Beyond the catalogue: 67 new computable markets in one page

Finance theory allows exactly two sources of excess return in a market with no ticker. Be cheaper to operate: your trucks, cold room, licence or exit channel cost less than the marginal bidder's, so a deal that is negative for them is positive for you. That is comparative advantage, and four of HyperC's seven reject reasons live there — scale, operational fit, capacity, time. Be right where others are wrong: their estimate, rule table or attention left value behind. That is the other three — attention, prediction, complexity — and it is the only side that leaves a recorded history of passes, which is what makes it learnable.

P34 learns both from the same two tables. The money tape (realised prices, fees, holding costs, write-offs) prices your costs, so the model learns the operator's value of a deal, not the market's. The historical menu with its declined rows prices their mistakes. In low-margin markets the larger half of the edge is the false positives you refuse: in our synthetic benchmark the same 20-date menu that bled a tuned regressor to −$62.6k left P34 at +$25.8k, mostly by declining.

The screen found four ways a market fails the thesis, each with a live example:

• No consumer, clean comps, open access — graded coins, vacant land: nothing removes supply, so capital competes the edge down to fees.
• One irreducible shock — used mining rigs (hashprice), bulk wine (a multi-year oversupply): a one-line rule does as well as a model.
• Feedback slower than the regime — timber, small-business acquisitions, Reg CF: outcomes arrive in months or years; only pooled histories can reach the data floors.
• Structure without access — hotel bed-bank inventory (contract-barred), and the licensed markets (pawn, cannabis, alcohol, medical devices, carbon) that belong on the enterprise track with counsel first.

Next: (1) publish the 67 as blueprints on the catalogue's waiting list with their evidence state; (2) build the snapshot crawlers for the top markets whose declined options are not yet recorded by the venue; (3) run shadow tests with one operator each in government surplus, the flower clocks and used-phone lots — the three where the venue already records every pass. Nothing here is a forecast of returns; the evidence base is one production deployment and a synthetic benchmark, and the thesis is that a market can fail the test.

Sources: HyperC catalogue, P34 API docs — candidate waiting list, Computable Markets working paper, Royal FloraHolland 2025, GovDeals by the numbers, B-Stock mobile storefronts, Faire; full per-market sources in the workbook P34_candidate_markets.xlsx.