Instead of market buying YFI as suggested in banteg, et al, utilize an auction system in which purchases can be made from the open market, thereby increasing liquidity for YFI and not disrupting normal trading quantities should institutional / bulge bracket transactions need to occur.
This also solves the issue of end users psychological hesitance in buying YFI (due to price), futures index would be 1/10th of the spot YFI price (see graph below)
Pick a continuous time period (e.g. some governance epoch, or as proposed in my other proposal for fees, a ‘roll over date’ see, ‘Dynamic Surcharge Fees’, for which the treasury/yearn commits to purchasing ‘X’ number of YFI (where X is dependent upon some % of allocation of the budget).
This creates the market perception similar to that of an ‘earnings call season’, whereby traders will take positions. This improves liquidity and price discovery as well.
TL:DR: use kee3pr bots e.g. turbokeeper to run the auctioning or liquidation engine for the auction. Can be extended to utilize gnosis oba solver which computes off chain for example order book auctions.
A naiive (re: hacky) representation of what a constructed, panama-style, futures index for YFI would look like: https://sambacha.github.io/yfi-futures/