A penalty killer has one primary objective: preventing the opposing team from scoring. He also has a secondary objective: scoring on the opposing team. He should never let his secondary objective get in the way of his primary objective.

There are different ways of accomplishing that primary objective: blocking shots, intercepting passes, clearing the zone, preventing the opposition from taking shots to start with, forcing them to take low-quality shots, winning face-offs, and drawing penalties to name a few. And while a PK that shuts down the opposing PP is great, the ability to generate offensive chances while short-handed shouldn't be ignored either.

Even in this age of "advanced stats", some of these are easier to measure than others. However, I hope that advanced stats can at least help us identify which players are preferable to have on a PK to others.

Metrics

Let's identify a few key metrics first. To narrow our field of players, we'll limit our data to the past two seasons (2013-15), looking at skaters with 200+ 4v5 SH minutes. If a player can't reach 200 minutes in two seasons (max 164 games), he probably isn't relied upon as a penalty killer. This gives us 104 forwards and 116 defencemen, for a total of 220 skaters.

Although PK success is measured by results, results aren't coachable. Process is. And process shows up in Corsi, Fenwick, and SOG. So let's have a look at what each can tell us.

CA/FA/SA60 - The player's contribution to raw shot-suppression. The difference between Corsi and Fenwick shows the shot-blocking ability. The different between Fenwick and SOG shows... the player's ability to make the opponent miss the net?

CF/FF/SF60 - The player's contribution to raw shot-generation. Since these will be relatively infrequent on the PK, we'll only consider CF60.

CF/FF/SF% - Since these events are more frequent, Percent-For has more value; however, it can still hide a player's good shot-suppression ability due to his poor shot-generation ability. For example, Maxim Lapierre's CA60 ranked 82nd out of 220, but his CF% ranked 211th due to his 219th ranked CF60.

CF/FF/SF%RelTM - The player's CF/FF/SF% relative to the team. The same pros and cons of GF%RelTM apply here.

CA/FA/SA60RelTM - The player's CA/FA/SA60 relative to the team. The lower the better.

CF/FF/SF60RelTM - The player's CF/FF/SF60 relative to the team. The higher the better.

Ideally, a player should first and foremost limit CA/FA/SA, secondly without hurting CF/FF/SF disproportionately. To identify players who do this well, we'll look for players with negative CA/FA/SA60RelTM, but a positive CF/FF/SF%RelTM. Moreover, the best penalty killers should be able to do this consistently, while playing against high quality opposition and in tough zone starts.

So we'll set the following criteria:

1. CA60RelTM must be negative - Team gives up fewer shot attempts with the player than without.

2. CF%RelTM must be positive - Team's overall play (offence + defence) is not disproportionately impacted by the player's shot-suppression ability.

3. DZFO% must be high - Player takes on the toughest zone starts.

4. Either criterion 1 or 2 above must be true in each season - Performance must be consistent year-over-year for the elite.

The Good

To identify our elite-level penalty killers, we'll pick out players who meet all four criteria: negative CA60RelTM, positive CF%RelTM, and a DZFO% over 80% (or at least within 2 DZFO's of it). That results in this list:

To get a look at merely-great penalty killers, we'll examine two different groups. The first is those who are great shot-suppressors, at the expense of shot-generation (i.e. met first criterion but failed the second):

And the second list is those who are great shot-generators, at the expense of shot-suppression (i.e. met second criterion but failed the first):

There are also players who met the first two criteria in easier circumstances (i.e. failed the third criterion). This suggests either a) they are being utilized in a suitable way, or b) they are ready for a bigger challenge. Here's a list of them (29) who had at least 60% DZFO:

Beyond the most-frequent penalty killers, there is a group of players who can handle penalty-killing duties in specific situations, whether to offer breakaway SHG threats or to handle neutral-zone starts. As a result, they generally log fewer PK minutes, and likely wouldn't be as effective if given top PK duties. Below is a list of 58 players who have logged between 100 and 200 4v5 SH minutes over the past two seasons, still met the first two criteria above, with the third and fourth criteria relaxed or ignored to reflect the reality of their specific, limited, situational usage (and subsequent reduced sample size):

The Bad

Conversely, the same metrics can be used to identify the penalty killers who have most underperformed. We can identify these players by looking for:

1. Positive CA60RelTM - Team gives up more shot attempts with the player than without.

2. Negative CF%RelTM - Any improvement in shot-generation with the player does not sufficiently compensate for the decline in shot-suppression.

Applying these two criteria on the 220 skaters with 200+ 4v5 SH minutes during 2013-15 yielded a list of 95, which is rather large. The nature of RelTM metrics means that there is bound to be at least one player on each team on this list, which isn't exactly what we're looking for. We're not looking for the worst penalty killers on each team; we're looking for the worst across the league.

That means we need more stringent criteria. I propose using the following:

1. GF60RelTM is negative - Team scores fewer goals with the player than without.

2. GA60RelTM is positive - Team gives up more goals with the player than without.

3. SF60RelTM is negative - Team has fewer shots on goal with the player than without.

4. SA60RelTM is positive - Team gives up more shots on goal with the player than without.

5. FF60RelTM is negative - Team has fewer unblocked shot attempts with the player than without.

6. FA60RelTM is positive - Team gives up more unblocked shot attempts with the player than without.

7. CF60RelTM is negative - Team has fewer shot attempts with the player than without.

8. CA60RelTM is positive - Team gives up more shot attempts with the player than without.

These 8 criteria are all variations on the same theme: team offence and team defence are both absolutely worse with the player on the ice, than with him off the ice, in every way that we can measure. Applying them to the list of 220 yields a set of 24:

There are some surprises here: the likes of Anze Kopitar, Zdeno Chara, and Patrick Marleau being detrimental to a team's PK. Kopitar ranked dead last in GF%RelTM because in the past two seasons, over 161 games, he had been on the ice for zero SHGF.

Flames Verdict

Since this is a Flames fanpost, we now turn our attention to the Flames players in the lists above.

Leading the way is captain Mark Giordano, listed above as a great shot-generator on the PK. In fact, aside of Corsi-Against, the Flames are better in every other metric (Goals, Shots, Fenwick, and Corsi) with Giordano on the PK than without him. A healthy Giordano is crucial to the Flames' success.

Next up is Giordano's regular partner, T.J. Brodie, on the list of good PK'ers with easier zone starts. He actually took on tougher zone starts in 2014-15, with 76.5% DZFO over the previous year's 68.9%, while improving in metrics across the board, to the point where he is a positive influence at both ends of the PK by any measure. The hope is that he will continue to improve.

Joining Brodie on the same list is Mikael Backlund. Injury interrupted his season in 2014-15, but his shot-suppression ability remained on display, and he scored 6 of the Flames' 17 SHGs over the past two seasons; only one other Flame had more than 1 (Lee Stempniak). His 6 SHGs is tied for second in the league over 2013-15, with Rick Nash and Brandon Sutter, one behind Brad Marchand.

We find three 2014-15 Flames among the list of PK'ers who had been effective in limited usage. Two of them played 2013-14 with another team (or teams): Mason Raymond with the Toronto Maple Leafs, Raphael Diaz with the Montreal Canadiens, Vancouver Canucks, and New York Rangers. While both had been used on the PK regularly through their respective careers, neither saw any significant PK time with the 2014-15 Flames.

Paul Byron was the third Flame on that list, and one can only hope that he sees more time on the PK in the future. Widely derided for his wastefulness on breakaways, he nevertheless has the highest GF% among the Flames over the past two seasons, and his 5 SH first assists is tied for second in the league over that span, with Patrik Elias and Victor Hedman, one behind Derek Stepan.

All this brings us to the final Flame, and the only one on the naughty list: Lance Bouma. I hope that, for all his eye-catching and heart-warming bravery, the numbers will finally put the debate to rest once and for all: Bouma is not the guy you want as your top penalty-killer. The Flames simply do worse in every measurable way with him on the PK. And not just a bit worse; Bouma is an anchor to the PK. For all his shot-blocking prowess, his SH FA60 remains highest among the Flames. That he continues to receive the most PK time of all Flames forwards is indefensible.

Look, I understand that he gets the toughest assignments on the PK, and somebody has to eat up those tough minutes. But those circumstances do not justify his horrendous numbers. Other forwards must be given a chance to step up on the PK: Byron, Raymond, Michael Frolik, Josh Jooris, even Joe Colborne. David Jones is another player who had previous PK experience (with Colorado Avalanche) but has seen only 6 minutes of PK time since joining the Flames two years ago. Ditto Drew Shore, formerly of the Florida Panthers. Hell, let Sam Bennett have a go.

Just don't try to convince anyone that we have no better option than Lance Bouma.

(All stats courtesy of Puckalytics.com)