‘Randomized’ Seeding in Fantasy FIRST Drafts
Issue:
Seeding order assigned in any draft, especially those of District drafts greatly impact the performance of a Fantasy FIRST team in their respective District Championship.
Important points:
- The difference between two consecutive seed positions at a given event will not necessarily have a resultant score that differentiates in a linear fashion.
- Seeding positions are also not comparable between events. For example, certain events have more dominant teams at the top end, resulting in a wide swing in the max vs. minimum point potential. Other events may have a wider distribution, with a deeper roster, allowing late seeds to pull in more fantasy points from their position in the draft, compared to other shallower events.
Data Analysis:
I compiled a
spreadsheet of 2015 Fantasy FIRST data to analyze the impact of seed order on draft outcomes.
- While scoring data is easily available for manipulation at any given event, seed data is not. *I'm tracking seed position this year*
- The important piece of data to note, is that fantasy teams are not statistically likely to move very far from their original rank, based on initial seed.
-In a 10 person draft, a fantasy team will on average only move ~3 seed positions (absolute), in rank from their initial seed.
Proposed Solution:
In order to get one step closer to similar seed benefits across events, the average seeding order should be statistically uniform. This means, if a team receives a first seed at one event, they should be statistically likely to receive a last seed at another to match. Their average seed order across the total events in the specified pool should be nearly the same as all others Fantasy teams. While the proposed solution is not a perfect one, it should help to remove some of the statistical advantages given out to specific teams, based on a system with ‘entirely’ random seeding generation.
Here’s the algorithm that would need to be developed:
You have 7 Districts, to place 10 Fantasy teams (A-J) in. There can’t be multiple instances of a Fantasy team in any given District. The goal is to maintain a similar average seed across all fantasy teams. You don’t care how the distribution resolves itself, but it will likely naturally result in a bell curve. The algorithm should randomly generate solutions to the table with a *small* average seed deviation.