Chapter 28: Key Takeaways

The SMART Framework for Market Questions

Specific: Every question must identify a precise event, entity, and measurement. "Will the economy improve?" fails; "Will US real GDP growth exceed 2.0% in Q3 2026 per the BEA advance estimate?" succeeds.
Measurable: Resolution must depend on objective, quantifiable criteria with an identified data source.
Assessable (Resolvable): The answer must be determinable in practice, not just in theory. Avoid questions whose resolution data does not actually exist.
Relevant: The question must attract informed traders who have genuine private information to contribute.
Time-bound: Every market must have a clear resolution date or a triggering event with a backstop deadline.

Resolution criteria define a binding agreement between the platform and its traders. They must be explicit, objective, and robust to edge cases.
Use the resolution source hierarchy: Tier 1 (government statistics, regulatory filings) is preferred over Tier 2 (institutional data), Tier 3 (major media), Tier 4 (domain-specific sources), and Tier 5 (platform/community judgment).
Always specify a backup resolution mechanism and conditions under which the market resolves N/A.
Stress-test every market against the edge case checklist: subject ceasing to exist, metric discontinuation, metric redefinition, partial outcomes, reversals, temporal technicalities, reasonable disagreement, and self-referential influence.

Ambiguous "or": Always specify inclusive ("at least one of") or exclusive ("exactly one of") disjunction.
Scope creep: Pin every key term to a specific standard or definition at market creation time.
Moving goalposts: Define success criteria with objective, pre-specified thresholds.
Self-referential markets: Avoid markets whose outcome depends on their own price.
Insider-triggerable markets: Focus on outcomes that no single party can unilaterally determine.
Temporal ambiguity: Use explicit dates, times, and time zones (e.g., "before 11:59 PM ET on March 31, 2026").
Negation: Prefer positive framing; let the price reflect the probability of the negative outcome.

Binary markets maximize liquidity concentration and are simplest to understand. Best for yes/no and threshold questions.
Multi-outcome (categorical) markets must be mutually exclusive and exhaustive. Always include an "Other" catch-all. Limit to 5--10 outcomes for practical liquidity.
Scalar markets capture full distributional information without requiring pre-specified thresholds but are more complex to implement.
Bracket markets discretize continuous outcomes. Use equal-probability brackets (based on prior distributions) rather than equal-width brackets to maximize information content.
Validate mutually exclusive and exhaustive properties formally: $\bigcup O_i = \Omega$ and $O_i \cap O_j = \emptyset$ for $i \neq j$.

The lifecycle phases are: Creation, Seeding, Active Trading, Approaching Resolution, Resolution, Settlement, Archived.
Creation timing should align with periods of meaningful uncertainty and public interest.
Seeding requires initial liquidity (via AMM) and an initial price reflecting a reasonable prior -- not a default 50/50 split.
Resolution should be prompt, transparent, and appealable with a defined dispute window (24--72 hours is standard).

The cold start problem is fundamental: markets need traders for liquidity, but traders avoid illiquid markets. Subsidization via AMM funding is the primary solution.
Initial subsidy should reflect the base rate: setting a 50/50 price when the base rate is 5% wastes subsidy to easy arbitrage.
Declining subsidies using exponential decay $s(t) = s_0 \cdot e^{-\lambda t}$ gracefully transition from subsidized to organic liquidity.
Subsidy allocation across markets should maximize $\sum V_i(s_i)$ subject to a budget constraint, weighting by information value, expected participation, and marginal liquidity impact.

Participation incentives (sign-up bonuses, trading rewards) overcome the barriers of time cost, capital risk, and uncertainty aversion.
Accuracy incentives (Brier score leaderboards, accuracy bonuses) align trader behavior with the market's information-aggregation purpose.
Liquidity incentives (fee rebates, liquidity mining) compensate market makers for adverse selection risk.
Reputation systems serve multiple purposes: signal quality, governance, access gating, and social proof.
Gamification should reward accurate forecasting, not just trading volume.

Participation metrics: Unique traders, trading volume, trader diversity (Herfindahl Index $= \sum s_i^2$; lower is more diverse).
Liquidity metrics: Bid-ask spread ($\frac{p_{\text{ask}} - p_{\text{bid}}}{p_{\text{mid}}} \times 100\%$), depth, slippage.
Accuracy metrics: Brier score ($BS = \frac{1}{n}\sum(f_i - o_i)^2$), log score, calibration curves.
Resolution quality metrics: Dispute rate, resolution time, N/A rate.
Composite quality score: $Q = \sum w_j \cdot \tilde{m}_j$ with recommended weights: calibration 30%, participation 20%, liquidity 20%, low dispute rate 15%, low N/A rate 15%.

Template-based generation can produce hundreds of consistent-quality markets from parameterized templates covering common question patterns.
LLM-assisted question writing can draft questions, detect ambiguities, enumerate edge cases, and score questions against the SMART framework -- but always requires human review.
Auto-resolution is feasible for markets tied to structured data sources (APIs for BLS, BEA, election results) and should still include a dispute window.

Concept	Formula
Brier score	$BS = \frac{1}{n}\sum_{i=1}^n (f_i - o_i)^2$
Herfindahl Index	$HHI = \sum_{i=1}^n s_i^2$
Bid-ask spread	$\frac{p_{\text{ask}} - p_{\text{bid}}}{p_{\text{mid}}} \times 100\%$
Exponential subsidy decay	$s(t) = s_0 \cdot e^{-\lambda t}$
Volume-based subsidy decay	$s(t) = \max(0,\; s_0 - k \cdot V_{\text{organic}}(t))$
Composite quality score	$Q = \sum_j w_j \cdot \tilde{m}_j$
Reputation score	$R = w_1 \cdot \text{Accuracy} + w_2 \cdot \text{Volume} + w_3 \cdot \text{Consistency} + w_4 \cdot \text{Tenure}$

Clarity above all. Ambiguity is the enemy of information aggregation.
Resolution is a contract. Make it explicit, objective, and robust.
Design for the trader. Every decision should serve clarity, trust, and efficiency.
Liquidity requires investment. Markets do not bootstrap themselves.
Measure and iterate. Use data-driven quality metrics to continuously improve design.
Automate carefully. Balance scale with quality control and human oversight.