Thanks to the internet, there’s a plethora of information available about the NBA draft. We have more observers than ever — more prognosticators who say they can predict the future. Some NBA writers are defined by the draft itself, but which ones perform the best? Which mock drafts are the most accurate?
Tabulating the predictions before the draft started, which included last minute tweaks from DraftExpress and Chad Ford. We can run through the numbers on how they compare to reality. The first method of comparison is to simply look at how many accurate “hits” each draft predictor collected. This was much easier to do in the lottery where the players stand apart more, but it becomes nearly impossible to nail any down consistently after the lottery.
Sources:
Ford: Chad Ford
DE: DraftExpress
BR: BleacherReport
HH: HoopsHype
NET: NBADraft.net
NET (C): NBADraft.net (Consensus)
SB: SBNation
DraftExpress (DE) looks strong here even without a second round pick correctly predicted. NBADraft.net (Net) was the only one to predict multiple second round guys. But DraftExpress cleaned up in the lottery at a 50% rate, which is fairly remarkable for such a notoriously crazy night.
| Hits | Ford | DE | BR | HH | Net | Net (C) | SB |
|---|---|---|---|---|---|---|---|
| Lottery | 5 | 7 | 4 | 5 | 6 | 3 | 4 |
| First round | 6 | 8 | 5 | 5 | 6 | 3 | 5 |
| Second round | 0 | 0 | N/A | N/A | 2 | 0 | N/A |
*The consensus mock draft appears to be using crowd-sourcing; it used 2276 people.
However, the best method of comparison is probably using root-mean squared error (RMSE). Essentially, this is an “average” prediction error (where error is defined as how many picks you were off by) where a greater penalty is given to the larger mistakes. For guys drafted who were not in the mock draft, they were assumed to have been picked in the fictional 70th slot for a respective mock draft. (This is a sizable but not severe penalty for failing to predict someone will even be drafted.)
By RMSE, DraftExpress (DE) stands out. They have the lowest RMSE of the second and first rounds. Only Chad Ford fared better in the lottery, thanks to only one notable mistake (Nik Stauskas moving up) compared to two for DraftExpress (Stauskas moving up and Nurkic dropping lower). Both BleacerReport (BR) and SBNation (SB) held their own with respective scores, but HoopsHype (HH) and NBADraft.net (Net) were the clear losers.
| RMSE | Ford | DE | BR | HH | Net | Net (C) | SB |
|---|---|---|---|---|---|---|---|
| Lottery | 1.46 | 1.93 | 2.85 | 4.59 | 3.36 | 4.54 | 2.98 |
| First round | 5.56 | 4.37 | 5.53 | 7.12 | 6.81 | 7.84 | 5.48 |
| Second round | 14.9 | 12.8 | N/A | N/A | 13.4 | 15.8 | N/A |
As a final note, what separates Chad Ford and DraftExpress from the field is the ability to make multiple last second corrections. I already had their mock drafts uploaded from an hour to the draft, but they made a few corrections later on like Vonleh correctly dropping. It made all the difference for DraftExpress to take a significant lead. (Chad Ford’s updated mock draft only improved by a fraction.)
