Introducing The 5 Newest Creation Stats

From Self-Creation to Assist conversion rates, and a brand new advanced Assists metric, 5 new creation stats driven purely by tracking data. Database can be accessed on the 'Stats' page on the home menu, or by clicking here.

Morey Assist Rate

Definition: % of a player's assists that are Rim, 3PT, and FT assists.

All assists are not made equal and all shots are not equally efficient. Assisting a late shot clock, contested mid-range jumper isn't considered as efficient as an open corner 3. While this stat isn't designed to showcase creation or playmaking ability and doesn't account for shot quality, it does attempt to quantify "Morey Assists" (also known as High Value Assists) into a one number percentage. Morey Ball, named after former Rockets GM and current Sixers GM Daryl Morey, consists of statistically the 3 most efficient shots in basketball: Rim, 3PT, and FTs. This stat takes those shots and applies it for players' assist locations.

Assist Conversion Rate

Definition: % of a player's potential assists converted to assists.

Traditional assists are a very flawed stat. While I went into which shots are more efficient than others above, there's also a possibility that a player's assist numbers are heavily tanked playing next to bad teammates who are unable to convert on passes. This is evident with Shai Gilgeous-Alexander and his 46% assist conversion rate, which is 7% below league average and 9% lower than Chris Paul's rate. This stat isn't perfect either, no one number metrics are. It's flaws are that it tends to make frequent lob passers like CP3 and Trae Young's conversion rates seem inflated, especially in comparison to drive and kick playmakers like SGA and Fox, when that's likely not the case. However, when used in the right context, the stat can still provide valuable information.

Self-Creation Rate

Definition: % of a player's FGA self-created, defined as where the ball was held for 2 or more seconds and dribbled before the shot attempt.

While the NBA's % of FGM unassisted stat is good at adding context to how many of a player's baskets are being directly assisted or not, I didn't think it's the best way to measure how many of a player's buckets are self-created or not. Using PBP tracking data pulled from NBA stats, I was able to filter for shots where the ball was held for 2+ seconds and dribbled 1+ times. Enter self-created FGA, which when divided by a player's total FGA, gives us their self-creation rate. Total self-created FGA and self-created eFG% are both also available on my stats page. The stat tends to undervalue the amount of post-ups that are self-created shots, especially face up attempts, which should be noted as a drawback. Thus, post hubs like Jokic and Embiid have lower self-creation rates than they likely should.

Shots Created On-Ball

Definition: The number of shots directly created through a player for a team, using tracking data.

Using the very new self-created FGA, potential assists, and secondary/hockey assists, I was able to come up with a formula to track the total number of a team's FGA a player is directly contributing to with the ball in is hands. While passing/assisting ≠ creating or playmaking, and the self-created FGA stat I've made is not perfect by any means, this stat simply attempts to estimate the amount of offense statistically running through a single player while they're on the court using tracking data. Naturally, there is a correlation between players time of possession and shots created on-ball. The stat should not be used to interpret the skill or overall value a player is providing on the offensive side of the ball, but instead just something that's interesting when used with context and understanding the boundaries of tracking stats.

Modified Assists

Definition: A player's assists adjusted for league average conversion rates and weighted for 3-point makes being statistically 1.5x more valuable than 2-point makes, using a model similar to eFG.

As explored through Morey Assist Rate and Assist Conversion Rate, traditional assist numbers can often be misleading. Once again, all assists are not made equal and all shots do not have the same statistical output, and a player's assist numbers could also be tanked playing next to bad play finishers unable to capitalize on potential assists at a competent rate. While still unable to account for shot quality, Modified Assists attempts to bridge the gap between situations and add context when comparing players' assist numbers.

The Modified Assists formula first extrapolates 3-point potential assists using a player's 3-point assist rate (amount of assists that are 3-pointers) and total potential assists. Then, it converts for league average conversion rates. After that, it weighs 3-point makes being statistically 1.5x more valuable than 2-point makes using an eFG model.

The stat isn't designed to showcase creation or playmaking ability, and like any other metric is not perfect. It has it's flaws. It doesn't account for shot quality or specific shot location, and also doesn't have an exact number of 3-point vs 2-point potential assists and instead has to estimate the number by assuming trends continue with 3-point assist rate. These are both potential future enhancements for the stat, however, there is no data yet publicly available that track these.

Despite this, Modified Assists is still subjectively better to use than traditional assists or raw potential assists. It's designed to create a level playing field for all players by using league average conversion rates from teammates, and also weights for the greater statistical output of 3-pointers in comparison to 2-pointers. While it does have its drawbacks and limitations, it better depicts the value of a single assist in my opinion. Again, it isn't created to show players' creation or playmaking through a single number. But when used in the right context, not just this stat but even all 5 of them, they can all provide valuable info that the traditional assist simply cannot.