Stephen Curry will Break the NBA Career Three Point Record

By Benedict Brady

Stephen Curry is in the middle of the single-best three pointing shooting stretch in NBA history. While Curry has been one of the greatest shooters in the league throughout his career, this year he has taken it to another level, averaging over five threes made per game. While these peak numbers will probably not continue for the rest of his career, Curry is still well ahead of pace to break the NBA career three point record of 2973, set by Ray Allen. Motivated by this, I attempted to get a better handle on the pace of Curry’s three pointers by modeling his career off other three point shooting greats.

There are three major input variables to estimate when modeling career three point numbers: average three pointers made per game (3PM/G), games played per year, and years played total. This model uses an aging curve to project 3PM/G for Stephen Curry for each year that he plays in the NBA. It will then assign probabilities to both games played per year, and years played total to generate percentages for Total 3PM values.

The first step is to construct an aging curve for Steph Curry’s 3PM/G. To do this, I considered a set of similar players and the Delta method. This method is essentially measuring season by season change for every player, and using these values to determine how Curry’s career will change from year to year given his age. In order to build a curve that more accurately reflected Curry’s three-point heavy style, I looked at all players who had made over 175 threes in a season (71 in total).

Next, I needed to adjust the magnitude of this curve so that it properly reflects Curry’s career. The best way to do this to minimize the square of the distance between the Curry’s current curve and the the projected values through adjusting the y intercept. This minimization does include the current year, but it is weighted to reflect the fact that only 62 games have been played. Finally, there was not enough data available to model the age 39 season, so I approximated the slope by repeating the decrease from the age 37 to age 38 season.

The final step is to simulate Curry’s career three point totals. I approximated the number of years that Curry will play and the number of games he will play per year by simulating his career 10,000 times. To model his career, I used the comparable players from FiveThirtyEight’s CARMELO projections, as well as the top ten three point shooters of all time to get a range of data points. I then assumed that games and years are distributed normally with a few exceptions. Games are obviously capped at 82, and I limited years to the age 39 season as there is very little data for Curry’s comparables in their age 40 seasons. For games, the set of 19 similar players averaged about 70 games per season with a standard deviation of 16 games per year. This matches closely with Curry’s 69.3 games played per year up to this point in his career. For age, the retired similar players played until 37.4 on average, with a standard deviation of 1.7 years. We can then simulate Curry’s career from this point out 10,000 times in R.

Curry breaks the current career 3PM record in roughly 86% of the simulations. Additionally, Steph has taken an unprecedented jumping in three point shooting volume this year, and because of this the projection is currently underestimating his performance. It is seeming increasingly likely that Curry will have no trouble breaking the NBA three point record.