Draft-to-Funding Markets: Putting Speculative Liquidity to Work
Prediction markets aggregate beliefs. Threshold crowdfunding coordinates contributions. Dominant assurance contracts solve the free-rider problem but need an entrepreneur to commit a subsidy. Draft-to-Funding (D2F) asks a different question: when a prediction market prices a funding outcome, the capital sitting inside it is denominated in the same units as the capital required to cause the outcome. Can the protocol act on this?
The motivating example
Imagine a $10,000 community fundraiser stuck at $7,000. A YES bettor has staked $2,000, a NO bettor $4,000. The neighbourhood has $13,000 of capital — enough to build the thing — just allocated to speculation. A single YES bettor could rationally top up the missing $3,000 themselves, win the bet, and walk away ahead. But with fifty potential Bettys, nobody moves: classic free-rider problem.
D2F automates Betty’s move. When pool depth D+Y+N≥G, the protocol drafts the minimal amount R=max\{0,G−D_{total}\} from the losing side (NO first, YES as backstop) and pushes the outcome across the line. The winning side splits the residual.
The mechanism in one breath
A campaign is a tuple (G,T,b,β): goal, deadline, donor perk per unit, draft-perk fraction.
Three primitives:
-
Donation: pay stake, receive perk
bon success, full refund on failure. -
YES bet: pay stake, split residual
Y+N−Rpro-rata on success, forfeit on failure. -
NO bet: pay stake, forfeited on success (drafted share earns partial perk
βb), refunded with bonusY/Non failure.
Why does anyone participate?
The mechanism reveals a dependency chain:
donations / mission-aligned capital → NO bets → YES bets
-
Donations enter motivated by perk
b, independent of pool state. -
NO is most attractive when the pool is thin (high
Y/Nfailure bonus prospects, high failure probability). -
YES is most attractive when the pool is deep (overshoot residual to extract).
Each layer underwrites the next.
Three regimes, one parameter
The perk b governs the mechanism’s character:
| Regime | Application | Distinguishing feature |
|---|---|---|
| b=0 | Curator action markets | Donation collapses into NO; market is YES vs NO |
| b>0 small | Threshold crowdfunding | Creator perks; donation and NO coexist |
| b>0 large | Token launchpad | Bonding-curve perks; donation dominates |
Donation dominates NO iff p⋅b⋅(1−βE[q_draft])>(1−p)⋅Y/N, where q_draft=C_N/N. Perk advantage on success vs failure bonus on the loss branch.
The bootstrap problem and the donor tax
In regimes B and C, thin early-stage Y means a weak Y/N bonus, leaving NO under-recruited. The fix: a tax k on donations, levied only on failure, routed to NO bettors.
π_{failure}^{NO}=n_i⋅(1+\frac{Y+kD}{N})
Donations seed NO failure-bonus depth before YES capital exists, igniting the dependency chain earlier. β<1β<1 rebalances if NO becomes over-recruited.
Curator-led action markets (b=0)
The most distinctive case. A local authority has legitimacy but lacks discretionary capital. An external funder has capital but won’t transfer it directly. The authority opens a prediction market on a concrete outcome (“will the dump on XYZ Street be cleared by 30 June?”), articulates the question, verifies resolution, never touches funds.
Mission-aligned NO from the funder seeds the market. Either a doer emerges, buys YES, performs the action, and collects NO on resolution — or the authority drafts NO capital and procures the outcome directly. The market itself motivates action. Capital flow never routes through the authority.
Token launchpads (b large)
Bonding-curve graduation: G≡L liquidity threshold, donations ≡≡ curve purchases with token perk b. The perk is structurally embedded and endogenously priced, placing the launchpad firmly in regime C. Phase 3 dynamics are rich: post-lock donations continue along the curve, with overflow flowing to liquidity.
What’s in the talk
-
The full settlement rules (success, failure, crowd-out from late donations)
-
Equilibrium analysis and the terminal-state tension (YES improves with depth, NO deteriorates)
-
The endogenised-DAC comparison
-
Three worked applications with curator design parameters (k,β)
-
Extensions: multi-threshold markets, continuous funding, coupled settlement across competing projects
IFT Research call recording: [link]
Shorter IRL talk from ETHPrague: [link]
Happy to discuss in the thread — particularly interested in feedback on the bootstrap problem,applicatiosn and extensions.