random numbers
Draftee
I have done an analysis comparing 8 of our phantom drafters to the actual result. I only chose those that did a full draft. I chose in order in which these Phantoms currently appear on the board. Apologies to don Patch, Jeff Poulter and Chris (and maybe others) but I got tired of further analysis and besides which the overall pattern was clearly consistent.
I wasn’t concerned with exact position, but rather whether the phantom drafters were more or less in sync with the thinking of the professionals. Thus as an arbitrary guide, I judged a phantom call within 10 positions of the actual as accurate.
Because the file was too big I broke it into 2 uploads draft positions 1-40 and above 40
1. Yellow were those that slid by more than 10 positions from actual (with actual position numbered)
2. Green were Bolters who rose more than 10 positions from actual (with actual position numbered)
3. Red were undrafted
4. Blue went to rookie
Overall, for the top 20, the phantom drafters were very good ie. They had a good idea of who the top candidates were.
From there, the predictions became progressively less reliable.
The 20-40 range had about a 50% accuracy (mean 10.1/20, median 10/20, mode 10/20) with a high of 14/20 and a low of 6/20. The inaccuracies were dominated by sliders ie. those who the phantom chose at a higher position than they actually went. The great majority of the sliders actually went later in the draft with only a couple failing to be drafted or going to rookie.
In the 41 and above, the phantoms were totally unreliable with the great majority being inaccurate. About half the inaccuracies were due to bolters (ie. Those who were actually drafted higher than the phantoms predicted). The other half were those who failed to be drafted or were drafted in the rookie.
The most notable bolters were Long, McLarty, Fisher and Lipinski.
The most notable sliders were Scharenberg, Sproule and Walker
I wasn’t concerned with exact position, but rather whether the phantom drafters were more or less in sync with the thinking of the professionals. Thus as an arbitrary guide, I judged a phantom call within 10 positions of the actual as accurate.
Because the file was too big I broke it into 2 uploads draft positions 1-40 and above 40
1. Yellow were those that slid by more than 10 positions from actual (with actual position numbered)
2. Green were Bolters who rose more than 10 positions from actual (with actual position numbered)
3. Red were undrafted
4. Blue went to rookie
Overall, for the top 20, the phantom drafters were very good ie. They had a good idea of who the top candidates were.
From there, the predictions became progressively less reliable.
The 20-40 range had about a 50% accuracy (mean 10.1/20, median 10/20, mode 10/20) with a high of 14/20 and a low of 6/20. The inaccuracies were dominated by sliders ie. those who the phantom chose at a higher position than they actually went. The great majority of the sliders actually went later in the draft with only a couple failing to be drafted or going to rookie.
In the 41 and above, the phantoms were totally unreliable with the great majority being inaccurate. About half the inaccuracies were due to bolters (ie. Those who were actually drafted higher than the phantoms predicted). The other half were those who failed to be drafted or were drafted in the rookie.
The most notable bolters were Long, McLarty, Fisher and Lipinski.
The most notable sliders were Scharenberg, Sproule and Walker