Which Marlies goaltender should start Game 7?

160514-sm-13-TSG_9534
Photo Credit: Christian Bonin/TSGPhoto.com

In Game 6 against the Albany Devils, the Toronto Marlies decided to continue to run with Antoine Bibeau. Bibeau allowed four goals on 28 shots, causing the Marlies to lose 4-1. This sends the series back to Toronto tied 3-3, with Game 7 on Monday the 16th. 

There’s no question a lack of scoring has plagued the Marlies in portions of this series. This was very apparent in Games 1 and 6, where the Marlies were only able to score one goal in each game. This is in contrast to Games Four and Five where they tore off for seven and five goals in those two games. 

The Devils play a trap style defense with a roster of AHL veterans who can shut down the dynamic offense of the Marlies pretty effectively. Because of this, it’s imperative that the Marlies get the best goaltending they possibly can. Goals for will be rare, so you have to do everything you can to limit goals against. Is starting Bibeau the way to accomplish that? I’m here to answer that question.

Keefe has undoubtedly seen Bibeau as the preferable option so far, starting him five of the six games in this series, and 7 of the Marlies’ 9 games so far. The other two starts have gone to Garret Sparks. Wondering why Sparks hasn’t gotten as much favour in the AHL playoffs led me to diving into this post. 

One stat that gets touted is Bibeau’s splits against Albany, where he went 4-0 in the regular season. He had a mixed bag of results, with two starts of only 90% save percentage, with other two being shutouts. That’s fine, and all, but splits stats are stupid.

There’s no reason to believe success against a particular team in the past is in any way predictive of future performance. Four games are such an incredibly small sample size that it’s worthless for real analysis.

So let’s do some of that real analysis and see how the goaltending situation shakes out.

GOALS SAVED ABOVE AVERAGE

Before we start, we all know that goaltenders are voodoo. There’s been very little progress made in the realm of analyzing goaltenders. As such, it’s incredibly hard to explain this position, but there are a few metrics that can be utilized. Unfortunately, most of the fascinating stuff that does exist exists only at the NHL level. So, to analyze Bibeau and Sparks at the AHL level, we’ll have to rely on the basic stats that exist for the AHL.

It would be far too easy to say “Sparks has a higher save percentage over the season, so he should start.” I would be remiss if I were to leave it at that. While there is some merit to simply looking at save percentage, we can make use of another stat.

This post from In Goal Magazine by Greg Balloch (@GregBalloch on Twitter dot com) describes a stat called GSAA – Goals Saved Above Average. It approaches Saves like WAR (Wins Above Replacement) in baseball. The basics of it are that it looks at the shots faced by a goaltender, how many they stopped, and compares that to league average. This tabulates an amount of goals that that goaltender saved compared to the league average. The problem with this is it very clearly favours goaltenders who played more. Since Bibeau played in almost double the games that Sparks did, directly comparing this total is unfair.

I went to the trouble of calculating GSAA for the two Marlies goaltenders for this season, including playoffs. I then divided this by their games played to get a Goals Saved Above Average per Game. This should provide a more balanced version of GSAA. The league average save percentage (calculated by me) is 90.90%. Let’s see how Bibeau and Sparks fared.

Goalie GP SV% GSAA/G
Antoine Bibeau 47 91.10 0.053
Garret Sparks 24 93.20 0.69

Stats include playoffs, found on www.ahl.prospect-stats.com and www.theahl.com

This means that over the course of a seven game series, Sparks would have saved 4.46 goals more than Bibeau. The Marlies have scored six more goals than the Devils in this series, and yet they are still tied. That 4.46 goal difference may have been enough for a win here or there to close it out in 5 or 6 games. Instead, the Marlies head to Game 7 with a team they eclipse in offensive skill. It’s not outrageous to suggest that an improvement in goal could have allowed the Marlies to show their dominance over Albany.

QUALITY STARTS

Another stat we can use for goaltenders is Quality Start Percentage. This is a stat developed by Rob Vollman and is explained in this Lighthouse Hockey post. The basic idea is that if your goalie has an above average save percentage in a game, or if he has higher-than-replacement-level save percentage in a game where they allowed fewer than two goals against, he played a good game, and your team has a high probability of winning. That post displays that, by their definition of the numbers for a Quality Start, a team has a 75% chance of winning the game when their goaltender posts a quality start.

The numbers Vollman uses are 91.2% for league average save percentage, and 88.5% for replacement level goaltending. We showed above that league average in the AHL for the 2015-16 season was 90.9%. Given that I’m not sure how they calculate replacement level goaltending, we’ll adjust the replacement level goaltending by the same ratio as the average goaltending number decreased. This gives us a replacement level save percentage of 88.2%. Realistically, this won’t affect the data at all, unless you got into ~60-70 shots against games.

Once again, I’ve gone through the trouble of tabulating the quality starts for the two Marlies goaltenders this season, including playoffs.

Goaltender GP QS QS%
Antoine Bibeau 47 27 57.4
Garret Sparks 24 14 58.3

Stats from www.theahl.com

We see here that Bibeau very nearly matched Sparks for Quality Starts. Bibeau benefited substantially from the inclusion of 88.2%+ starts if you only allow two goals. Additionally, in the playoffs, Bibeau has made five quality starts in 7 games of the playoffs so far. Perhaps some of his criticism is unwarranted. 

We can draw from this that both goaltenders have a nearly equal expectation of achieving a Quality Start. This makes it not very valuable for analysis, but an important item to check off the list for our analysis.

CONCLUSION

The conclusion here is two-fold. One, Garret Sparks had a significantly better overall performance in this AHL season. Second, there an equal likelihood that Sparks or Bibeau will post a “Quality Start” when they play.

When choosing a goaltender for Game 7, I would lean towards Sparks. Based on the data above, and the higher level experience Sparks has had in the NHL, I’d be comfortable with this choice ten times out of 10. But, given Bibeau’s QS%, it won’t be a coaching disgrace to put him in there given how he’s played in these playoffs so far.

  • Mitch92

    Bibeau had his chance to close out the series and instead put his team behind early and they couldn’t come back. So perhaps it is time to give Sparks another look? He has played great since returning from the Leafs where he unfairly was deemed subpar given that he was not provided an adequate amount of team support most nights. Sometimes a change in goal can provide a jump start which is what the Marlies need heading into game seven.

  • DragLikePull

    Good analysis. Strongly agree that Bibeau should get the boot. He’s getting way too much hype for being “clutch” or “good at hockey” from that memorial cup win, even though his AHL stats have been lackluster.

    If you look at the top goalie Sv% over the past two years, Sparks is in 3rd behind Matt Murray and Jacob Markström. Bibeau is in 49th.

    One thing I noticed is your replacement level save percentage is a bit low though. Consider that not a single goaltender in the AHL had that a save percentage below .890 with min 1440 Minutes played. I’d say it makes more sense to define the top 30 goalies in SV% as your starters and create a bin for the rest of the goalies as replacement level. This gives you a replacement SV% of .901.

  • Mitch92

    There is the confounding factor that Bibeau was terrible in the first half of of the season, then quite good in the second half. So I don’t know if using his whole season stats is completely fair.

    That said, I would have liked to have seen Sparks get an extra start earlier in the series do it’d be a little clearer who’s playing better at the moment..