Why Bobby Clarke's Offensive Performance in the Playoffs Was Better Than It Looks
I've noticed there is a general tendency to compare one-way players to two-players by comparing their offensive results and then trying to estimate whether the additional defensive impact "makes up the gap". I think this approach undervalues two-way forwards used in a shutdown role. The reason is that a shutdown forward's offensive contributions will almost certainly be more valuable to his team on average (in terms of increasing the chances of actually winning hockey games) than the offensive contributions of a forward who scores the same regardless of whether the score is 8-0 or 0-0. This is because the shutdown forward is only going to be trying his hardest to score in the most high-leverage situations, while trading offence for defence in other scenarios (especially when his team is ahead), meaning that he can't really take advantage of the juicy counterattack chances that tend to result when the other team is pressing to try to catch up on the scoreboard.
From a past analysis, I happened to have situational stats by score for the Philadelphia Flyers with Bernie Parent in net from the Hockey Summary Project box scores, so I went back and filled in a few gaps to get the data for Bobby Clarke's playoff career. Here is how his scoring broke down during his playoff peak from 1973-78 based on the score of the game, along with the minutes his team spent in each score situation, my estimate of Clarke's points per 60 minutes assuming he played one-third of his team's ice time in each situation, and the percentage of Philadelphia goals that Clarke got a point on in each scenario:
Bobby Clarke Situational Playoff Scoring, 1973-78:
| Score | Pts | TOI | Pts/60 | % of Tm |
| Trail by 2+ | 8 | 541.7 | 2.66 | 25% |
| Trail by 1 | 21 | 698.1 | 5.42 | 40% |
| Tied | 27 | 1884.8 | 2.58 | 29% |
| Lead by 1 | 12 | 1032.1 | 2.09 | 24% |
| Lead by 2+ | 9 | 939.9 | 1.72 | 21% |
[TBODY]
[/TBODY]
The main takeaway is that Clarke's numbers are best with the game tied or especially with his team down a goal, but that they drop off significantly with his team already in the lead. That's exactly what we'd expect to see for a strong offensive player being used in a shutdown role.
Did Clarke's scoring patterns differ in the regular season? I don't have the full numbers but I'm confident the answer is yes, considering that just from March 5-26, 1975, he scored 12 points with his team already leading by 2 or more goals (in just 11 GP), even though he only had 15 such points in his entire 136 game playoff career. In other words, while deployed in a standard top line offensive player mode (including being given the opportunity to pad his stats in blowout scenarios against weaker opposition), Clarke had competitive scoring numbers with other elite forwards. The difference is that come playoff time he was used in a different way which lowered his offensive numbers (even though I think more than enough of Fred Shero to be quite confident it didn't actually hurt the Flyers one bit).
If Clarke scored at the same rate with his team leading as he did in all other situations, he would have scored an extra 14 points in those 83 games, improving his PPG from 0.93 to 1.10. That's still not quite at his 1.31 regular season average, but not only did the Flyers face strong playoff opposition on average (as pointed out by Kyle McMahon), it's actually very conservative to assume that a forward would score the same with his team leading as otherwise. Offensive forwards usually score at their best with their teams already in the lead. Take, for example, Clarke's contemporary Phil Esposito:
Playoff points by score:
Peak Esposito, 1968-1975 (71 GP): 17 trailing, 33 tied, 52 leading
Peak Clarke, 1973-78 (83 GP): 29 trailing, 27 tied, 21 leading
I don't have the full TOI by game score for Esposito's teams, but it is likely that his teams spent a bit more time in the lead than Clarke's. For example, Gerry Cheevers spent 45% of his career playoff TOI with his team in the lead, while Clarke's Flyers spent 39% of their total TOI with a lead from 1973-78. But even with those adjustments, I think it's pretty likely that Clarke actually had a better per-minute scoring rate with his team trailing than Esposito did. Even with favourable estimates for Boston's distribution of TOI by score, I'd estimate that peak Clarke scored up to 20% better than peak Espo while their teams were trailing, Esposito scored up to 80% better with the game tied, and Esposito scored at least 125% better with his team already in the lead.
Now it should be noted that plenty of those points scored while leading still had a very positive impact on win probability. Giving your team a two-goal lead has diminishing returns depending on the time remaining, but nevertheless holds real value at pretty much any point in the game other than the final seconds. But nothing is nearly as important as game-tying and go-ahead goals, particularly late in the game. Also, scoring with your team behind by a lot of goals is admirable but not really all that relevant to win probability either (although that doesn't really apply here since Clarke didn't score many of those goals, with only 5 points in his playoff career coming with his team already down by 3 or more).
If you're doubting how much this is a matter of usage and whether it could merely reflect some sort of clutch play or whatever, I can provide further evidence using advanced stats from Natural Stat Trick, with Patrick Kane and Jonathan Toews as stand-ins for a player who is used in an offensive role regardless of situation and a two-way player who is often called on to matchup against opposing centers and defend leads. We see that Kane's individual Corsi events (the total number of shots on goal, missed shots, and shots that were blocked taken by that player himself) were consistent regardless of score at 5-on-5 in the playoffs (15.4 per 60 trailing, 15.2 tied, 15.3 leading), whereas Toews had more individual even strength attempts when his team was not in the lead (11.1 per 60 trailing, 12.5 tied, 10.2 leading). Kane also had more high danger and rush chances with Chicago in the lead, leading to a better finishing rate on his chances, whereas Toews' high danger and rush chances actually dropped with his team in the lead, which is unusual for forwards but reflects the fact that it usually wasn't Toews' job to take offensive chances in that scenario:
5-on-5 Playoff Stats, 2008-2018 (source: Natural Stat Trick):
| Leading | Tied/Trailing |
| Kane Pts/60 | 2.83 | 2.06 |
| Toews Pts/60 | 1.51 | 1.66 |
| Kane iCF/60 | 15.3 | 15.3 |
| Toews iCF/60 | 10.2 | 12.0 |
| Kane iHDCF/60 | 3.8 | 2.2 |
| Toews iHDCF/60 | 2.8 | 3.2 |
| Kane Rush/60 | 0.9 | 0.5 |
| Toews Rush/60 | 0.2 | 0.3 |
[TBODY]
[/TBODY]
(iCF - individual Corsi events, iHDCF - individual high danger chances, Rush - rush attempts)
So even though it has been claimed that a focus on stats causes voters to overrate offensive stars, I think the exact opposite should be true. A rational, detailed analysis of all the relevant numbers should generally favour two-way centers over one-dimensional offensive forwards. They deserve not only recognition for their defensive contributions and matchup deployment, but they should actually get extra credit for their offence because a majority of it comes in high leverage situations compared to offensive players who often pad their stats with their team's victory already mostly secured.
