Ranking algorithm
The ranking algorithm will score each fund in each distribution period, with the overall fund score (that determines ranking) calculated by summing scores across each period of the fund’s existence, and applying a discount function that boosts the importance of more recent scores. Each fund would thus have a score.
The percent of tokens distributed to a specific fund j in any period would be determined by dividing the fund’s score by the sum of all n fund’s scores (1 ≤ jn):
The proposed scoring algorithm:
Where t is the most recent period, a is the discount rate.
Note that the current Sharpe ratio is applied to each period’s score, and longevity is accounted for by incorporating a greater number of non-zero terms.
The table below outlines how the following factors are considered in order for the performance mining program to encourage investors to invest in the best funds from a more objective, sustainable perspective:
Fees generated
This measure incorporates both performance, as well as the nominal AUM of a fund. By including this as an important metric, fee-generating funds important for the protocol’s prosperity are supported.
Risk-adjusted returns
By taking stock of the risks that volatility of returns presents to investors, a measure of risk-adjusted returns makes sense as the basis of any long-term ranking. Measures such as Sharpe or Sortino ratios encapsulate both nominal returns and risk — as proxied by observed volatility (downside volatility in the case of the Sortino ratio).
Fund age
A longer track record typically makes a fund more suitable for investment, as it provides a signal as to the sustainability of an investment into the fund. A common phenomenon on Wall Street is the steady performance of short volatility funds that generate fantastic Sharpe ratios but that discount catastrophic tail risk (that eventually wipes out the fund). Fund age helps address this concern. Furthermore, older funds are more likely to have fund managers who have accumulated significant quantities of protocol tokens — exactly the cohort of users that we want to keep from moving to a potential competitor.
Copy link