With AFL clubs preparing for the Draft, Victoria University (VU) researchers are proposing a new approach to trade draft picks based on their true value according to teams' future performances.
Dr Jemuel Chandrakumaran used his PhD research to look at ways to modernise the Draft that will benefit both players and clubs. This study, 'Deferring draft picks: Empirical analysis of the AFL draft,' introduces a novel discount rate model based on player performance data, rather than historical trade behaviour.
Using data from over 900 AFL draftees between 2003 and 2016, the study finds that:
- Discount rates in the first year vary by pick number, ranging between 42% and 51%, and decrease over time as underperforming players are delisted.
- First-year player contributions are a poor predictor of long-term value, suggesting that clubs may benefit from deferring picks to future drafts.
- Historical trades show a consistent bias toward current-year selections, even when future picks offer greater strategic value.
Dr. Chandrakumaran explains, "Our research shows that if a team is determined to select a certain current year draftee, they are generally very quick to offer up future picks. Historical trade data shows that teams advancing up often discount future picks by as much as 120% in pursuit of immediate returns. This research helps to put a more accurate value on a future draftee."
In order to determine a discount approach and how it may function, researchers applied it to the AFL's existing Draft Value Index (DVI) (below) as the basis for the value of a pick in the current year.
For example, if team A wished to obtain team B's fifth pick in the current year (valued at DVI 1,878), they would need to release their pick nine in the current year (valued at 1,469) together with their first-round pick from the next year.
At best, if team A finishes last in the ladder this year and obtain pick one next year, team B will make a significant profit of 1,166 DVI points in the trade (1,575 (pick one next year) + 1,469-1,878) and at worst, breakeven with a slight margin of 85 (494 (pick 18 next year) + 1,469-1,878).
While the proposed metric might not necessarily allow teams to understand the exact value exchanged in each trade at the time it was executed due to the inability to predict next year's drafting order, it would allow them to hedge themselves according to their risk tolerances.
"Our model provides a more objective framework for evaluating draft pick trades. It empowers decision-makers to better understand the long-term value of their selections and potentially exploit inefficiencies in the market."
"Getting a fair distribution of amateur talent is not an easy task, that's why we need to rework the approach. For trades to always have parity in exchange, accurate denominations of value are required. We hope this research can support this and in turn reinforce the true purpose of the Draft," Dr Chandrakumaran said.
This research builds on previous Draft research by Dr Chandrakumaran.
