In the realm of sports, a longstanding principle asserts that “a team that commits the fewest errors in a game is likely to emerge victorious.” It is this very postulate that served as the catalyst behind the inception of hockeyfreeforall.com (HFFA) and the subsequent development of the Advanced Bracton (AB) family of metrics. On June 3rd, 2015, following the conclusion of the 2014-2015 NHL season, HFFA formally entered the hockey analytics community, unveiling the initial articles elucidating the proprietary methodology underpinning the AB score. Over the ensuing nine years, the foundational concepts and ideas evolved into a more comprehensive, intricate, and multifaceted system that transitioned from mere individual player analysis to encompassing discussions of organizational philosophy.

Through the accumulation of data spanning over 17 years, it became evident that the philosophical foundation of HFFA, as embedded by the AB score, is fundamentally sound. Engaging with insights from industry experts and professionals underscored the importance of developing one’s unique strategies or ideology in the game, thereby setting individuals apart from casual fans. This differentiation arises from the remarkable accuracy of retrospective data analysis and its potential applicability in shaping prospective managerial decisions. The process entailed the conceptualization of ideas and an exhaustive examination of every crucial facet of hockey operations. These facets included player drafting, development, scouting, contract negotiations, talent acquisition via trades and free agency, as well as executive personnel decision-making. These ideas were subsequently structured into a proprietary format and have been published over the years. Importantly, the principles clarified herein possess a universal versatility that can be effectively conveyed in a professional context, spanning all levels of the game. Furthermore, this body of research extended beyond the realm of player analysis. It encompassed various industry analysis exercises, including the evaluation of the hockey agency business, financial assessments for all 32 NHL teams, and an in-depth analysis of the NHL arbitration process. Consequently, the study presented here serves as an exposition of the underlying philosophy, its resultant findings, and a prelude to the proposition that past experiences often serve as a prologue to future outcomes. These insights have been cultivated over nearly a decade of dedicated industry research.

The Advanced Bracton Score (AB)

The Advanced Bracton score represents a meaningful quantification of the impact of errors on outcomes of hockey games. Just as there are positive multiplicative effects of multiplying two negatives in mathematics, the same holds true in hockey. As a result, two metrics arose; the first one being the Bracton Score “Bracton”, and the second the “Advanced Bracton Score.” The “Bracton” numerically calculates the positive or negative contribution of a player to his team vis-a-vis mistakes taken or generated via the penalty assessment process. The Bracton Score can also be used when aggregating the ability of an entire team to generate more mistakes from its opponent than it commits. In formulating the “Bracton,” the following assumptions and questions about hockey in general to deduce a numerical conclusion were considered.

• If the only differentiator in a game pitting world-class athletes against one another is the combination of skill, intelligence, and the desire to win, can these intangibles be measured?
• Through an analysis of the process of taking and drawing penalties, is it possible to partially evaluate these intangibles?
• Assuming players are most often only desperate enough to gain some extra-rule advantage when opposing players are either working harder for the puck, are demonstrating superior skill, or are making more intelligent decisions on or off the puck. Conversely, the players that consistently take minor penalties are the ones who are either slower, are being outplayed, are out of position, consistently playing outside the system of the coaching staff, or worse, playing injured. These situations only arise due to an asymmetry of skill, intelligence, desire, injury, or a combination of all four.
• When a team either takes or draws a penalty, the difference between goals produced and goals allowed is far higher in both directions. Therefore, players who possess a positive “Bracton” score allow for an additive contribution to the success or failure of a team.
• Players that generate penalties, even if they are normally not on the power play unit, maximize ice time for the assumedly most skilled star players who are on the power play unit. Conversely, and most significantly, players who take more penalties than they draw maximize ice time for opposing premier players. This is because 5×4/60 minutes is much higher than 5×5/60 minutes and saves goals by not taking penalties 4×5/60 minutes.

Even though the above points may seem obvious, we must ask ourselves, why hasn’t a metric such as the “Bracton” been either advanced, widely reported, or become a mainstay in hockey for decades, not just in the dawning era of “Moneyball” in hockey? While the “Bracton” is a powerful metric, if not somewhat a priori, it must be understood that it only provides a springboard to its more highly correlated predictive brethren; the Advanced Bracton (AB) Score. In this proprietary statistic, various data points related to retrospectively positive or negative outcomes within the play of hockey games aggregated for a season, were compiled within a highly complex formula in order to assign every player a numerical value. These statistics were added to the Bracton score to uncover an even more meaningful quantification of whether a hockey player impacts his team positively or negatively. Unlike the “Bracton”, these results are in no way obvious even though for seasoned hockey enthusiasts and managers they may be intuitive. For example, for someone to say, “I knew Auston Matthews was a good player, but I didn’t really know how good relative to the rest of the league” the AB successfully measures ostensible intangibles that appear to relate to what I term “unscored goals at the margin (UGM).”

By mathematically deducing the contribution of a player toward UGM, we can assign an actual value (or in many cases a negative value) players have toward the overall team performance. For example, in the case of Auston Matthews, he scores among the league-best AB at 9.42. Under the AB philosophy, in addition to the goals and assists produced by Matthews, he was responsible for another 9.42 goals UGM for Toronto that neither he nor the Leafs scored directly but avoided by either limiting mistakes or being “in the right place at the right time with the right mindset,” to borrow a tenet of eastern religious thought. In this regard, to earn a 9.42, Matthews undoubtedly had his head in the game, played hard the majority of his shifts, and played within the system employed by Toronto’s coaching staff. As such he is what the sports community terms a “character player.” A character player is not only someone who plays with heart and high energy, but if he is someone who provides UGM, he is possibly worth the money he is being paid to play hockey – a statistic also tracked herein by our proprietary arbitration analyzer.

In contrast to success stories like Matthews, it was quite surprising just how many players in the NHL not only produced negative UGM but recorded lower-than-replacement level production for their respective clubs – a replacement player analysis has recently been completed in the context of the AB metric as well (see below). By virtue of their presence, negative Advanced Bracton players take away UGM and actually cost their team goals, wins, and points in the standings, and above all, strike the teams’ salary cap with detrimental returns on investment.  In fact, it was also striking to note that roughly half of the players in the league could be replaced by a lesser-known talent and possibly have an overall positive impact on the team replacing them. Also, it became apparent that there are some teams in the NHL that use a metric similar to the AB in constructing their teams. The Pittsburgh Penguins, Tampa Bay Lighting, Colorado Avalanche, and Vegas Golden Knights stand out as notable recent examples. Of the teams in the middle of the pack, generally, it was concluded that even if they had several high UGM players, their success was muted if they had several in the lowest quintile (like Detroit, Ottawa, and Nashville). Finally, the worst teams in the league are literally loaded with negative UGM Advanced Bracton players with few or no players to counteract them at the top (Montreal, Philadelphia, Anaheim, San Jose).

From a general management philosophical standpoint, teams who have adopted or will adopt the Advanced Bracton method of team construction enjoy more playoff success and longevity than teams who do not. The remainder of this paper will demonstrate further analysis into the roster construction of NHL teams since the 2007-08 season, as well as provide more insight into the concepts addressed to this point.

Replacement Level Player Analysis

In order to adequately analyze the roster construction of NHL teams, the value of a replacement-level player (RLP’s) in accordance with AB must be determined first. The relevant statistics were derived as an offshoot of the formation of the AB itself, over sixteen NHL seasons. In theory, one would think a RLP should be a zero-to-slightly-positive AB player. However, in conducting the research, that theory would be debunked, as the following analysis details.

The following tables and graphs depict the changes across the study period in AB Score and RLP average in forwards and defensemen, as well as skaters as a whole.

The findings above lend keen insight into the evolution of hockey over the past 16 seasons. The quality of play by defensemen in relation to AB has improved by 13.8% since the beginning of the 2007-08 season. This could be the result of an influx of talented, smart defensemen entering the league over this span, such as Tampa Bay’s Victor Hedman, Dallas’ Miro Heiskanen, Colorado’s Cale Makar, New York’s Adam Fox, or Nashville’s Roman Josi. Regardless, it is noteworthy to view this trend from a standpoint of mistake minimization, as in theory, defensemen typically play more minutes than forwards, meaning they would supposedly have more time to make mistakes – even when calculated at a rate per sixty minutes. The same outcome was true of forwards as well, at a pace that was even more accelerated. especially pronounced over the last six seasons. It is also fascinating that the RLP average for both position groups is a negative AB contributor, as this illustrates just how valuable these character players are to their team’s success.

The second part of the hockeyfreeforall.com method of replacement player analysis according to AB consists of a shift from overall position group evaluation to a highlighted focus on individual player performance. Over the 16-year sample size, 11,612 season evaluations were created, reflecting the play of 2,295 unique individual players. Similar to the baseball statistic of WAR, the Advanced Bracton Above Replacement (ABAR) was born to illustrate these contributions. For all 11,612 points, individual ABAR was calculated by finding the difference between the player’s recorded score from the average numbers detailed in the graphs and tables above. These numbers were then added together by the player’s name to determine who performed exceptionally well above replacement over the course of their careers. Not surprisingly, the top-five players according to Career Advanced Bracton Above Replacement (CABAR) are all likely future Hall of Famers Patrice Bergeron (+95.71 CABAR), Brad Marchand (+73.53 CABAR), Pavel Datsyuk (+70.76 CABAR), Jonathan Toews (+61.83 CABAR), and Sidney Crosby (+59.84 CABAR). After completing this analysis, it was staggering to learn that only 978 of the 2,295 unique players in this study have CABARs above 0 (42%), and 1,084 of them above the replacement level skater average from the previous page (47%). These findings tell us that essentially half of the world’s best, most competitive hockey league can be replaced – likely with less expensive, younger, more salary cap friendly, and potentially lesser-known (scouted) alternatives.

This point intensifies the requirement of organizational depth, accurate player development and evaluation, and adoption of winning philosophies such as the AB approach would indicate.  Theoretically, GMs around the NHL (or any other level) can target underappreciated, yet talented players and neither be required to mortgage the future, be bogged down in burdensome long-term contracts to below-replacement players, and maneuver within the labyrinth of salary cap rules. The next several sections of this writing will discuss where talented AB players originate from, as well as dissect layers of the NHL draft through one of the most important HFFA metrics, the Tentative AB Score – where analytics meets prediction.

Tentative AB Score (TAB)

The NHL draft is one of the most exciting events on the hockey calendar, as it is a monumental occasion for all parties involved. The fruition of a lifetime of work for the prospects, years’ worth of scouting and evaluation for all thirty-two organizations, and the continued active engagement of hockey fans around the world are on full display for the entire hockey universe to witness and celebrate. However, in speaking with various industry professionals (GMs, scouts, coaches, agents) there is a consensus belief that the draft is essentially a “crapshoot”. No player selected, regardless of draft slot, is guaranteed to be successful, and an undrafted free agent may go on to have a Hall of Fame career. Any advantage, no matter how minuscule, that would increase the likelihood of drafting an NHL-capable player during any round of the entry draft would be beneficial to front offices and scouting staffs around the league. This is the purpose that the Tentative AB Score (TAB) serves in the Advanced Bracton and Hockeyfreeforall.com family of metrics.

Essentially, TAB is an estimated AB score based on the available statistics by league, as well as an analysis of current NHL players that have a similar profile with respect to background, experience, country of origin, and pedigree. There are major discrepancies in the availability of statistics in hockey leagues lower than the NHL, as they tremendously lack advanced statistical presences, thus providing an immense opportunity for right-minded organizations. The TAB Score, when calculated across all major junior leagues produced an 85% correlation between the TAB projection and the actual AB result. To clarify, if a player recorded a positive TAB score, there was an 85% chance they recorded a positive AB score as well. This discovery led to a full expansion of the HFFA database collection by several years, as Tentative AB NHL numbers from 2000-2006 were also created to complement the 16 seasons of complete data. In addition to this, one of the most important studies on hockeyfreeforall.com was created, as a deep dive into the NHL draft from a probability standpoint and was published prior to the 2022 Draft in Montreal.

The analysis below examines the 2007-08 NHL Draft and ends with the 2023 NHL Draft. Despite the fact the 2023 draft class has not played any NHL games yet, these players were included. In total, 3,321 players were viewed during that time span, and a pre-NHL TAB score was calculated for each player, regardless of whether said player actually played a game in the NHL at any point in the future or not. The 3,321 players were then split into their respective draft classes to collect the following statistics by year (all of which exclude goalies selected) and then found the averages of each across the 17-year set.

• Total Drafted Players in the NHL.
• Total Drafted Players to play NHL games.
• Percent of Players to Play NHL games.
• Average TAB Score of Respective Draft
• Average TAB Score of Players who have played NHL games per draft.
• Percent of Players with Positive TAB to play NHL games per draft.
• Percent of Players with Negative TAB to play NHL games per draft.
• Percent of Positive TAB Players in total draft to play NHL games.
• Percent of Negative TAB Players in total draft to play NHL games.
• The Difference between bullet point eight and bullet point nine.

The results of this section are depicted in the table below.

The table above consists of information compiled pertaining to the Tentative AB Score results of each draft class dating back to 2007. We see that on average, excluding the 2023 draft who has yet to play a game, 81 of 192 non-goaltender draft picks (42.49%) on average play at least one game in the NHL. In contrast, most of the skaters (57.51%) never play an NHL game with their respective teams regardless of their draft positioning. Of 42.49% that did play an NHL game, 73% of those players registered a positive TAB score at the pre-NHL levels, while only 27% recorded a negative score, a staggering difference. However, of all the positive TAB players drafted, on average, 45% of them played at least one NHL game, while only 38% of all negative TAB players drafted on average reached the same milestone. This 6% difference is the true value of the Tentative AB Score and the advantages it can provide, as this slight edge may be the difference in drafting a player who could help the franchise or someone who never contributes at the NHL level. If the goal is to make the NHL draft less of a “crapshoot”, perhaps the implementation of TAB research can identify some previously hidden diamonds in the rough for management groups across the league – and somewhat refine the current value system underneath player development and evaluation.

Following these discoveries, data previously discussed by round was collected for the following statistics (all of which still exclude goalies selected). The purpose of this section is to identify how many prospects by round ended up playing NHL games, and what percentage of those prospects were actually positive TAB players.

• Total First Round Picks.
• Total First Round Picks to Play NHL Games (%).
• Total First Round Picks with Positive TAB Scores to play NHL games (%).
• Total Second Round Picks.
• Total Second Round Picks to Play NHL Games (%).
• Total Second Round Picks with Positive TAB Scores to play NHL games (%).
• Total Third Round Picks.
• Total Third Round Picks to Play NHL Games (%).
• Total Third Round Picks with Positive TAB Scores to play NHL games (%).
• Total Fourth Round Picks.
• Total Fourth Round Picks to Play NHL Games (%).
• Total Fourth Round Picks with Positive TAB Scores to play NHL games (%).
• Total Fifth Round Picks.
• Total Fifth Round Picks to Play NHL Games (%).
• Total Fourth Round Picks with Positive TAB Scores to play NHL games (%).
• Total Sixth Round Picks.
• Total Sixth Round Picks to Play NHL Games (%).
• Total Sixth Round Picks with Positive TAB Scores to play NHL games (%).
• Total Seventh Round Picks.
• Total Seventh Round Picks to Play NHL Games (%).
• Total Seventh Round Picks with Positive TAB Scores to play NHL games (%).

The results of this section are depicted in the table below.

The results are very much in favor of TAB, as every round’s percentage well exceeds the majority. This is especially apparent in the later rounds, as 62% of the 18% of total seventh-round picks that played NHL games were positive TAB players, 74% of the 24% of total sixth-round picks that played NHL games were positive TAB players, and 71% of the 27% of fifth-round picks that played NHL games were positive TAB players. These advantages are extreme, and possibly transformational to NHL front offices. Sincerest at the deadline for rental players could provide a distinct future impact. This also could cause management groups to shift their attention away from allocating large salary cap portions to players below replacement as previously discussed, but instead spending more resources on drafting and developing high TAB players with an increased likelihood of finding a gem – the essence of solid asset management.

Following these conclusions, the data was further dissected (all of which still exclude goalies selected) for averages of each player across the seventeen-year set. The purpose of this section is to detail which clubs have drafted the best according to the TAB methodology.

• Total Number of draft picks per team.
• Picks who made NHL team (%).
• Average TAB Score of all draft picks by team.
• Picks Who made the NHL with positive TAB score (%).
• Total Score (sums of the ranks (1-32) of bullets 2,3,4).

These results illustrate which teams have been the most successful at drafting and development over the last 17 drafts in accordance with TAB. The criteria used to determine this ranking include a sum of rankings formula of three specific qualifications (Picks who made the NHL, AVG TAB Score of Draft Picks, and Picks who made the NHL with positive TAB Scores). Each team was ranked on those three criteria, and then those ranks were added and sorted from least to greatest, as lower rankings meant better performance. It was not surprising to see recent Stanley Cup Champions and perennial contenders of the last 17 years top the list, as the Colorado Avalanche, Washington Capitals, Anaheim Ducks, and Chicago Blackhawks have had unquestionably great success over this time span. It’s also very interesting to see that of the 39% average of total picks to make the NHL by each organization, 72% on average recorded positive pre-NHL TAB scores.

The above research under this section clearly details the benefits of selecting a higher TAB player rather than a negative one. Despite these theories, one must not forget or neglect the importance of traditional scouting methods when it comes to the drafting and development process. In no way are TAB and its findings aimed to diminish the many hours of research, interviews, film study, and other scouting processes undergone by departments across the league. However, the goal of TAB in a professional setting is to encourage debate among players who are unanimously liked amongst the members of the locker room, or to perhaps signal which types of players scouts should be focusing time and energy on. In this way, it is easy to envision TAB to be used as a “cheat sheet” in making more informed decisions as an organization.

The next section of this research consists of another study aimed at helping organizational decision-making by highlighting which pre-NHL organizations, universities, or development programs help produce the most talented players in accordance with the AB and its philosophies.

Junior Hockey and Pre-NHL Organizational Analysis

Since the creation of the Advanced Bracton Score in 2014 and the beginning of my data collection in 2007, one of its biggest flaws was predicting the success of the incoming rookie classes. These players entered the league through various levels of the game, whether it be major junior leagues such as the OHL, WHL, and QMJHL in Canada, National Development programs in the United States, or professional men’s leagues in Sweden, Russia, Germany, or Finland. The database contains data on 2,295 NHL individual players since the beginning of the 2007-08 season, and over 2,100 of them are included in this study. The purpose of this research was not only to determine a better idea as to where these 2,100+ players played prior to the NHL but to analyze which specific programs produce the most quality players according to my metric.

In conducting this research, 2,100+ individual players were examined for pre-NHL hockey according to data gathered from HockeyDB.  This was more difficult than it may seem, as some players spent time with multiple programs, and this was adjusted if the player was in close proximity to his eventual NHL season. To be included, each program needed at least two players, as 169 teams qualified to be in this study. The last major condition of inclusion was that the players in this study must have been drafted in 2007 or made their NHL debut during the 2007-08 season. Players such as Joe Thornton, Patrick Marleau, Alex Ovechkin, and Sidney Crosby, are not included despite the fact we have over half or the majority of their career data. Naturally, this hurts some program’s standings in our rankings (obviously the Rimouski Oceanic would’ve been higher had Crosby’s numbers actually counted).

Teams were evaluated on how many of the 2,100+ players were developed by their program according to the criteria in the previous paragraph, what their rookie season AB score was, and whether or not those players were actually career-positive AB players. For the players such as Thornton, Marleau, Ovechkin, and Crosby, a point was awarded for having produced an NHL-caliber player. However, their career AB and 1^st year AB in the overall calculation were negated according to the lack of full information available. The point system is called the Sum of Ranges (which was briefly mentioned in the previous section in relation to TAB). In order to accurately count the number of players who were developed by a certain program, the CountIf function in Excel was employed and created frequency distribution charts for all 169 teams. A point value to the following ranges was then applied. If a player scored a -1 AB score as a rookie or lower, they were assigned a value of 0. If a player scored within the range of -1 to 0, it was attributed a value of 1. If a player scored within the range of 0 to 2, it was scored as a value of 2. If a player scored within the range of 2 to 4 was given 3 from 4 to 8, then 4 and above +8, was assigned a 5 This was used to determine which of the players a program produced had effective starts to their NHL careers, and whether or not said program adequately prepared the player for NHL play. Lastly, the overall Career AB numbers from each of the players were combined into one sum according to the following formula: (Formula= (Sum of Ranges+ Total players) +/-career AB). It is also worth noting that if an AB score happened to fall exactly on the edge of one of these ranges, it received the higher score. For example, Max Jones of the London Knights registered an even 0 AB score during his rookie season in 2018-19, and I gave the Knights 2 points as opposed to 1. In total, there were 11 total players that were eligible for this distinction. It is also important to note that there were various teams across the WHL, OHL, and QMJHL that rebranded or relocated to different markets, as well as some expansion franchises. The teams included in this study are representative of the league’s landscapes in today’s world, with players on older franchise teams being included with the modern ones. For example, players from the Niagara Falls Thunder were factored into the Erie Otters’ overall calculation, and players of the Owen Sound Platers were factored into the Owen Sound Attack’s overall calculation. It would be as if I factored Atlanta Thrashers’ team history into an overall calculation of the Winnipeg Jets. To give an example of one of these players, Matt Stajan played for the Belleville Bulls from 2001-2003. The Bulls later moved to Hamilton and became the Hamilton Bulldogs. Oftentimes, these specific players, such as Stajan, played NHL seasons dating prior to 2007, which caused their AB numbers and range numbers to be omitted. In total, there were 27 programs that have combined scores of organization’s pasts. The table below shows the top 10 pre-NHL organizations or universities according to their NHL graduates’ AB success.

Since the goal of the Advanced Bracton Score is to identify players with unique intangible abilities, especially intelligence, the determination of which semi-pro teams, universities, or organizations produce the best talent in terms of NHL AB production is paramount. Three of the top five and five of the top ten teams are elite hockey universities. To attend these schools, maintain eligibility through GPA requirements, as well as balance work and school illustrates traits and skills that normally underpin at least a replacement-level NHL player. Obviously, not all player development paths are the same, however, the college avenue is one that seems to produce dividends earlier according to the data.

As the table shows, these universities have a large number of players with solid average first-year AB scores. This could be the result of a more controlled development plan, as NHL teams are known to communicate and coordinate with college coaches about their preferred development strategy for specific players, ages, or both. In addition to these collegiate teams, four of the top ten are Ontario Hockey League franchises (OHL), which illustrates that the quality of players entering the NHL from this CHL development system clearly stands out by AB standards. Lastly, Mike Johnston’s Portland Winterhawks, one of the best, most well-run organizations in the WHL, cracks the list at #9, as their NHL alumni are some of the more complete AB players in the overall study. This list is not to discredit or diminish the accomplishments of the other organizations in junior or pre-NHL hockey, as there are plenty of development options outside of these ten that have produced quality NHL talent in terms of AB. However, similarly to TAB, this list and study can provide scouts with supplemental information to their traditional methods, and potentially help the allocation of resources towards making the NHL draft less of a “crapshoot.”

AB Score, Goaltender Edition

Evaluating goaltenders is a difficult task, even in scouting departments with abundant resources. This section needs further study as advanced goaltending statistics aren’t as readily available as forward numbers. When polled, multiple NHL scouts provided specific features or traits a goalie can possess that differentiate them from the pack. In these conversations, the most notable quote was “Can he stop the puck? Yes, or no? It’s really that simple.” This section of the AB research therefore compares goaltenders relative to the league average in a variety of statistics, including the AB performance of the team. Similar to the skater AB score that is the basis behind all of the research in this paper, the goaltender AB score simplifies over 40 statistics down to just one number, assigning each NHL goalie an individual player value. To compare goalies at league averages, the determination of a depth chart was undertaken. Goals against average, total goals allowed, and save percentage were considered by finding averages for all #1’s, #2’s, and #3’s. However, a crucial part of the formula is the AB of the team itself, adjusted for the percentage of games played by one goaltender. For example, last season, the Boston Bruins finished with a +115.67 team AB score. When adjusted for Linus Ullmark specifically, who played in 59% of the Bruins’ games this season, the number decreases to +67.71, which is then divided by a factor when determining the final number.

While it Is possible this metric inflates the actual results of their respective teams, and rewards goaltenders on better teams simply because they are winning, this actually is not the case universally. In fact, when looking at last season, for example, the top goalies according to the metric were all players on playoff teams. However, it was a fair mix of standings finish, as Ilya Sorokin, the goaltender who played on a wild-card Islanders squad finished with a higher overall score than every goalie in hockey except Ullmark, Jeremey Swayman, and Jake Oettinger. The three goaltenders nominated for the Vezina Trophy each season over the course of the 16-year sample size were ones that scored highly on my metric during that season. Last year for instance, the three nominees were Linus Ullmark (1^st overall in metric), Connor Hellebuyck (8^th overall in study), and Ilya Sorokin (4^th overall in study). More study is needed in this area, but preliminary results are encouraging.

Advanced Bracton Coaching Evaluation Metric

            One of the major flaws AB has faced over the years was the inability to project how a player would do under a different coaching staff or system. Positive AB players would move teams and become negative, while negative players moved to become positive. One would think that the coach and his system are the primary reason for this sudden change, which is why this section of the Advanced Bracton family of metrics is among the most significant. Throughout this pursuit, every head coach or interim head coach has been analyzed with respect to success or failure with his team in each given year, as well as the period of his tenure. In total, there were 106 different head coaches in the NHL from 2007-present. A list of the names included in this study will be displayed below.

The methodology was as follows; First, of the 16 years of potential coaching time, several coaches received credit for years in which they did not actually finish the season employed by the organization. In these instances, the coach under examination either coached a majority of the season’s games, making the player’s individual scores a greater representation of his work or took over the position in the second half of a season despite coaching fewer games, potentially making the player’s individual scores a more recent reflection of said coaches’ work. A perfect example of this dilemma is the case of the 2020-21 Buffalo Sabres. The Sabres fired coach Ralph Krueger 28 games into the 56-game regular season, appointing Don Granato with the interim tag. Due to the fact that Granato’s 28 games were the last 28 games of the season, we infer that the players’ individual scores were more of a reflection of Granato as opposed to Krueger due to the recency. After thoroughly examining all of these cases, specific coaches were allocated to each player’s AB score recorded in the database. This was done with the aim of determining which coaches were responsible for guiding each player and when this guidance took place. In total, there were 11,624 players included in this database. There are some novel statistics specifically designed for this project as well. These statistics encompass Player Movement Trackers (PMTs), Advantage Percentage, Return Improvement Percentage, and Average Improvement Number.

The Player Movement Trackers diligently monitor the movements of NHL players under different coaches throughout a 16-year time frame. They also track the positive or negative changes in individual AB scores resulting from coaching switches. These trackers shed light on whether players showed improvement under the same coach in consecutive seasons and identify which coach yielded better performance if a player played under different coaches within a two-year span. “Movement” is also recorded when a coach is dismissed, as this results in a player having two separate coaches within the same two-year period. This depiction helps us ascertain whether the decision to change coaches ultimately benefited or hindered the team, both in the short and long term. The Advantage Percentage quantifies the percentage of players who improved under a new coach compared to their previous one. Meanwhile, the Return Improvement Percentage signifies the percentage of players who improved by returning to their coach from the previous year. This movement data holds significant importance for our next major project at HFFA, where we will seek to quantify General Managers’ decision-making regarding roster construction through trades and free agency, as well as their choices in hiring coaches. Finally, the Average Improvement Number represents the total number of individual AB improvements made on an annual basis by a coach’s returning players, divided by the total number of players who returned. Instead of analyzing individual AB scores individually to determine effectiveness, we opted to employ a point-based system similar to the sum of ranges statistic used in the Junior Hockey study. Each individual AB score recorded by players under a specific coach was assigned a point value, ranging from +5 to -5. Scores above +10 were assigned 5 points, while those between +4 and +9.99 earned 4 points, and so on. In addition to the aforementioned information, there are other intriguing but unrelated statistics that do not factor into the overall equation but may prove useful. Below, you’ll find an example of what the final statistic card looks like for each individual coach in this study.

Obviously, not all coaches have the same system or even ways to motivate and get the most out of their players. The goal of this study was to identify which coaches were able to generate results as far as AB is concerned, and I took team AB performance into account during this evaluation. Below is a ranking of the league’s active coaches who are not entering their first season. Anaheim’s Greg Cronin, Calgary’s Ryan Huska, Washington’s Spencer Carbery, and Columbus’ Pascal Vincent will not be included in this ranking for that reason, as there is yet to be any AB data on them.

The results according to AB corresponded with the consensus rankings amongst fans and people within the hockey industry. The top of the list includes multiple coaches who have won Stanley Cups or have had long, prosperous careers with a variety of different organizations. Often times throughout the course of the study, the bottom tier of coaches on the AB annual rankings are released for higher options – as published on hockeyfreeforall.com on July 27, 2021. At the time, the bottom six coaches were Detroit’s Jeff Blashill, San Jose’s Bob Boughner, Anaheim’s Dallas Eakins, Nashville’s John Hynes, Vancouver’s Travis Green, and Edmonton’s Dave Tippett. Two years later, all six of these organizations have new coaches behind the bench. This can’t be coincidental, and the removal of coaches who tolerate mistakes has been an emerging trend. There are a variety of methods to determine how successful or unsuccessful a coaching hire was according to AB. and the overall rankings list may have predictive power.

In the case of the six aforementioned individuals, the San Jose hockey Sharks underwent a coaching change when they replaced Bob Boughner with David Quinn, who held a considerably higher rank, 28 positions higher according to the AB metric. Initially, the AB metric would suggest that this was a prudent hiring decision. However, the team’s performance witnessed a significant decline, experiencing a 17-point decrease in standings points from 2021 to 2022. This discrepancy underscores that the AB Coach’s metric is not perfect and cannot predict all outcomes with certainty. Nevertheless, as with other sections of this research, this information can prove valuable to general management in their endeavor to select the most suitable coach for their existing roster. For instance, a General Manager (GM) might seek a coach with a track record of success in developing young players or one capable of elevating a maturing team to the next level. The AB coaching data in this study can provide insights into which coaches have demonstrated success over the sixteen-year period, offering management a foundation for considering potential candidates for coaching positions or interviews. It is worth acknowledging that the sport of hockey is inherently a people-driven business, and GMs often gravitate toward coaches with whom they have established trust and previous working relationships. This element of trust can play a significant role in hiring decisions. Nevertheless, it is imperative to recognize that coaches wield substantial influence over the implementation of management’s decision-making strategies within the organization. Consequently, selecting the right individual to lead from behind the bench holds paramount importance in the pursuit of constructing a championship-caliber team.

In conclusion, the AB coaching metric can serve as a valuable analytical tool to aid GMs in making more informed coaching decisions. While it may not provide foolproof predictions, it can enhance the likelihood of making choices that align with the team’s objectives and aspirations for success.

Advanced Bracton General Management Evaluation Metric

This part of the study presents the Advanced Bracton (AB) General Management Evaluation Metric, which extends the analysis of individually registered AB scores beyond the realm of on-ice performance. The GM evaluations represent a crucial component in the formulation of an overarching organizational philosophy within the context of hockeyfreeforall.com research. The examination encompassed every decision made by each NHL franchise’s general manager from 2007 to the present, including trades, free agency acquisitions, and draft history. The CapFriendly.com General Managers tool played a pivotal role in facilitating this analysis. The methodology employed for GM evaluations paralleled the approach used in the previous Coach’s study conducted during the preceding summer. General Managers were assessed and compared regardless of their tenure, with specific predetermined criteria outlined herein. To address cases where GMs departed mid-season, credit was often attributed to them for the entire season, as interim GMs were deemed unlikely to make substantial roster changes during their brief tenures. Four key evaluation dimensions were employed, which included trade history, free-agency history, draft history, and overall AB team success during their tenure.

For more tenured GMs like David Poile and Lou Lamoriello, the analysis extended as far back as CapFriendly’s documentation allowed, as data regarding trades and free agent signings before the year 2000 was often limited. Consequently, draft assessments considered the year of the final trade recorded on CapFriendly. For instance, Glen Sather’s last documented trade occurred on March 6th, 2004; thus, only his 2004 draft with the New York Rangers was included, while drafts from 1980-1999 with the Edmonton Oilers, as well as 2000-2003 drafts with the New York Rangers, were excluded. To ensure consistency in the analysis, all players drafted in rounds beyond the 7th were classified as 7th round selections for formatting purposes. Notably, due to their recent appointments and a dearth of available data, Calgary Flames General Manager Craig Conroy, Nashville Predators General Manager Barry Trotz, and Philadelphia Flyers GM Daniel Briere were not included in this study, but they will be incorporated as more data becomes available. In total, this study encompassed seventy-nine current and former general managers, as delineated in the list provided below. The AB General Management Evaluation Metric, in conjunction with the Coach’s study, contributes to a comprehensive understanding of the factors influencing organizational success in the realm of professional hockey.

The research involved with this study was aimed to identify which GMs value similar traits and characteristics of hockey players as dictated by AB. The decision-making of every GM was evaluated in trades, signing off free agents, making draft selections, and the results of their coaching moves. The full list of statistics compiled on all seventy-nine men is shown below, as the graphic actually depicts all of Arizona GM Bill Armstrong’s decisions in relation to AB and overall.

Similar to coaches, all GMs are different in their managerial actions due to a variety of factors. These could include ownership situation, salary cap configuration, talent pool, or simply managerial tendencies or strategy. Below is a ranking of the active GMs in the NHL today based on a variety of criteria shown in the graphic on the previous page.

Once more, it’s essential to clarify that this ranking should not be interpreted as a criticism of the general managers (GMs), scouts, or team members within these NHL organizations. The purpose of this analysis was simply to identify which GMs have led their organizations in alignment with AB principles or similar guiding philosophies. It’s noteworthy that the individuals at the top of this list have enjoyed long-standing tenures with their respective organizations, a rarity in an industry characterized by high turnover rates each season. This correlation with sustained success, as seen in teams like Tampa Bay, St. Louis, and Boston, is likely not a coincidence. Industry experts have personally shared with me that ultimately, a GM’s role involves assembling a roster capable of making the playoffs, with outcomes beyond that stage being unpredictable. The information presented in this study thus far can be viewed as a blueprint for achieving sustained success in the NHL. I remain eager for the opportunity to test these theories in a professional setting. The upcoming section of this paper will delve into agency relationships and contract negotiations, offering insights from both perspectives, leveraging tools stemming from AB research.

Arbitration Analyzer and Agency Analysis Research

This part of the study introduces a significant advancement in managerial decision-making resulting from nearly a decade of research conducted at HFFA. The primary innovation is the development of the Arbitration Analyzer tool, which integrates the AB metric with actual pay-performance data and incorporates an intricate examination of various parameters. These parameters encompass the AB score, AB score/actual salary ratio, salary versus points produced, points/AB ratio, salary adjusted for AB score, league-wide pay rank, AB rank, salary per AB rank, salary per point adjusted for AB score, pay rank relative to AB rank, and pay rank relative to salary adjusted for AB rank. Each parameter was evaluated using a “pass/fail” criterion based on the top and lower deciles within each category. The net result of passes minus fails for each player was tabulated, yielding a rank on a scale of 1 to 10, which allows for a comparison between actual salary rank and the number of net passes. This comparison aids in determining a comparable salary for players entering into contract negotiations.

While comparing the salaries of similar players provides a valuable strategy for contract projection, the Arbitration Analyzer formula presents a more robust approach for determining true player value. This is achieved by combining the AB metric with adjusted actual points to accurately represent a player’s production, cost, or contribution to their team. This approach provides a more comprehensive view, as it quantifies previously intangible factors. Notably, the analysis indicates that the best value in the NHL is typically found among high-performing young players during their rookie contract period. This insight allows for precise projections of what these players are likely to command once their rookie contracts expire, along with the associated cap implications. While players like Auston Matthews or Conor McDavid are expected to command top salaries, the market valuation of players like Anton Lundell, who scores highly on this ranking, may not align with their perceived worth. This discrepancy presents an opportunity for NHL teams, particularly those with lower scores on our scale.

The data strongly suggests that success in the NHL is contingent on assembling a roster with players who provide both tangible and intangible value. Teams lacking players who contribute intangible value are likely to achieve only mediocrity or worse unless they adopt a roster construction philosophy akin to peers who base their decisions (partially or wholly) on data similar to the AB metric at HFFA. The introduction of the Arbitration Analyzer tool complements the information discussed in the preceding section, serving as the primary instrument for contract analysis and guiding GM decision-making in the free-agent market. The Arbitration Analyzer tool offers a unique opportunity for player representation firms as they negotiate contracts on behalf of their clients. Having access to this information could prove invaluable during arbitration proceedings, as many NHL agents have indicated that arbitration often hinges on the presentation of compelling numerical evidence, even in cases where the arbitrator may lack in-depth knowledge of hockey. This represents the most immediate practical application of my research at a professional level, as it enables the preparation of detailed arbitration cases using AB information. Furthermore, akin to the GM and Coach study, this research investigates which individual agents represent the most successful AB-rated players and which agents are particularly adept at securing favorable contracts for their clients using the Arbitration Analyzer and AB metric. The list provided below is not a ranking of specific agents but rather a compilation based on the number of clients each agent represents and the average AB metric of their clientele.

It’s interesting to see that of all qualified agents with over ten clients, Shawn Hunwick, Philippe Lecavalier, Mark Gandler & Todd Diamond, Matt Keator, and Allain Roy are the only ones with positive average AB clientele. However, Allain Roy and his team at Roy Sports Group have almost double the number of clients as any other positive agent. His notable clients include Tampa Bay’s Brandon Hagel, Bufalo’s Dylan Cozens, New Jersey’s Nico Hischier, and Pittsburgh’s Ryan Graves, all of whom are career-positive AB players who have signed lucrative long-term deals.

What does a Stanley Cup Championship Team Look Like in Terms of AB?

Recently, a study aimed at identifying potential trends among the 16 Stanley Cup Champions concerning the Advanced Bracton (AB) metric was completed. Since the previous study on the Colorado Avalanche in 2022, there have been modifications made to the overall AB score formula. However, for the sake of maintaining data consistency within this article, the previous formula was utilized.

Upon conducting this analysis, it became evident that the Golden Knights met or exceeded the established thresholds for three out of the four AB trends typically observed in championship rosters. They narrowly missed meeting the criteria for the fourth trend. Remarkably, only two playoff teams managed to surpass the criteria for all four trends – the Seattle Kraken and the Boston Bruins. That each was projected to be the Eastern and Western Conference champions before the playoffs, respectfully, is a testimony of the potential for the AB to be used as a predictive analyst. Notably, the AB metric once again successfully predicted the eventual champion when the competition narrowed down to the final two teams. In the 2022-23 season, the difference in team AB scores between the Golden Knights and the Florida team exceeded sixty AB points, significantly favoring the Golden Knights. The table below presents the team AB scores of the past fifteen champions, along with the AB score of last season’s Golden Knights team.

Regular Season Team AB Scores of the Sixteen Stanley Cup Champions of the AB Era

Year	Champion	Score	Highest (Y/N)
2007-2008	Detroit Red Wings	+85.79	Yes
2008-2009	Pittsburgh Penguins	+27.38	No
2009-2010	Chicago Blackhawks	+119.55	No
2010-2011	Boston Bruins	+124.51	Yes
2011-2012	Los Angeles Kings	-11.17	No
2012-2013	Chicago Blackhawks	+94.52	Yes
2013-2014	Los Angeles Kings	+59.83	No
2014-2015	Chicago Blackhawks	+53.02	No
2015-2016	Pittsburgh Penguins	+74.84	No
2016-2017	Pittsburgh Penguins	+77.96	No
2017-2018	Washington Capitals	+27.91	No
2018-2019	St. Louis Blues	+26.26	No
2019-2020	Tampa Bay Lightning	+76.27	No
2020-2021	Tampa Bay Lightning	+28.50	No
2021-2022	Colorado Avalanche	+122.52	No
2022-2023	Vegas Golden Knights	+97.81	No

The average AB score of a Stanley Cup championship team of +65.846 was the first trend the Golden Knights eclipsed, as their +97.81 adjusted team AB score is the second highest team AB score of a Stanley Cup champion (behind only the 2021-22 Avalanche) dating back to 2010-11. This was also the fourth-highest AB of a championship team since the metric’s inception. They were one of nine teams to meet this mark prior to the playoffs, as those teams were Boston (+233.92), New Jersey (+128.13), Carolina (+84.83), Toronto (+72.90), New York Rangers (+75.09), Seattle (+105.13), Vegas (+97.81), Edmonton (+93.03), Dallas (+96.34). Also, like all but two prior champions before them, the Golden Knights also did not record the highest overall team AB score at the conclusion of the regular season, as this year’s Boston Bruins team had a +233.92 score, which was the highest recorded score of all time.

The AB Positive to AB Negative Player Ratio of the Sixteen Stanley Cup Champions of the AB Era

Year	Champion	Ratio (Positive/Negative)	Percentage Positive
2007-2008	Detroit Red Wings	13:9	60%
2008-2009	Pittsburgh Penguins	17:11	60%
2009-2010	Chicago Blackhawks	21:2	91%
2010-2011	Boston Bruins	17:7	70%
2011-2012	Los Angeles Kings	7:17	29%
2012-2013	Chicago Blackhawks	19:4	83%
2013-2014	Los Angeles Kings	14:7	67%
2014-2015	Chicago Blackhawks	15:8	65%
2015-2016	Pittsburgh Penguins	21:5	78%
2016-2017	Pittsburgh Penguins	17:6	74%
2017-2018	Washington Capitals	12:6	67%
2018-2019	St. Louis Blues	13:13	50%
2019-2020	Tampa Bay Lightning	13:12	52%
2020-2021	Tampa Bay Lightning	15:6	71%
2021-2022	Colorado Avalanche	18:8	69%
2022-2023	Vegas Golden Knights	20:7	74%

The graphic above shows the ratio of AB positive to AB negative players on the 16 Stanley Cup championship teams, as well as the percentage of total positive players on each team. The average prior to this season of 65.73% positive players per team was the second mark the Golden Knights exceeded. The Golden Knights’ positive to negative AB player ratio was 20:7 74%, the highest ratio recorded since the 2016-17 Pittsburgh Penguins recorded the same during their second straight Cup Finals win. Prior to the playoffs, of the nine potential championship teams, only the Toronto Maple Leafs (63%), and the New York Rangers (56%) failed to eclipse that mark.

Players on Stanley Cup Championship Teams with a +10 AB Score Or Higher

Year	Champion	Player Names	Respective Scores
2007-2008	Detroit Red Wings	Datsyuk, Zetterberg, Lidstrom, Samuelsson, Cleary, Filppula, Rafalski	+29.5,+18.07,+16.08,+13.47,+12.14,+10.84,+10.67
2008-2009	Pittsburgh Penguins	Malkin, Kennedy	+12.27, +10.17
2009-2010	Chicago Blackhawks	Toews, Hossa, Kane, Sharp	+14.73, +13.79, +12.60, +11.72
2010-2011	Boston Bruins	Chara, McQuaid, Horton, Krejci, Marchand, Lucic, Ference, Bergeron	+15.37, +14.77, +14.34, +12.49, +12.43, +11.05, +10.44, +10.32
2011-2012	Los Angeles Kings	Brown	+12.66
2012-2013	Chicago Blackhawks	Toews, Hossa	+16.17, +10.78
2013-2014	Los Angeles Kings	Kopitar, Toffoli	+18.36, +11.26
2014-2015	Chicago Blackhawks	Toews	+16.65
2015-2016	Pittsburgh Penguins	Kunitz, Maatta, Crosby, Hornqvist	+14.60, +12.06, +10.79, +10.30
2016-2017	Pittsburgh Penguins	Sheary, Schultz	+12.99, +10.11
2017-2018	Washington Capitals	NONE	NONE
2018-2019	St. Louis Blues	O’Reilly, Parayko	+14.75, +10.27
2019-2020	Tampa Bay Lightning	Cirelli, Point, Palat, Hedman, Kucherov	+17.97, +16.32, +13.33, +13.21, +11.96
2020-2021	Tampa Bay Lightning	NONE	NONE
2021-2022	Colorado Avalanche	Toews, Makar, Rantanen, MacKinnon, Nichushkin	+25.36, +24.48, +18.31, +13.90,+12.00
2022-2023	Vegas Golden Knights	Eichel, Martinez	+13.42, +14.61

The third trend in this study was that 73% of Stanley Cup championship teams had a center with a +10 or higher individual AB score recorded in the regular season, as well as two or more players who exceeded that mark. This year’s Vegas Golden Knights had center Jack Eichel record a +13.42 individual AB score, as well as defenseman Alec Martinez record a +14.61 score. Prior to the playoffs, of the seven remaining potential championship teams to this point, only the Carolina Hurricanes failed to accomplish this feat.

Average and Combined Individual AB Scores of Top-Pairing Defensemen of the last Sixteen Stanley Cup Championship Teams

Year	Champion	Pairing	Pairing Total AB/Average
2007-2008	Detroit Red Wings	Lidstrom (+16.08)/Rafalski (+10.84)	+26.92 (+13.46 AVG)
2008-2009	Pittsburgh Penguins	Orpik (+1.48)/Gonchar (-3.75)	-2.27 (-1.135 AVG)
2009-2010	Chicago Blackhawks	Seabrook (+8.82)/Keith (+9.20)	+18.02 (+9.01 AVG)
2010-2011	Boston Bruins	Chara (+15.37)/Seidenberg (-0.14)	+15.23 (+7.615 AVG)
2011-2012	Los Angeles Kings	Scuderi (-5.02)/Doughty (-2.9)	-7.92 (-3.96 AVG)
2012-2013	Chicago Blackhawks	Seabrook (+4)/Keith (+6.22)	+10.22 (+5.11 AVG)
2013-2014	Los Angeles Kings	Muzzin (-0.41)/Doughty (+4.05)	+3.64 (+1.82 AVG)
2014-2015	Chicago Blackhawks	Keith (+3.34)/Hjalmarsson (+9.08)	+12.42 (+6.21 AVG)
2015-2016	Pittsburgh Penguins	Letang (+2.06)/Maatta (+12.06)	+14.12 (+7.06 AVG)
2016-2017	Pittsburgh Penguins	Dumoulin (-0.12)/Letang (+0.26)	+0.14 (+0.07 AVG)
2017-2018	Washington Capitals	Orlov (+6.34)/Niskanen (+7.61)	+13.95 (+6.975 AVG)
2018-2019	St. Louis Blues	Edmundson (-2.26)/Pietrangelo (+1.77)	-0.49 (-0.245 AVG)
2019-2020	Tampa Bay Lightning	Hedman (+13.21)/Sergachev (+4.10)	+17.31 (+8.655)
2020-2021	Tampa Bay Lightning	Hedman (+1.45)/Sergachev (-0.28)	+1.17 (+0.585)
2021-2022	Colorado Avalanche	Makar (+24.48)/Toews (+25.36)	+49.84 (+24.92)
2022-2023	Vegas Golden Knights	Martinez (+14.61)/Pietrangelo (+6.60)	+21.21 (+10.60)

Prior to this season, the average combined top defense paring score of Stanley Cup Championship teams was +11.48. As Vegas’ Alec Martinez and Alex Pietrangelo only combined for a +10.60 average, this was the only one of our four trends that the Golden Knights did not exceed. Despite this, their +10.60 average was among the higher pairing averages of championship teams recorded to this date, as they fall only behind last year’s Colorado Avalanche (+24.92), and the 2007-08 Detroit Red Wings (+13.46). This year, they were not alone however, as the New Jersey Devils, Dallas Stars, and Edmonton Oilers also failed to reach this mark, leaving the Seattle Kraken and Boston Bruins as the only two teams to surpass all four benchmarks this season as mentioned before. In conclusion, the 2022-2023 Vegas Golden Knights were adequately constructed in accordance with the standards of AB championship teams of the past.

AB Results with Spearman’s Rho and Regression Analysis

In this research endeavor, an evaluation of the performances of all 32 NHL teams was conducted in accordance with the Advanced Bracton (AB) metric. This evaluation encompassed both retrospective and predictive aspects over the course of the study. Traditionally, at the conclusion of each NHL season, members of the hockey analytics community strive not only to assess the accuracy of their regular season predictions but also endeavor to make precise playoff forecasts based on regular season performance. Various methodologies are employed for this purpose. Since the inception of this study at HFFA, the Spearman’s Rank Correlation Coefficient Test, commonly known as Spearman’s Rho, has been utilized to assess the effectiveness of predictions, both retrospectively and predictively.

Spearman’s Rho is a non-parametric statistical test utilized to quantify the strength of association between variables. In this context, a correlation coefficient (r) value of 1 signifies a positive correlation, whereas r=-1 denotes a negative correlation. To execute this analysis, data spreadsheets were formulated for AB player statistics spanning sixteen NHL seasons. Special attention was given to ensuring the accurate assignment of players to their respective teams immediately following the conclusion of each specific NHL regular season. This meticulous data management process was undertaken to mitigate potential errors in the overall calculated team scores for each of the league’s 30 teams (2007-2017), 31 teams (2017-2021), and 32 teams (2021-present). Following the alignment of players with their respective teams, and after consolidating data for traded or released players, the player scores for each team were aggregated to derive a single numerical representation of the team’s AB score. This process was iterated to acquire sixteen years’ worth of team scores for every NHL team. Subsequently, all team score data was compiled into a master spreadsheet labeled “Team History 07-23.”

To elucidate the retrospective order of finish, the team score numbers from the end of the regular seasons spanning 2007 to 2023 were organized. Regression analysis was then employed to forecast the order of finish ahead of each season’s commencement. For the sake of data accuracy and test validity, the first year of predictive analysis was initiated in the third year of the study (2009-10). This decision was made to ensure a minimum of two years of data were available for regression modeling. The actual results were manually observed from the “Team History” spreadsheet and transcribed onto a separate sheet of paper, juxtaposed with the predicted rankings. Subsequently, this data was input into Microsoft Excel spreadsheets to perform the Spearman’s Rho tests.

A total of 68 Spearman’s Rho tests were conducted, comprising two tests per conference for each year over 16 years. An additional two tests were conducted for the divisions during the COVID-affected season. These tests were structured within Microsoft Excel spreadsheets featuring thirteen columns, each corresponding to elements of Spearman’s Rho formula. Column A listed the teams within their respective Eastern and Western conferences. Column B contained the predicted ranks based on the AB metric, whereas Column C reflected the actual final standings positions. To expedite data collection, the actual order of finish data was imported from the NHL.com standings pages and consolidated into a separate spreadsheet labeled “Team Standings 07-19.” Column D represented the difference between predicted and actual standings (B2-C2), while Column E squared the values in Column D. An auto sum operation was executed in Column E to determine the sum of squared differences between ranks. Column G represented the sample size, which varied by year due to NHL division realignment and the addition of expansion teams, such as the Las Vegas Golden Knights and the Seattle Kraken, to the league. Typically, the sample size equaled 15 teams in the Eastern and Western Conferences, resulting in N=15 for the majority of tests. Column H denoted N squared, equivalent to 225. Column I calculated N squared minus 1, yielding 224. Column J held the constant value 6, while Column K represented the numerator rho (the sum of squared differences from Column E, after auto sum, multiplied by the constant value in Column J). Column L represents the denominator rho, which is equated to N multiplied by N squared minus 1. Column M calculated the numerator divided by the denominator rho, yielding the value to be placed in Column N. Column N was the crucial determinant of predictive accuracy and was used in subsequent statistical analysis.

The ensuing table presents the results of the AB score analysis on both retrospective and predictive bases, spanning the period from 2007 to the present.

Obviously, the retrospective numbers were higher than the predictive ones, as explaining the past is easier than predicting the future. There’s something to be said for these results, however, as positive-performing teams have a better chance at making the playoffs. Below is a table illustrating a positive AB-performing team’s chances of qualifying for the playoffs over the course of the study.

As the table shows, the significant majority of AB-positive teams qualify for the playoffs, an average of 93% throughout the sixteen-year period. 94% of champions over the course of the study were positive AB teams, and the higher of the two finals teams has won the Stanley Cup in 14 of the 16 seasons (the two Los Angeles wins being exceptions).

Final Remarks – Epilogue

This paper was a combination of the research and methodology behind my life’s work, the Advanced Bracton Score family of metrics, and hockeyfreeforall.com. The philosophies behind my work are ones I believe will absolutely be successful in a professional setting, whether that be as the GM of an organization, a player agent, a coach, a scout, or a consultant. If any of my work interests you as a potential employer, please feel free to contact me via cell phone at 757-345-8128, or by email at abellabr@shu.edu. It’s been a lifelong dream of mine to work in the hockey world in some capacity, and I hope the information in this paper will help earn me an opportunity to aid an organization to the best of my ability. Thank you for your time and consideration.