The Flames Should Keep Hiller

I’ve made this point numerous times since rumours emerged that the Flames were shopping Jonas Hiller – he’s the best goalie in the organization right now. I don’t mean in terms of overall value (clearly Ortio and Gillies have higher ceilings). I just mean in terms of a guy who has proven he can stop the puck at the NHL level. Ramo is a clear step below. The kids are unknown commodities.

If the Flames want to make the post-season again this year, getting rid of Hiller is a big gamble. Here’s the proof. 

A narrative amongst the MSM who cover the Flames has emerged that Jonas Hiller is prone to allowing early goals, which is bad for the team and one of the reasons Hartley would like to move on. I’m skeptical of this for a few reasons:

1.) I doubt that this is something goalies can actually control. Goals are fairly random events, so their timing in a game is mostly out of a goalies hands. I’ve never seen any evidence that this is an ability (or persistent weakness). Which is to say, even if Hiller gave up a higher rate of “early goals” last year, that doesn’t necessarily mean it will happen again. 

2.) That said, since nobody tracks “early goal” rates, we don’t even know if Hiller is above average or below average at this from a league perspective. 

As for the deleterious effect something like that might have the team, we can investigate it by comparing a couple of broad brush metrics between Ramo and Hiller: quality starts and win%. 

Quality starts a simply games in which a goalie allows two or less goals or achieves a SV% of .917%. Win% is simply the amount of wins versus games a goalie managed. 

Hiller 2014-15: QS% 52.3%, WIN% 50.0%

Ramo 2014-15: QS% 43.4%, WIN 45.5%

So even if we grant that Hiller does a have a tendency to give up weak, early goals, we can see that he still gives the Flames a better chance to win than Ramo on average. By the way, Ramo’s best QS% in the league so far is just 51.4% – so, still below Hiller’s rate last year (which was his lowest QS rate in his career). 

The reason for this is simple: Hiller stops more pucks than Ramo. 

Here’s every Flames goalie’s adjusted vs. unadjusted save rates from last year:

Hiller had a better save rate)s) than Ramo last year. He also saw more shots per 60 and played way more minutes. Also, that’s Joni Ortio way down in the bottom left corner. He didn’t play enough to really worry about his results, but it’s also not like he tore the cover off the ball when he was in the league either. I like him as a prospect, but we still don’t really know how good he is.

Just to put an exclamation point on things, let’s look at an expected goals model for both guys. This model from Don’t Tell Me About Heart looks at the quality of shots faced by a goalie and then projects an “average” save rate according to those shots. It then looks at the player’s average save rate to see if how he performed against this expected save rate. 

Just for fun, here’s Ortio:

The black, horizontal line in each graph is the player’s save rate. The bar represents his actual save rate. If the bar is below the line, he under performed expected SV%. If it’s above, then he outperformed it.

Again, Hiller is consistently the superior option. In fact, Ramo has frequently under performed his expected save rate by this model. Hiller has consistently outperformed his.

You can do this sort of exercise with various other metrics from various other angles and the answer is always the same: Hiller is better.

The one reason to worry about Hiller is his age. At 33 going on to 34, he’s entering the point in his career where goalies start to drop off. Nevertheless, everything we know about him right now says he’s the best goalie for the starters job this season. 

I don’t know why Bob Hartley doesn’t seem to like Hiller anymore (he played him more than any other Flames goalie last year) and I don’t really know why the Flames bothered to re-sign Ramo, but here we are. If the club decides to waive or trade Hiller at the start of the season, they are making a very sizeable gamble.