Some notes for those unfamiliar:
- Percentiles are not linear! The difference between a 90th percentile player and an 80th percentile player is much larger than the difference between 80 and 70, or 55 and 45. Good illustration here:
- Patrick Bacon found that Replacement level is 19%. That is, anyone below 19% adds negative value to the lineup.
- The absolute worst players in the league are almost as detrimental to a team as the best players are positive.
- One-dimensional defensive defenceman and forwards are nearly always underrated by these metrics, but not to a huge extent. The opposite for one-dimensional offensive talents.
- With hockey following a Pareto distribution of 19/81, the rule of thumb is the top 20% of players in the league contribute 80% of the wins. Pro-rated to an equal team level, the top 4 players on each team account for 80% of the team's relative strength.
- With that in mind, it is extremely difficult to become a legitimate contender without high-end talent, particularly at the centre position.
- Free agent splashes on mid-roster players seldom live up to their contract because they get premium money for middling performance, especially once accounting for age curves. Well managed teams identify these players before they become "mainstream" to extract maximum value. Poorly run teams pay a premium for past performance, often to declining players. July 1, 2016 is the perfect example of this.
- Do nearly whatever it takes to obtain and retain elite talent, and surround them with cost-effective two-way depth. Paying elite money, especially with term, for non-elite players is, by far, the most sub-optimal move a team can make (Hi, Darnell Nurse!).