Open ryantuck opened 7 years ago
I had the same issue. I concluded there's really no consistent way to determine the position based on stats either (how would you tell a WR from a TE? Receiving stats + height/weight ratio? Still unreliable)
I ended up running a script that parsed retired player positions out of Pro Football Reference's csv files (http://static.pfref.com/years/2010/fantasy.htm) and updated that player's position in Postgres. I joined the PG data with the csv's on unknown player's full names and PG birthdates equaling Pro Football Reference ages. Kinda an invasive solution, and not 100% accurate (I'm sure I updated some retired WR named Chris Johnson to have the RB position or something like that).
Overall it made my life a lot easier, but note there are costs if you decide to go this route. Let me know if you'd like to use my code.
@ryantuck feeling your pain. @CPOConnor would love if you would share your code... Would be much MUCH appreciated. anm5030@gmail.com
I noticed that retired players have both position and team labeled
UNK
:This is problematic when attempting to run historical performance analysis for all QBs and Donovan McNabb is nowhere to be found (after filtering my query on
position = 'QB'
).No immediate solution to this problem presents itself after perusing the data model. I imagine I could remove the requirement that a player is a
QB
and simply analyze anyone who had meaningful passing yards in a given game, and back into whatever team they were on at the time a la the solution provided in #176. This would be a good-enough workaround for my purposes.Am I missing something here, or is there no way to get a historical position?