Part 2 of 2 — Gradual Distributional Shifts · ← Part 1: All-or-Nothing Simulator
In the 2014-15 season, NBA players made about 35% of their three-point attempts. What if the mix of shot conditions had been slightly different — a few more open looks, a few fewer deep pull-ups?
Unlike the All-or-Nothing Simulator, this tool lets you shift the mix of shot conditions gradually — moving 20% of tight-defense shots to open rather than moving every shot at once.
How to test what gradual shifts in the shot-condition mix do to three-point make rates in the 2014-15 NBA season
Start with All players to understand the league-wide model, or switch to a player such as Stephen Curry to re-fit the six-variable model on that player's shots only. The shot count and baseline make rate under the selector update to show the sample you are analyzing.
The preset buttons apply a realistic partial shift (e.g. Shift toward open shots moves some contested attempts into open and very open categories rather than zeroing out contested entirely). To set your own mix, drag any slider and the other levels in the same group rebalance so the total stays near 100%.
Each variable shows a running total next to its sliders, in green when the mix adds to 100% and in red when it doesn't. You can't simulate until every group is valid. The reset link on any group snaps that variable back to the observed baseline mix without touching the others.
Click Simulate Make Rate to estimate the overall make rate under your new mix. The result card shows the predicted make rate, the shift from baseline, a 95% interval, and an estimated count of made shots. The quick banner above translates the result into plain English.
① shows the shot mix you set next to the observed mix. ② shows how the overall make rate moved. ③ shows uncertainty through 10,000 repeated simulations. ④ compares the full distribution of shot-level make probabilities. The calibration panel ⑤ shows how well the fitted model matches the observed data.
Use Reset All — or the Reset to baseline preset — to return to the current player's observed mix. Because shifts here are gradual, it is often worth running several variations (5% shift, 10%, 20%) to see how the effect scales before drawing a conclusion.
Try a scenario:
These presets apply a partial shift — nudging the distribution in one direction rather than replacing it. Drag the sliders yourself for finer control.
| Variable | Level | Shot % | Make % | N |
|---|---|---|---|---|
| Defender Distance | Very tight (0-3 ft) | 7 | 26 | 2,329 |
| Tight (3-6 ft) | 55 | 34 | 17,918 | |
| Open (6-9 ft) | 26 | 38 | 8,546 | |
| Very open (9+ ft) | 12 | 40 | 3,718 | |
| Shot Clock | Hurried (0-2 sec) | 6 | 20 | 2,102 |
| Pressured (2-4 sec) | 6 | 32 | 2,040 | |
| Open clock (4+ sec) | 88 | 36 | 28,369 | |
| Catch or Dribble | Off catch | 75 | 36 | 24,539 |
| Off dribble | 25 | 31 | 7,972 | |
| Period | Q1 | 24 | 37 | 7,696 |
| Q2 | 24 | 34 | 7,833 | |
| Q3 | 25 | 36 | 8,208 | |
| Q4 / OT | 27 | 33 | 8,774 | |
| Shot Distance | Corner 22-24 ft | 34 | 38 | 10,984 |
| Wing close 24-25 ft | 36 | 36 | 11,853 | |
| Wing mid 25-26 ft | 18 | 34 | 5,848 | |
| Deep 26+ ft | 12 | 27 | 3,826 | |
| Venue | Away | 50 | 35 | 16,111 |
| Home | 50 | 36 | 16,400 |
Use this section to see the bottom line first, then the explanation of what changed, then the charts below for the supporting detail.