We have recently tested the Advanced Bracton (AB) score against the current construction of the National Hockey League (NHL) with respect to original draft round and time in the league. Our thesis was that players with higher AB scores would experience greater career longevity in the NHL than players who scored poorly. We were fully expecting to find a close correlation of the variables, given the astounding, albeit retrospective, association we found between the AB and success in the league over the 2014-2015 and 2013-2014 seasons.
To our surprise, we found absolutely no correlation whatsoever between either AB and career stint or AB and round drafted……..That conclusion is where the study gets VERY interesting and the data have profound implications. Most notably, the lack of correlation of these variables indicates that teams are roughly equally likely to achieve high levels of mistake minimization from younger (and assumedly less expensive) players as they are from older ones. As players performance in their ninth year or greater begins to lapse as age advances (table 3), NHL teams may be habitually paying outsized salaries to older players unnecessarily and are doing so in approximately 125 cases heading into the 2015-16 season. Said another way, more than one sixth of the players in the NHL should not be commanding anywhere near what they are getting paid or at least could be replaced by players that contribute positively at the margin at assumedly a far lower cost to teams and, by extension, the fans. We think our data supports these conclusions (see below).
We complied data linking AB score to draft round as well as numbers of years in the league a player has played. We used hockeydb.com and other sources to generate the information. We then amassed the following tables (n=659):
The first table depicts the number of players currently in the league, how many were drafted as sorted by round, how many have an AB score of less than -2.50 the representative percentage, the number of players who have AB scores higher than 2 and their percentage.
The data show that even though there was a 38% chance that a player drafted in the first round would make an NHL team he had a reasonably equal likelihood to impact an NHL team in either an outsized positive or negative manner. That means negative AB players could be replaced in theory with positive ones with similar point production with similar experience at nearly all levels.
|Round||Players||AB < -2.50||AB > 2|
The second table shows the number of players sorted by years in the league relative to the rounds that players were picked (or undrafted) with associated AB scores of either below -2.5 or above 2.0 respectively. The data suggest that at nearly every experience level, the number of outlying negative AB players outweighs the number of outlying positive ones. Additionally, most first round picks seem to earn playing time at the NHL level with roughly half the teams employing such a player for at least 10 years. The numbers for the second round are about half of the first round and so on. Of note, there is a better probability of an undrafted player making an NHL team than a third rounder or below.
|YEARS IN LEAGUE VERSUS AB|
|Year||Players||Round 1||Round 2||Round 3||Round 4||Round 5||Round 6||Round 7||Round 8||Round 9||UND||AB <-2.5||AB>2|
The final table shows the number of players in the NHL broken down by positive or negative AB scores. The data suggest that even though there does not appear to be a correlation between AB score and time in the league, there may be a point in time, roughly year nine, where a negative correlation arises and continues as a player ages. The negative AB scores in the first years in the league can probably be explained by the learning curve process.
|YEARS IN LEAGUE VERSUS POS OR NEG AB|
|Year||Players||NEG AB||POS AB|
A discussion of the results from this study relative to the AB score include the following points;
- Although the AB score is retrospectively powerful in determining player and team success at generating goals at the margin either scored or avoided, there appears to be no correlation between mistake minimization and rewards from general management in contracts for players who consistently post a positive AB score over the course of careers. The reversal of how and why this persists could represent the biggest sea change for hockey general management in the period prior to upcoming labor negotiations.
- Players that post an outsized AB score are rarer than ones that cost their teams goals at the margin. Should they be paid accordingly or at least replace low AB players?
- Players entering their ninth year in the NHL, for whatever reason (age, physical deterioration etc), begin to “lose a step” relative to their younger brethren
- Players drafted in the first or second rounds of the draft on average enjoy more longevity than those drafted in the third round or lower. While this may seem obvious, there are more undrafted players in the league currently than third rounders or below.
- Stockpiling early draft picks increases the odds that a team will draft, and secure, high AB players early in their careers and have the options to move players with consistently low AB scores after roughly year 3.
Conclusions by point
- Point 1 implies that several teams in the NHL place relatively less value on character traits such as (but NOT LIMITED TO) net penalty margin than they do other factors. The inattention to things like the AB score, if it eventually demonstrates its power as a prospective predictor of success, costs teams goals at the margin and by extension hockey related revenues related to regular season and playoff gates as well as the factor of HRR that is derived as a multiple of tickets sales. We are not sure why teams would rather employ and pay players prone to making more mistakes than they generate instead of players who do not, assuming a replacement players exists that could breach any point production gap.
- Costing goals at the margin is less hurtful to teams if they are not paying top decile salaries to players who are consistently making more mistakes than they generate.
- Older players often times can be replaced by younger less expensive ones who are equally as adept at either producing or conserving goals at the margin as their older counterparts.
- Undrafted players are twice as likely to be low AB players as high AB players. This result is not wholly dissimilar to that found in third round players and below. Could (and should?) AB be used as a differentiator in late rounds when considering a player? Why is it not, all other things being equal? If scouts are doing their jobs, why not gamble on late round picks with high AB players rather than low ABs?
- Teams with three years of experience with a player are well positioned to decide whether such player is a fit within their organization at roughly that time frame.
It appears from the data that while scouting in the NHL is reasonably good for the top 60 draft eligible players in any one year, it seems to falter in rounds 3-7 to the point where teams have as much if not more success with undrafted players than they do with the ones taken at later levels. In order for scouts to add depth to their analysis, we are advancing the AB metric as manner by which players may be differentiated at a young age given that such players are rarer in the NHL than are their more mistake prone counterparts. Moreover, while the data do not correlate AB score with either time in the league or round drafted, we cannot surmise how and why a team would rather employ and pay, sometimes handsomely, players that make more mistakes than they generate if a less expensive similar point producing players exists as a replacement. Teams that have amassed rosters replete with such players have been very highly retrospectively correlated with making the playoffs. We think the drafting process, given the success of first round picks at making NHL clubs, should be aligned with the philosophy that mistake minimization wins hockey games.