Watchouts of MyHockeyRankings

While MyHockeyRankings (MHR) has many benefits, I’ve seen and heard enough parents use the site in a manner for which it was not intended. Instead of being used for good, it can be used for evil to the detriment of player development and their game.

Let’s get the most obvious one out of the way first…

  1. Focusing on your team ranking

Coaching games to maintain or improve your team’s rank/rating goes against the original intent of the site. By putting a focus on ‘goal differential’ over playing games to your team’s fullest capability is essentially poor sportsmanship. One example is not pulling your goalie late in a close game to minimize the risk to lowering your rating even more. Another one is to keep/only play your best players late in the 3rd period even when the outcome is clear. Going into a game knowing the EGD and playing to match or exceed that difference should not be on the mind of any coach before or during a game. 

2. Using MHR rating or ranking as the measurement of team success

Like in business, a team cannot just look at a single metric to see determine the how well it is performing. Usually you’ll need 2 or 3 attributes to get the full picture of how an organization is performing. Things like player development, win/loss record during league or tournament play, and learning to compete are much more important than any single rating metric.

3. Playing for a highly ranked vs middle-of-the-pack team

Coaching certainly plays a role in the development and success of a team. However, the size of the pool of players in an area and the multi-year commitment to player development of a club or region is really the biggest factor in how good a team is. This is why regions like Toronto, Boston and Minnesota have so many strong teams. They have both robust club programs to develop players from Mites to Midget as well as a deep group of players in their programs to choose from. Thus, as a parent, it really shouldn’t matter if your child’s team is highly ranked, what matters is that they continue to develop on a path to help them be the best hockey player they can be.

There are also several weaknesses to the actual algorithm using only goal differential for team ratings. Here are a few of them:

4. Lack of uniformity in game format and duration

Not all games are created equal. While USA Hockey tries to standardize games across divisions, the reality is that a large portion of games that are included in the rankings do not follow those guidelines. These can include games from tournaments, exhibition and pre-season game. The attributes that are not consistent across games can include game time, how regulation ties are handled (e.g. overtime vs. shootout vs. no extra time).  Last year, we were at a tournament with 90 second penalties while the total game was only 75% of a real game. This season our pre-season games were two twenty minute running-time games. There is no way to normalize scores based on the running time of a game.

5. Games scheduled between teams with Expected Goal Differential (EGD) great than 7

Per the original MHR manifesto, only scheduling games between teams that will be competitive makes perfect sense. However, in some regions having division where there is a large discrepancy between the top and bottom teams may occur.  Since MHR max goal differential per game is 7, I have seen several times where the lower rated team’s rating went up even though they lost by 10 goals, since the teams ratings difference was 8 or 9 goals heading into the game. I would recommend changing the algorithm to not include games between teams that have a 7 or more goal differential.

It is my experience that the MHR rating should be taken with a grain of salt and statistically there is probably some reasonably standard deviation between 0.50 and 0.75 rating points. Once again though if you are using the site for its intended purpose, then it shouldn’t matter what the actual rating is for your team. Furthermore, the natural standard deviation makes the rating even more meaningless.  Here are some additional factors that contribute to the standard deviation:

6. Tired teams

Most tier teams regularly play 4 to 6 games in a weekend.  While fatigue is something all the teams need to deal with, when the key metric for MHR is goal differential, it is very likely that final scores between two identical team will not be the same at the end of a 6 game weekend as they would have been on the first day. I have been surprised on many occasions when I expected to see a blow-out between two teams, but it was clear that the higher rated team couldn’t maintain the same level of play for 3 full periods in their final game.

7. Backup goalie dynamic

Ratings are a weighted average of both goalies.  But on many teams there can be a big gap between the top goalie and the second goalie. On others, there may be only one goalie. One season, one my kid’s teams had a goal differential rating of about 1.5 difference between the two goalies.  In this situation, wins vs. losses is a much better indicator of the team’s success instead of goal differential.

7. Asymmetric Actual Goal Differential

In my experience the EGD vs actual goal differential appears asymmetric when the EGD is about 4 or more. Usually this happened when team from different division play each other (i.e. when the higher ranked team has played most of their games against higher ranked teams and the lower ranked team traditionally plays lower ranked teams). So, the rating don’t reflect an apples-to-apples set of teams they have played and when the two teams play, the higher rated team can significantly exceed the EGD.

I am sure there are several other factors I have missed that contribute to the rating not being as precise as possible.

So how should you look at the ratings?

As mentioned above, take it with a grain of salt and don’t focus on the specific number, but more the peer group you are grouped with to see how your team is doing relative to others.  In addition, don’t be concerned about any number rating or ranking, focus instead on player and team development because at the end of the day that is what youth hockey is all about.

The Benefits of MyHockeyRankings

I spend a lot of time on the MyHockeyRankings (MHR) website.  I don’t use the site because I care about the rankings of my kids’ teams, but I do see the rankings, so I am aware of what they are. However, there are many more valuable reasons to use the site that I find really insightful that I wanted to share. I will also put together my thoughts on what the watchouts and drawbacks are for the site in a separate post.

To start, let’s remind you of why MyHockeyRankings was created and how it is intended to be used. As stated explicitly on their About Us page, the site was set-up to help with scheduling competitive games between clubs. Using a pretty simple algorithm based on goal differential, a rating is created for each team.  The difference in ratings between two teams is their expected goal differential (EGD) between the two teams if they were to play each other (subtracting the lower rating from the higher rating). This methodology is used to normalize quality of opponent and calibrate one team versus another. I won’t go into the statistical analysis of the legitimacy of EGD, but the goal differential is only an expected value, and thus there will be a lot of variation in actual game scores. If the EGD is small, then if the two teams would play each other, then it likely would be a competitive game. If the goal differential is large (for me 4 or 5 goals is a significantly large gap) then it likely would not be a competitive game and it may not make sense for the two teams to play in the first place. This is especially helpful if there is a large tournament and the organizers are trying to group teams into competitive divisions or for leagues to draw the line between A, BB and B levels. There is a lot more detail to how and why, but that is the gist of the ratings.  Well, now that each team has a rating, it is only natural to rank them.  This is where much of the controversy starts with the MHR site, but we’ll discuss the use of rankings separately.

  1. Schedule, Scores and LiveBarn

I track about 10 different teams at various clubs, levels and age groups and I find it much easier to see the schedule and scores in one single location thanks to MyHockeyRankings than going to each individual team’s website or leagues site to find the schedule and/or score.  If I want to dive into a particular game then I might go elsewhere, but being able to see which of those teams has/had games in a particular weekend is very helpful.  As a bonus, having the LiveBarn icon next to a game (especially if one of the teams is playing at an away tournament) it lets me know that I could watch the game if I wanted to. This feature was remarkably helpful for watching games for a prep school team that we are considering for my daughter. Without the MyHockeyRankings/LiveBarn partnership, I might never have been able to see several of the school’s games.

2. Scouting and Researching

When heading to a tournament it is highly likely that most of the teams we will play we won’t have ever seen or played before. We have used the MHR ratings to help decide which goalie to start in which games. If on the first day of a tourney we have two games and one team is clearly rated higher than the other on MyHockeyRanking, then the coach usually has some insightful information on who to start in each game. For a playoff game one coach asked me to download the most recent games of the team they will be playing, thanks to MHR I was able to see exactly when and where to find the game on LiveBarn. Finally, as mentioned above, being able to find games from potential prep schools for my daughter and watch their play has allowed me to research potential future schools/teams for her to play on based on the quality of their teams, including the LiveBarn video identification.

3. AA vs. AAA?

I care about my kids’ hockey development not the number of letters they have for their level of play. Since there really is no standard of what is AA vs AAA (which is a topic for another post), MyHockeyRankings also helps to compare AA vs AAA teams in an apples-to-apples manner. For example, I have one of my kids playing on a AA team this season that is rated slightly higher than the AAA team in our area. This is not a surprise, and while having a higher rating doesn’t really matter, there is a big difference between the two teams. The AAA struggles to compete and loses most of their games and their players are usually chasing the puck. Instead, nearly all the AA games my kids team plays are competitive and the players are developing a lot more both on defense and offense, especially when the games are close. While MHR didn’t really play a role in the decision on which team my child would play on, the ratings have provided validation that they are playing at the right level for their development even though it has one less ‘A’.

4. Scheduling Games

As mentioned at the start, this is the original intent of the MyHockeyRankings. Last season our team played about 30 games.  Which is fine, but a few more would have been nice.

Since there weren’t any local teams that were available in the Spring, we looked to find a few teams that were 5 hours away for weekend exhibition games against 3 or 4 teams. Well, of course I used MHR to find the teams which would be most comparable to our team to reach out to (even though they were a level lower by letter).  Once again, this is exactly the intended use of MHR, to help schedule and ensure competitive games.

5. Triangulation Between Levels

My daughter plays on a youth (boys) team and at some point she will switch to playing with the girls. One of the challenges is for me to compare her current level of play with the boys to the girls of her age.  However, thanks to MyHockeyRanking and using their methodology, I can triangulate her team’s ratings to other girls teams.  Since the local girls team plays enough games in the boys division, I can determine their equivalent boys rating and see the difference to my daughter’s team. It provides an additional piece of data to inform our decision on the where and when she should play in the future. I have used the same methodology to compare between levels between teams from Squirt to Peewee, from Peewee to Bantam and Bantam to Midget. For example, how does a Peewee AA team compare to a Bantam B team? Thanks to MHR I am able to figure out the answer to this question.

6. How are you trending? Last 10 Game Ratings

Figuring out if your team is trending up or down during the season is a pretty important insight.  Almost all teams improve throughout the year, but how is your team improving relative to others. Thanks to MHR you can see if the last 10 games are accretive or dilutive to your rating. Also, with the help of a basic spreadsheet you take any time period and figure it out for yourself.  As a data geek, I like the ability to analyze this kind of stuff.

Finally, as I admitted at the start, I am aware of the rankings for my kids’ teams. The ranking usually doesn’t vary too much from the start of the season to the end of the season, so once it has been established, not much point paying too close attention to it.  But knowing where the team ranks on a national and state/regional level is good to know as a parent. It basically helps me set realistic expectations for where my kids are in their development and what goals to help them set for the coming year.

These are the benefits I have found from using MyHockeyRankings and when used properly it has provided helpful insights for several important decisions for my kids’ hockey development. However, while I have used site for ‘good’, it is pretty easy to use the site in the wrong way. My next post will discuss the watch-outs and using the site in a manner that goes against its original intent.

In the NHL, Drafting Well Matters – But Isn’t a Silver Bullet

One would think that drafting quality players (players with >160 games played in the NHL) would nearly guarantee success in the NHL.  However, while the data shows it certainly leads to a consistently competitive team, it does not guarantee a Stanley Cup or even a trip to the finals.  Pittsburgh is a middle-of-the-pack team when it comes to drafting high quality players (15% of draft picks turn into quality players) and only have had 10 them, yet their success stems from more than just their picks.  At the same time, teams like Anaheim, Los Angeles and Washington are in the top quartile when it comes to consistently picking well and their continued success in fielding playoff-making teams can be traced to their draft and develop capability.


Who Many Quality NHL Players Are There In Each Draft Year?

To really judge which how well players and teams performed relative to their draft position, just looking at who made it to NHL to play a single game really isn’t the best metric, since the bar needs to be higher in order for a player to have a real impact for their organization.  So what is the better, next level cut to provide a good measurement of a quality draft pick who made an impact in the NHL.  After looking at the players selected from 2005-2016 the bar for players I recognized and having reasonable careers and stuck around for a few years seemed to be about 160 games.  Two years of games, even if it was over several years looked about right.  Then when I researched that the NHL players are eligible for a pension when they played 160 games, I felt confident that this second cut seems quite reasonable to confirm that the scouting organizations for teams did a good job at picking a ‘winner’ – or better said, a quality player.

Here is a summary of the quality players (as defined as having played at least 160 games in the NHL) by year from 2005 to 2016. As you can see,  each draft typically has 20%-25% of the players drafted end up as quality players, with a typical deviation of about 6%.

NHL Draft Picks Who Played in the NHL by Round

Everyone pays attention to first round draft picks and likes to point out the busts that are sure to be found in the top 30 (or 31) picks every year (e.g. Griffin Reinhart or Jarred Tinordi). In addition, they also like to point out the later-round break-through players (e.g. Jamie Benn or Patric Hornqvist). However, most people don’t realize that after the first round,  the drop off in the likelihood of playing at least one game in the NHL is pretty significant.  Once you get to the 5th round there really isn’t a big difference in the probability that a prospect will make the NHL and thus there is only a minor difference in the value of those picks (to be discussed in more detail in a later post). Players drafted in the first two rounds make up half of all the players to play a single game.

While playing a single game in the NHL is a good first cut at achieving ‘success’ in the NHL (and fulfilling childhood dreams), that probably isn’t the best metric to determine how good a team is at drafting.  Just playing a single game sets the bar a little low.  In the next post we will discuss a better second view of drafting success.

Less than 50% of players selected in the NHL entry draft play at least one game in the NHL

Being draft by an NHL team is certainly no guarantee of ever making it to the show.  Since 2005 only about 43% of players picked in the draft ever played a single game at the NHL level. The draft class of 2011 had the most success with almost 56% of players drafted that at least had a cup of coffee with the big club.  Clearly, the more recent draft years will continue to increase the penetration of prospects who break the single game barrier, but the on average the number seems to consistently asymptotically approach the 50% limit with a couple of expectations.


% of Drafted Hockey Players Who Make it to the NHL

Note: As of the end of the 2016-17 NHL regular season

How to evaluate NHL Draft team success

Over the past few weeks I have been analyzing the results of NHL Entry Draft since 2005.  Over the coming weeks I will be trying to answer the following macro-level questions:

  1. What is the best way to evaluate a team’s drafting success factoring in when they have picked and how many draft picks they have selected?
  2. Which teams have been the most successful at drafting since 2005?
  3. What is a draft pick worth when trading to move up or down?

In order to answer these questions, there is some basic groundwork that needs to be covered to understand the norms or benchmarks of player success during that time.  As a result I will start by providing some general draft insights to help understand the sliding scale of outcomes that can be seen at various stages in the draft.  To start here are some foundational questions which will be answered:

  1. What percent of drafted players make it to the NHL by draft round?
  2. How many of the players who make it to the NHL turn out to be ‘quality’ players?
  3. What are all the various perspectives in which to evaluate a team’s draft success?
  4. Why does a team’s draft success not directly correlate to on-ice success? i.e. which non-draft factors

Hopefully the data and analysis I provide will show fans a more balanced perspective on where their expectations should be set for the draft and their current prospects.

2012 NHL Playoffs – First Round Predictions

This is our first attempt to predict the outcome of playoff series based on our analysis of the value of the players on the team.  Since our approach looks at the body of work of team over the entire year and does not adjust for any other factors other than injuries it will be interesting to see how good our methodology is a predictor of success.  Because our methodology only determines the total value of the team (as the sum of its players) we will not be in how many games a team will win by, since that would be replacing data-driven conclusions with opinion. Please keep in mind that injuries to key players, in particular any top goalie will likely affect the outcome of a series by the reducing the true value of a team.

Eastern Conference

Ottawa Senators ($59.6M) over the New York Rangers ($57.2M)

We have to admit this one immediately surprised us, in fact we had to go back and double check our numbers to make sure this wasn’t a mistake.  But despite some questionable goaltending, Ottawa has some a balanced offence compared to the Rangers excellent goaltending, but weaker defence and overall offensive threat.


Boston Bruins ($64.7M) over the Washington Capitals ($56.2M)

Even without Nathan Horton, the Bruins balance and depth is quite impressive.  Tim Thomas’s performance this year was just okay, but given the team in front of him, the Bruins will put the puck in a lot more then they will take the puck out of their net. Washington’s injuries at the goalie position and below-average scoring ability really don’t bode well for them to upset the Bruins.


Pittsburgh Penguins ($65.9M) over the Philadelphia Flyers ($66.1M)

Statistically these two teams are equal in terms of their 2011-12 overall performance.  However when you adjust for the injuries to key Flyers like Van Riemsdyk, Meszaros, Briere (and Pronger even though his absence is already reflected in the numbers) their team isn’t as valuable right now as it has been all year. When you combine that fact with the obvious, that a healthy Sidney Crosby for the playoffs and no key injuries to Penguins, Pittsburgh’s true value for the first round is higher than their regular season numbers show.


New Jersey Devils ($58.1M) over the Florida Panthers ($53.4M)

New Jersey has above average goaltending and scoring while the Panthers are below average in each category (Goaltending, Defence and Forwards). In particular, Florida is significantly challenged on offence and will likely struggle to score goals.


Western Conference

Vancouver Canucks ($61.5M) over the Los Angeles Kings ($46.8M)

The strength of Vancouver was not a surprise, but the weakness of the Kings was more than just a little shocking.  Our analysis showed not only the obvious, the lack of scoring ability for L.A., but also a lack of depth on the blue line.  The Kings have by far the lowest value defensive corps due to the departure of Jack Johnson.  Despite their quality goaltending, the Kings will continue to struggle to put the puck in the net.  The Canucks have the best goalie tandem in the league and solid defense. The only question for Vancouver is around scoring.


San Jose Sharks ($61.5M) over the St. Louis Blues ($55.5M)

Even though the Blues swept the Sharks during the regular season, the difference in team value is quite apparent.  Clearly the Blues have excellent goaltending and good defence, but a lack of scoring from their forwards especially compared to the Sharks balanced attack should be noticeable.  Despite average goaltending and a balanced defence, the overall depth of the Sharks should be enough to get by the Blues.


Chicago Blackhawks ($62.9M) over the Phoenix Coyotes ($58.3M)

Chicago is above average across the board, although not by much. The offense they can generate from their defence will be the key to their success.  Phoenix has some solid goaltending in Mike Smith, but the Coyotes lack of depth in their skaters is expected to show over the course of the series.


Nashville Predators ($63.4M) over the Detroit Red Wings ($62.1M)

This is another one that was closer than we expected.  Nashville has quality goaltending and potent offensive upside from their defence. Despite what most people think, their offence is also above average.  For the Red Wings they continue to be well-balanced across the board.  Detroit is also better than the mean across all three categories.  Nashville gets the nod just by a few hundred thousand dollars in each category, including the best goalie in the league, and is predicted to take the series. However, any injury to a key Predator player will likely make a big difference in being able to win the last game.

Total Salary Cap Spending by Position 2007/08 – 2010/11

To ground everything, let’s look at the basics of how big the pot of NHL compensation dollars was for each of the last four seasons segmented into three buckets:  forwards, defensemen and goalies.

The numbers above are the starting point of our analysis. We take the totals for each type of player and divide them between all the players based on their REGULAR SEASON performance.  For more details on about these calculation, please read our Puckonomics methodology.

Here are some key assumptions that went into these numbers:

1) Two-Way contract players needed to play a material number of games to be included in the analysis. If a 2-way non-goalie player played less than 15 games they were probably not included.  There are about  150 forwards, 50 defensmen and 15 goalies who had a cup of coffee in the big leagues each year who did not play enough games, generated sufficient statistical value or earn enough salary to be included in our analysis.

2) One-way contract players are included in the analysis even if they were hurt and only played a small number of games during the season.  The exception to this assumption are players who did not play a single game all year and were on Long Term Injured Reserved the entire season. Example of players who missed a full year are Mike Rathje in 07/08 and 08/09 or Mark Streit in 2010/11. Salaries for these types of situations are not included. However, Marc Savard who only played 25 games in 2011 had his full $4M  cap hit included. Savard’s 2011-12 $4M cap hit will not be included since it does not look he will play this year.

At this high-level there are a couple of interesting points of note:

– Average salaries seen above are pretty similar between the three groups, but not all that meaningful given it would just be an average and includes a wide range of player quality and number of games played.

– It seems the trend to paying goalies a smaller percentage of overall salary cap is indeed reflected in the numbers.

Playoff Adjustment

There are 1230 regular season games and about 88 playoff games (give or take a couple) each year.  Since the salary cap is directly tied to the total revenue of the NHL, it is important to remember that the pool of compensation dollars is a fixed pie of ~57% of the total revenue (depending on how many teams spend to the cap). Since playoff revenue is disproportionately higher than the 7.2% of number of games it represents, for simplicity sake we assume that Playoff-related Revenue generates ~10% of total league revenues. As a result, if a team/player makes the playoff we will also show an adjusted salary related to their playoff performance (if any).  It makes sense for consistently  deep playoff performers like the Detroit Red Wings to be making 20-30% more than their calculated REGULAR SEASON numbers, but that means that extra amount should be taken out of someone else’s compensation who is not playing or performing in the PLAYOFFS.  So if a player does not make the playoffs, their VALUE is adjusted downward by about 10%.