Benmocha: Predicting Stanley Cup Finalists With Historical Data

by David Benmocha | Posted on Saturday, May 10th, 2014

StanleyCupHandshakeLine(SlidingSideways) Predicting the Stanley Cup finalists is one of the toughest things to do because of the uncertain value of momentum swings. For example; how would someone predict that the Los Angeles Kings would come back from a 0-3 deficit and win their quarterfinal series 4-3? Not only did the Kings win four straight games versus the high flying San Jose Sharks, but they also added two more semifinal wins versus the the Anaheim Ducks, winning each of their last six games. This phenomenon cannot be explained by simply watching the game or crunching their Corsi/Fenwick statistics. (Photo/slidingsideways)

You must dig deeper to understand the characteristics of a Stanley Cup winner and how history has played a huge role in the adaptation of those characteristics in the past decade. 


In my research, I studied six statistics in the playoffs that I believe  indicate whether a team is primed to win the Stanley Cup or, at the very least, make it to the final. My theory is based on the CBA rule changes in the 2004-2005 season lockout, in which the NHL sought to increase goal scoring in the playoffs. The league implemented a zero-tolerance policy for hooking, slashing, tripping, cross-checking and inference. Players who used their stick or free hand to slow any opposing player were penalized. The following two years, there were huge spikes in penalty minutes per game, but after those years the numbers started to balance out due to adjustments that players and coaches made in their strategy as well as general managers in their roster construction plans.


The six statistics that I studied were PIMs/game (penalty minutes per game), goals per game, penalty kill percentage, power play percentage, 5v5 goals for vs. goals against and even strength goal percentage. The reason I choose these exclusive stats is because I wanted to know to what degree the “Man-Advantage” matters in the playoffs. Does it account for 30%, 60% or 90% of a team’s wins in the playoffs? The evidence I found was counterintuitive to what one might theorize about the trends of increased penalties in hockey. I studied three types of playoff teams in three different eras. The first was pre-CBA (five years: 1999-2004), buffer period (two years: 2005-2007) and post CBA (six years: 2007-2013). The three types of teams in the playoffs were Stanley Cup champions, Stanley Cup finalist and the average team in the playoffs. In PIMs/game, the Stanley Cup winner was nearly identical from pre-CBA (11.7) to post-CBA (11.6). The same result was true for the Stanley Cup finalist from pre-CBA (12.5) to post-CBA (12.4). What this is telling us is that to win the Stanley Cup, you must be disciplined enough to avoid handing out unnecessary opportunities to your opponent. In penalty kill percentage, the Stanley Cup winner was very similar in pre- CBA (87.8%) to post-CBA (86.1%). However, the Stanley Cup finalist dropped significantly from pre-CBA (85.1%) to post-CBA (81.3%). This trend means that the Stanley Cup winner must have a good penalty kill % to win the Cup.

I believe that the penalty kill percentage is the cornerstone of a successful playoff team, since penalties are more involved in this new NHL than in years previous. So counteracting a trend holds great value. 
In power play percentage, the Stanley Cup winner has completely flip-flopped from pre-CBA (18.4%) to post-CBA (15.7%). The same result has occurred with the Stanley Cup finalist from pre-CBA (14.5%) to post-CBA (20.5%). My explanation of this phenomenon has to do with penalty minutes per game. Think of hitters in baseball. A batter gets on average four at bats per game. If he goes three for four, his batting average goes up and he looks like an all-star batting .750. But what happens if you give him more at bats? His batting average will naturally decline because the sample size has grown larger and a .750 batting average is not attainable throughout a 162 game season. This is what happened to the NHL. Because less penalties were called, pre-CBA, teams relied on scoring on the power play as a way to counteract their weakness in even strength play. So, year after year from 1999-2004 the teams that won the Stanley Cup had a better power play than their finals opponent. This trend came to an end once more penalties were called. The new trend has stood solid in six out of the last seven years, dating from the 2006-07 season to last year’s season. In even strength goal percentage, the Stanley Cup winner has flip-flopped from pre-CBA (64.7%) to post-CBA (69.5%). This same outcome stood still for the finalist from pre-CBA (69.1%) to post-CBA (64.5%). This trend follows our last theory of PIMs/game and it’s correlation with power play percentage. The reason for the power play percentage decrease is because teams are getting more chances than they had previously.
But, relying on power play goals is a scary tactic because in a close game, that strategy relies on the referee(s) making calls in your favor. That strategy is simply unsustainable due to the fact that penalties called are out of a team’s control.
Since power play percentages are decreasing, that means even strength goal percentage must be increasing. There is a direct correlation between having the “man-advantage” and playing at “even-strength.” In 5v5 goals for/ goals against, the Stanley Cup winner has increased goals from pre-CBA (1.17) to post CBA (1.44). The Stanley Cup finalist has increased goals from pre-CBA (1.19) to post CBA (1.28). This statistic backs up even strength goal percentage, but also adds context by integrating a team’s defensive ability to keep pucks out of their net during 5 vs 5 play. Notice that pre-CBA finalists had a higher ratio than the Stanley Cup winner and that trend reversed post-CBA by a sizable amount. In addition, it’s important to note that the average team in the playoffs 5v5 goals for/goals against is 0.94 which has stayed the same over the past 13 years.  This trend is indicating the level of difficulty scoring 5v5 due to the quality and depth of teams that make the playoffs. In goals per game, the Stanley Cup winner has increased scoring from pre-CBA (2.76) to post-CBA (3.16). The Stanley Cup finalist has followed the same trend from pre-CBA (2.26) to post-CBA (2.89). This statistic doesn’t tell the audience a lot about how good a team is, as these numbers can be inflated because of a high scoring first and second round series. The best information we can extract from this statistic is that the league has been successful in their objective of increasing goal scoring. The average team went from pre-CBA (2.14) to post-CBA (2.52), but the most impressive outcome has been the strength of competition in the Stanley Cup final. I conclude with this inference because Pre-CBA the difference between the Stanley Cup winner (2.76) to the Stanley Cup finalist (2.26) was 0.50 goals, which is a large margin. That difference has shrunk nearly in half in the post-CBA era to 0.27 for the Stanley Cup winner (3.16) to the Stanley Cup finalist (2.89). This most certainly is a positive trend for the NHL, its fans, and the broader appeal. As of May 10, 2014, I have crunched the numbers for the probability of teams making it to the Stanley Cup final. I observed the statistics that I explained above and used the six categories as a tool to understand which team is well balanced and possesses the characteristics of a Stanley Cup champion.



  • 11.6 – 12.4
  • -Choose 3 penalty minutes difference
  • -This equals out to a penalty and a half in both directions (North and South)

Even Strength %

  • 64.5% – 69.5%
  • -Choose 64,0% and above

Penalty Kill %

  • 81% – 86%
  • -Choose 81% and above

Power play %

  • 15.7% – 20.5%
  • -Choose 3% difference (North & South)

5v5 Goals For/Goals Against

  • 1.28 – 1.44
  • -Choose 1.28 and above

Goals per Game

  • 2.89 – 3.16
  • -Choose 2.89 and above


DBenStats510 Chicago Blackhawks 5 out of 6

Pittsburgh Penguins 5 out of 6

Minnesota Wild 5 out of 6

Boston Bruins 3 out of 6

Los Angeles Kings 3 out of 6

Montreal Canadiens 3 out of 6

Anaheim Ducks 2 out of 6

New York Rangers 2 out of 6

*There will be discrepancies in judgment in some of these categories because of the inflated numbers from the first round of the playoffs. Teams that fall out of the standard deviation, or in other words have statistics that are too high or too low do not receive credit for their effectiveness and thus I penalized them for being an unbalanced team. My theory is that these numbers will normalize over a greater sample size through the second, third and fourth round of the playoffs.


The objective behind ranking teams in the playoffs is to gain insight on which teams are well balanced and possess the characteristics necessary to move to the next round. Teams that fall short in some categories will be faced with an ultimatum of improvement or elimination. It becomes a rocky road the higher a team climbs near the top.

The last two teams standing will be the ones that have adapted the best to the new NHL and its most valued commodity: Even strength play. If the numbers hold true in this year’s playoffs, the teams with the highest probability of advancing to the Stanley Cup Final are the Chicago Blackhawks or Minnesota Wild and the Pittsburgh Penguins. 
Follow me on Twitter @Bennymochs9 and please “like” LightningShout on Facebook. You can email us at

Send to Kindle
David Benmocha
About the Author

Writer, David Benmocha, attended South Kent Prep School and graduated from Manhattanville College with a degree in Finance. He is involved with the Tampa Elite Hockey Club and takes classes at John Hopkins Bloomberg School of Public Health in Biostatistics. He is interested in sports history and predictive analytics. His goal is to be the GM of a sports franchise.

Displaying 4 Comments
Add Your Comment
  1. I have a few problems with your analysis as you’ve designed and implemented it:

    1) This seems very much like a post-hoc analysis. You’ve decided after the fact by looking at a bunch of Cup Finalists what Cup Finalists tend to be, but there is no actual predictive power here. You’re merely comparing common elements of good teams in the playoffs without actually predicting which teams will show these elements beforehand.
    2) Even expanding your search into multiple years, the playoffs are a very, very small sample size. Random fluctuations in shooting and save percentages will grossly skew your data; bad teams will sometimes look good in a 7 or fewer game series, or vice versa. These common variations in shooting and goaltending will drastically affect 5v5 Goals For/Against Ratios and Goals/Game in small samples.
    3. “The reason for the power play % decrease is because teams are getting more chances than they did previously”. I’d quote you directly but LightningShout! doesn’t allow that apparently. I don’t think you’ve shown this at all, not statistically or otherwise.
    4. In fact, the exact opposite has been shown: Power play opportunities are WAY DOWN since the lockout, from almost 6/game down to under 3.5. Over that time, power play % league-wide has actually stayed fairly constant — even gone up a little. But teams get significantly fewer chances. Limiting your data to playoff teams or teams with a chance of being Cup Finalists must be drastically affecting your results.
    5. It seems like a lot of your conclusions and data collection are both based on bad assumptions so I’m not sure a lot of what you’re contending here is reliable in general.
    6. Casually dismissing Corsi and Fenwick is a mistake IMO, when both have been shown to have significant predictive power:
    7. Your methodology reads more like a hypothesis than a true explanation of your methods. You then prove your hypothesis wrong, but don’t really comment on that fact. What data are you pulling from? Where is this data? What methods did you use to cull the data down to the tables we see in this post?

    I’m not trying to be mean or rude, but this analysis is, IMO, misguided. There’s a lot of legwork that has already been done on predicting success in the NHL if you’re willing to delve in.

    If you have any questions about my criticisms or want to discuss this further, you can e-mail me:

  2. BigKahuna BigKahuna says:

    I don’t know advanced statistics but it seems that you simply used the readily available numbers to come up with your conclusions. I find it very interesting that the penalty kill % may be the biggest determining factor for playoff success. In my opinion, this is really good info, especially for the average fan like me. Thank you.

  3. David Benmocha David Benmocha says:

    Thank you for reading and commenting. Always good having other members of the Lightning blogosphere following along. Please feel free to express any comments, questions or concerns at any time.

    I have taken team statistics from pulling both regular season and playoffs. I computed the numbers using a calculator and taking the averages from 1999-2014. I decided to split up the years in three categories in order to understand the context of the statistics for the casual fan using history; Pre-CBA (1999-2004), Buffer period (2005-2007), Post CBA (2007-2014).

    I decided to take the averages of three types of teams in the playoffs, as an indicator of what a contending team’s stats should look like at the end of the tunnel. I chose the average team in the playoffs, the Stanley Cup Champion, and the Stanley Cup Finalists. The reason I choose the Stanley Cup Champion and Finalist is to analyze success. Winning the Stanley Cup is a result that every team seeks and it is why players want to play in the NHL; to be named the best. Any hockey fan casual or hardcore would agree that success is determined by winning the Stanley Cup and being in the final in one of the many steps in the process to achieve that result of success.

    I’m not suggesting that Fenwick and Corsi stat are not useful, nor possess significant predictive power. What I’m suggesting is that you cannot rely singularly on shots as an indicator of how successful a team will be. There are other characteristics important to learn how these goals are being scored and whether or not these patterns can be replicated.

    I agree that the playoffs are a very small sample size and that is why I stated that these numbers will normalize and balance out the further a team moves to a Stanley Cup final. Teams that reach the Stanley Cup final play anywhere between 20-25 games while the average team plays around 11 games. I agree that 11 games is very small sample size as it only accounts for 13.4% of a regular season. That sample size doubles to 24%-30% of a regular season for teams that make it to the Stanley Cup final. The more games these teams play the less likely they are able to hide their weaknesses with power play goals. Under the Pre-CBA rules, teams with the best “Man-Advantage” were valued higher, thus giving rise to their success. In the Post-CBA era, teams cannot rely on power play goals to help them win games every night. It becomes a matter of sustainability and endurance. The only guarantee a referee has to a team, is that the game will start with even strength play. Whether the game has 40 penalties or 0 penalties, even strength play will happen and it is promoted to encourage fairness as long as a player doesn’t cross the line. The context is simple, “Control the elements of the game that are controllable.”

    Elements such as emotions, work ethic, timing, shooting, passing, aggressiveness, positive thinking, and communication are all controllable player to player. The successful teams bring these characteristics together for a common goal. The Stanley Cup is representation of success, which has been acknowledged throughout hockey culture and history. Teams that follow the process the right way will be rewarded for their consistency and efforts, which history tells us is the ultimate predictor of success.

  4. ITMAN says:

    Not sure who peed in that guy’s Cheerio’s but I have no problem with the statistics used and the way the writer uses them. Way better than some of those advanced stats that are hard to understand and don’t represent reality. I also laud this site for protecting their copy and photo’s. Everyone knows that predictions are a crap shoot, but this writer has used everyday common stats to predict a winner through probability. Nice job and a good read!

Leave a comment

XHTML: You can use these html tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Most Popular Posts Outlet
  • Read Us On Your Kindle