Introduction

A few years ago, I created a metric (which I have now named DEEP—Draft Evaluation using Exponentially weighted comparisons and Probability scoring; I'm open to better acronym suggestions) that is designed to measure the ability of teams and executives to draft. Each player is scored by comparing them against the alternatives that were available, and measured according to whether they provided more or less value. A score of zero is roughly neutral, while positive scores indicate good drafting ability and negative scores the opposite. Over the years, the methodology has improved (becoming more complex in this process), and this is the newest and best version that has addressed many of the issues in previous versions.

The metric currently covers every player drafted from 2009 to 2020 as well as some undrafted players (details are in the methodology). In total, 849 players are scored, which are attributed to 76 executives from all 30 teams.

Quick Results

Team Scores

Team Scores Excluding Undrafted

In short, Toronto and San Antonio are excellent while Sacramento and Cleveland are terrible.

Distribution of draft scores

Top 5 Draft Picks

​

Bottom 5 Draft Picks

​

Active executive draft scores (minimum 5 players)

Dashboard to view data for your chosen team, player, or executive

Full article (with methodology at the end)

Updates Since Last Year

For anyone that may have seen last year's data, there have been 3 major updates in response to some of the feedback I received:

• The introduction of probability scoring to discount lottery picks according to how obvious they were. Consensus first overall picks (and other high selections) were previously receiving incorrectly high scores for consensus selections.
• The inclusion of undrafted players if they made a team in the season following their draft year. Note: many undrafted players don't find success with the team they're attributed to. The dashboard allows you to filter undrafted players out of team rankings if you desire.
• The creation of a new and improved dashboard to view results.

ELI5 Methodology

This is my best attempt at a no-numbers-involved explanation of the methodology.

Each player is compared against the alternatives that were available. For example, to draft a player 1st overall, that team must forgo the players drafted 2nd, 3rd, and so on. More weight is given to players that were more likely to have been realistically considered. For example, if I was drafting 5th overall, the player that went 6th overall was a very likely alternative. The player that went 55th overall was a very unlikely alternative. Therefore, much more weight is given to the 6th overall pick than the 55th (which would receive essentially zero weight). Early draft picks emphasize comparisons against a small number of players drafted close after while later draft picks compare more evenly against many players, reflecting the greater uncertainty of drafting in the late first and second round.

For lottery picks, there is also an adjustment reflecting how obvious the selection was. For example, Anthony Davis (1st overall in 2012) was clearly a good draft pick, but it was also a universally agreed upon pick. It did not take any unique talent or skill to identify Davis as the best choice. Therefore, the selection is scored close to zero since it is likely that almost any team would have made the same choice.

In summary, the score of each player depends on:

• Their own performance
• The performance of the players taken after
• For lottery picks, how obvious the selection was (did everyone agree this player was the best available? Or were was many plausible options?)
• Their draft position, but only to determine which players they are compared against. High draft picks are compared mainly against players drafted nearby, while lower draft picks are compared more evenly against a larger number of players

The score of each player does not depend on:

• How good a player drafted in that position normally is; this is not a measure of whether players met lofty expectations or not
• How long they spend with the team they are attributed to. This metric is not intended to measure how well teams manage assets after the fact
• The performance of players taken before them. Since these players were not possible alternatives, their performance is irrelevant

Full Methodology

Described here in the second half of the article