep themselves well positioned so that if a rebound does occur they’re in a good position to make a save. This can be quantified by simply calculating each netminder’s save percentage on rebound shots, or Rebound Save Percentage.

Season Pekka Rinne James Reimer
Rebound Save % Rank Rebound Save % Rank
20102011 72.3% 15/42 70.0% 23/42
20112012 78.9% 9/44 81.0% 6/44
20122013 72.0% 13/25 80.0% 4/25
20132014 66.7% 75.0% 21/43
20142015 72.7% 25/46 73.3% 23/46

By this metric, both goalies look slightly more mediocre. Reimer has been more or less dead in the middle of the league, with the exception of a two-season span where he posted back-to-back 80%+ Rebound Save Percentages. Rinne has been slightly better, although his numbers don’t match his world class freezing figures from above. What’s perhaps important to note here is that for goalies like Reimer and Rinne who have shown strong rebound prevention ability, posting a top-tier Rebound Save Percentage is a lot less critical, because they face fewer of these opportunities in general. Where it becomes more important is for a goalie like Cam Ward, who’s been in the bottom 5 in the league 4 times over the past 6 years. Fortunately for Ward, he’s compensated for his propensity to give up rebounds by placing in the top 10 in rebound save percentage in 5 of those 6 seasons, although it remains a dangerous game that the Hurricanes netminder is playing.

Perhaps the most interesting thing to note though is that rebound save percentage actually shows a modest level of persistence, with the year-to-year correlation coming in at 0.24, just slightly below the figure we noted for Adjusted Rebound Percentage. While this may not seem like a huge number, particularly when compared to the repeatability of a goalie’s Frozen Shot Percentage, the consistency of rebound save percentage is actually significantly higher than it is for All Shot 5v5 Save Percentage. What this suggests is that goalies have the ability to control whether they’re giving up high danger rebounds, or that some netminders are better able to keep their body well positioned to make a save on any subsequent shots.

No matter which method we use to measure rebound control, it’s pretty clear that James Reimer’s struggles in this regard are greatly overstated. While he may not have the elite glove-hand of Pekka Rinne, the Leafs’ 1A netminder still remains one of the better players in the game at freezing the puck or deflecting it out of harm’s way. Although there’s no “one-number” statistic to measure rebound control, it’s clear that each of these components forms an important piece of a goalie’s skillset, and that each metric that we’ve proposed here is at least partially talent. These metrics give us a more granular view of a goalie’s performance, and can help identify the differing ways in which certain goaltenders approach rebound control, and ultimately, how they try to keep the puck out of their net.

Seasonal Goalie Rebound Statistics from 2009-Present are available here. Statistics are 5v5 only. All goaltenders who faced at least 500 5v5 shots are included.

3 thoughts on “How can we measure a goalie’s rebound control? Examining Pekka Rinne and James Reimer.

  1. neat stuff. two things:

    1. would be interesting to see the full rankings. If persistence is that high for some of these , it could be a “skill” worth examining more.

    2. I wonder if there’s a level of defensive talent in “Adjusted Rebound percentage”. If the key component of that metric is ” total number of rebound shot attempts (shot attempts taken <= 2 seconds following a save)" then we dont know if the goalie prevented the rebound shot (by placing the puck away from the danger areas, or if a defender was able to clear the puck from the danger areas, right? That would also, to me, support the stats lack of persistence, since its adding a new variable into the outcome outside of the goalie's control (defensemen ability to clear pucks, basically)

  2. One thing I would like to see is how to maximize the correlation. For example, would it be possible to improve the year to year correlation by adjusting the two second windows or other parameters entering into the formulas?

    While too much fiddling with the formulas to maximize correlation can lead to a problem known as overfitting, the data here far outstrips the complexity of the formulas, so we are a long ways away from that problem.

Leave a Reply

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

You are commenting using your 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