Draft Evaluation Metric #619
Conversation
| duplicates_removed = [] | ||
| for pick in draftData: | ||
| pick_id = (pick.round_num, pick.round_pick) | ||
| if pick_id not in duplicates: |
There was a problem hiding this comment.
Would there ever be duplicates in this data?
There was a problem hiding this comment.
When I printed out league.draft, the draft picks would repeat so that the set was double the actual size, sometimes even triple. I couldn't find why, so I decided to just filter it out with this method.
| draftscore = actualsum - projectedsum | ||
| avgdraftscore = avg_actualsum - avg_projectedsum | ||
|
|
||
| #Weighting average points vs total points | ||
| real_score = 0.2 * draftscore + 0.8 * avgdraftscore |
There was a problem hiding this comment.
I feel like a draft score should incorporate the pick number in some way. Screwing up your 1st round pick should hurt you much more than screwing up your 10th round pick, for example. Simply using points wouldn't necessarily pick up on this, as a QB will have the largest raw points but probably isn't drafted until a few rounds later. Perhaps using pick value to weight each pick (or something similar to this).
There was a problem hiding this comment.
This is a great point, I'll incorporate this.
There was a problem hiding this comment.
This is a really cool idea, thanks for submitting a PR!
High Level Comment
- Make sure to remove your tester.py from PR and it has your creds showing
- remove .DS_Store files
- I think it would be good to move all of this function logic to utils.py so it doesn't bloat the league.py class. Then can just pass necessary data to function.
I developed a metric to evaluate how well someone drafted, based on how their projections at the time of draft compared to their actual results. More specifically, I collected the drafted players for each team, and summed up the total projection and total actual scores of all the players, to then reduce to a single team score.