Literature Readings · Prediction Markets · Paper A Proposal

Political Risk Prices and Cross-Asset Returns

Evidence from $5B in 2024 Prediction Markets

Working draft research proposal · May 19, 2026 · "Snowberg-Wolfers-Zitzewitz QJE 2007 redux at 100× scale"

📋 Executive Summary

The 2024 US presidential election generated $5B+ in PolyMarket and Kalshi trading volume — 100× the size of TradeSports during the 2004 election studied by Snowberg, Wolfers & Zitzewitz in their canonical QJE 2007 paper. They identified causal partisan effects on equity valuations of 2-3 percentage points. We propose a redux at the 2024 scale that exploits the much richer cross-section of industries, currencies, rates, commodities, volatility surfaces, and crypto assets, with cleaner identification from PolyMarket's 1-minute on-chain resolution, IV from whale-induced exogenous odds shifts (documented by Tsang-Yang Mar 2026), and triangulation via the PolyMarket-Kalshi-IEM divergence.

The crucial fact: none of Wolfers, Zitzewitz, Snowberg, Manski, Ottaviani, or Rothschild has a 2024-2026 academic working paper on PolyMarket / Kalshi. The canonical methodology has not been applied to the canonical new data by anyone with the canonical pedigree. This is a clean arbitrage opportunity.

Target: Journal of Finance / QJE / JFE. Timeline: 9 months to first submission. Data cost: low (PolyMarket fully on-chain via the Cong et al. Apr 2026 dataset release; Kalshi via academic API request; asset prices via WRDS subscriptions). Side product: the wallet-attribution + PM-shock-impact methodology becomes the toolkit for CFTC / SEC / DOJ litigation consulting in the Griffin-Shams (Tether 2020) style.

1. Research Question

When PolyMarket and Kalshi prices move on news during the 2024 US election cycle, do US asset prices respond causally, and through what channels?

Three sub-questions:

Q1 (Asset Pricing). What is the asset-class-level response to a 1-percentage-point increase in PolyMarket's Trump probability? Across equities (and by industry), FX, Treasury yields and term spreads, credit spreads, commodities, volatility surfaces, and crypto.

Q2 (Mechanism). Through which Trump-policy channels does the cross-sectional industry response operate — tax rate, regulatory burden, trade-tariff exposure, immigration/H-1B exposure, or geopolitical tilt?

Q3 (Wedge). Does the asset-pricing response differ when PolyMarket moves but Kalshi does not (idiosyncratic / whale-driven move) vs when both move together (informational shock)? Tests of price-vs-noise in the SWZ tradition.

The single headline statistic we are after:

The cross-asset analog of SWZ 2007's "2-3 percentage points of equity valuation" effect — but disaggregated by industry, FX, rates, and crypto. We will produce a "political risk price vector" for the US asset universe in 2024.

2. Positioning vs Existing Literature

The paper sits at the intersection of three literatures, and the contribution is sharpest when stated as "the gap each existing paper leaves."

Paper	What they did	What we add
Snowberg-Wolfers-Zitzewitz QJE 2007	TradeSports + IEM intraday moves on 2004 election day → equity-aggregate response (Republicans worth ~2-3% more for stocks). $30M scale.	100× scale, full cross-section, full election cycle, cleaner ID. Their template applied to a richer setting.
Flynn-Tarkom FRL 2025	Daily VAR of DJT (Trump Media) stock vs PolyMarket odds. Single asset.	Cross-section (industries, FX, rates, crypto). High-frequency identification rather than daily VAR.
Bartlett-O'Hara SSRN 6615739 (Apr 2026)	Adverse selection within Kalshi. Kyle-λ + Glosten-Harris of Kalshi order flow.	Different unit of analysis: outside the prediction market, into asset prices.
Tsang-Yang arXiv 2603.03136 (Mar 2026)	PolyMarket on-chain "anatomy" of 2024 election. Documents Fredi9999 wallets but concludes "not manipulation, heterogeneous beliefs."	Uses Tsang-Yang's documented exogenous whale-induced odds shifts as our IV instrument.
Diercks-Katz-Wright NBER 34702 (Jan 2026)	Kalshi macro forecasting accuracy. Kalshi beats Bloomberg consensus + Fed Funds Futures.	Election (not macro) markets; cross-asset response (not nowcasting accuracy).
Chernov-Elenev-Song NBER 33339 (Jan 2025)	Time-series factor model of PolyMarket vs polls.	Cross-sectional asset-pricing response, not voter-preference structure.
Mohanty-Krishnamachari arXiv 2604.01431 (Apr 2026)	Kalshi macro contracts → BTC/altcoin volatility.	Election markets → cross-asset (not macro markets → crypto alone).
Krause SSRN 6520278 (Apr 2026)	PolyMarket CLARITY-Act passage probabilities → BTC. Single-event, single-asset.	Multi-event, multi-asset. Cleaner industry-exposure measure.
Bauer-Swanson AER 2023; Nakamura-Steinsson QJE 2018	High-frequency identification of monetary policy shocks via Fed announcement windows.	Methodological template — same identification logic, election-market shocks instead of FOMC.

One-paragraph positioning: No paper combines (i) the cross-asset breadth of equity industries + FX + rates + commodities + crypto, (ii) the high-frequency identification of intraday windows, (iii) the IV identification using documented whale-induced exogenous shifts, and (iv) the canonical SWZ 2007 framing. Single-asset versions exist (Flynn-Tarkom DJT, Krause CLARITY-BTC); within-Kalshi microstructure exists (Bartlett-O'Hara); macro applications exist (Diercks-Katz-Wright, Mohanty-Krishnamachari). The cross-asset political-risk paper does not.

3. Data Architecture

3.1 PolyMarket on-chain (free)

PolyMarket settles on Polygon. Every order, trade, settlement, oracle event is public. The Cong et al. Apr 2026 dataset release (arXiv 2604.20421) provides the unified research infrastructure:

770K+	PolyMarket markets (Oct 2020 - Mar 2026)
943M	trading fills (timestamp, wallet, market, position, price)
~2M	oracle resolution events
Polygon RPC	direct chain access (backup if Cong dataset has gaps)
The Graph	indexed subgraph for fast queries
Goldsky / Dune	analytic SQL access to the same chain data

3.2 Kalshi public + academic API

Need to apply early — Kalshi has academic API access with ~3-week response time. Public data available without API:

Market-level OHLCV	per contract per day (full election cycle)
Volume reports	quarterly aggregate from Kalshi public reports
Trade-level data	requires academic API approval

3.3 IEM (free, low-frequency)

Iowa Electronic Markets data via University of Iowa direct request. Capped at $500/trader → useful counterfactual. Berg-Nelson-Rietz (Cambridge PS 2024) is the canonical reference for the 2024 cycle.

3.4 Asset prices (WRDS + Bloomberg)

Asset class	Source	Frequency	Sample
US equities	WRDS TAQ	1-second	S&P 500 + Russell 3000
Sector ETFs	WRDS / Refinitiv	1-minute	11 SPDR sectors (XLF, XLE, XLV, ITA, XLI, ...)
FX	Bloomberg / Refinitiv	1-second	MXN, CNY, EUR, RUB, JPY, CAD, INR, BRL, ZAR
Treasury	GovPX / NY Fed	1-second	2Y, 5Y, 10Y, 30Y on-the-run
Credit	WRDS / Refinitiv	1-minute	IG (LQD), HY (HYG) spread proxies
Vol	CBOE	1-minute	VIX, VVIX, term structure (1M/3M/6M)
Commodities	Bloomberg	1-minute	WTI, gold, copper, soybeans, corn
Crypto	CoinGecko / CryptoCompare	1-minute	BTC, ETH, SOL, ADA, LINK + DJT stock
News timing	Bloomberg Event DB + manual	second-resolution	9 major + ~30 minor 2024 election events

3.5 Pre-determined industry exposures

For cross-sectional analysis, we need pre-2024 measures of each industry's Trump-policy exposure to avoid look-ahead bias. Five exposure measures:

Channel	Measure	Source
Tax (corporate)	2017-2019 effective tax rate × statutory differential	Compustat
Regulation	Industry-level mentions in Federal Register; lobbying spend (OpenSecrets)	FedReg + OpenSecrets
Trade / tariffs	Import share from China + Mexico; tariff-exposure score	Atkin-Khandelwal + Census trade
Immigration	H-1B dependency + unauthorized-worker industry share	USCIS + BLS Borjas data
Geopolitical	Defense contract revenue share; Russia/Ukraine/Iran/China exposure	Govt contracts + 10-K Item 1A

4. Identification Strategy

The threat to clean identification is omitted-variable bias: the underlying news shock drives both PolyMarket prices and asset returns. We address this with three complementary identification strategies, triangulated for robustness.

4.1 High-frequency event-window identification (SWZ template)

Around each scheduled / unscheduled election news event at minute t_e, we compute:

first-stage shock ΔP_PM,e = P^PolyMarket_{t_e+30min} − P^PolyMarket_{t_e−5min}

where P^PolyMarket_t is the PolyMarket Trump-win probability at minute t (taken as the mid-quote / VWAP of the Trump "Yes" contract). The event-window response is:

main regression ΔY_i,e = α_i + β_i · ΔP_PM,e + γ_i · X_e + ε_i,e

where ΔY_i,e is the log return of asset i in the same window, X_e includes pre-event controls (lagged VIX, lagged S&P), and standard errors are clustered by event.

9 major events used as identification points:

Date	Event	Expected PM Trump Δ
Jan 15, 2024	Iowa caucus — Trump wins	↑
May 30, 2024	Trump conviction (NY hush money)	↓ (modest)
Jun 27, 2024	Biden-Trump debate (Biden collapse)	↑↑↑
Jul 13, 2024	Trump assassination attempt #1 (Butler PA)	↑↑↑
Jul 21, 2024	Biden drops out	↓↓ (uncertainty)
Aug 6, 2024	Harris picks Walz as VP	↓
Sep 10, 2024	Harris-Trump debate	↓ (Harris perceived win)
Sep 15, 2024	Trump assassination attempt #2	↑ (modest)
Nov 5, 2024	Election night	↑↑↑ (resolution)

Plus ~30 minor events (smaller polls, court rulings, VP debate, regional events). Total N ≈ 40 event-asset × asset-cross-section ≈ 2,000 observations per asset class.

Why HF identification works:

In a 30-minute window, the only macro variable that should plausibly move is the election-related shock. We absorb other macro variation via the controls X_e. The identifying assumption is:

identifying assumption E[ε_i,e · ΔP_PM,e | X_e] = 0

— i.e., conditional on controls, PolyMarket moves are orthogonal to other return drivers in the same 30-minute window. This is the standard high-frequency identification logic of Bauer & Swanson (AER 2023) and Nakamura & Steinsson (QJE 2018) for monetary policy, applied to election-market shocks.

4.2 Whale-induced exogenous shifts as IV (the Tsang-Yang instrument)

For non-event periods (continuous flow of routine PolyMarket trading), we need a different identification strategy. Tsang-Yang (arXiv 2603.03136) documented specific episodes where the Fredi9999 / Théo wallets moved PolyMarket prices on their own — i.e., independent of contemporaneous news.

We use these whale-induced PolyMarket moves as an instrument for the price level:

first stage ΔP_PM,t = π · WhaleTrade_t + δ · X_t + ν_t

second stage ΔY_i,t = α_i + β_i^IV · ΔP̂_PM,t + γ_i · X_t + ε_i,t

where WhaleTrade_t is the net flow from identified whale wallets in minute t. Following the wallet-clustering methodology of Sirolly-Ma-Kanoria-Sethi (Nov 2025) and Chainalysis (Nov 2024), we use:

The 11 Fredi9999/Théo wallets identified by Chainalysis
The 43,000-wallet wash-trading cluster from Sirolly-Sethi (separately tested)
Top-1% by volume per Akey-Grégoire-Harvie-Martineau (SSRN 6443103)

Identifying assumption for the IV:

exclusion restriction E[ε_i,t · WhaleTrade_t | X_t] = 0

— whale trades are driven by the whale's private liquidity needs and idiosyncratic beliefs, not new information about candidate quality that simultaneously moves both PM prices and asset prices. Validation: WhaleTrade should not predict polling changes (placebo) or campaign-event timing (placebo) at higher than chance.

This is the standard trade-flow-as-IV logic of Hasbrouck (1991), Brunnermeier-Pedersen (RFS 2009), and the order-flow literature. The novelty: the IV is built from fully observable on-chain data, not estimated from price-impact models.

4.3 The PolyMarket-Kalshi-IEM divergence triangulation

Three platforms price the same election in 2024. Their cross-platform divergence is itself a clean source of identification.

divergence shock Δdiv_t = ΔP^PolyMarket_t − ΔP^Kalshi_t

Interpretation. If both platforms move together (Δdiv_t ≈ 0 but levels change) → common information shock. If only one moves (Δdiv_t ≠ 0) → idiosyncratic / liquidity / wallet-specific shock to that platform.

We can decompose ΔP_PM,t into:

decomposition ΔP_PM,t = ΔP^common_t + ΔP^{idiosyncratic}_t

where ΔP^common_t ≈ (ΔP^PolyMarket_t + ΔP^Kalshi_t + ΔP^IEM_t) / 3 and ΔP^{idiosyncratic}_t is the orthogonal residual.

Two tests follow:

Test A. Asset prices should respond more strongly to ΔP^common (real information) than to ΔP^{idiosyncratic} (noise / liquidity). If they respond equally, we have a problem.
Test B. The Kalshi response coefficient should equal the PolyMarket response coefficient up to a noise-amplitude scaling, after controlling for the divergence. This is a Hausman-style overidentification test.

Triangulation. If all three identification strategies (HF event window, Whale-IV, Cross-platform divergence) yield similar β̂_i estimates, we have unusually strong confidence. Each makes different identification assumptions, so converging estimates suggest the true causal effect is what we measure.

5. Main Specifications

5.1 Asset-class-level response (Spec 1)

spec 1 ΔY_k,e = α_k + β_k · ΔP_PM,e + γ_k · X_e + ε_k,e

where k ∈ {SPX, USD-index, 10Y-yield, VIX, WTI, gold, BTC, ...}. Replicates and extends SWZ 2007 Table 1 at the asset-class level. Hypothesis: β_SPX > 0 (Trump favorable for equities); β_10Y-yield > 0 (Trump → higher deficits → higher yields); β_USD-index ambiguous; β_BTC > 0 (deregulation).

5.2 Industry-level cross-section (Spec 2)

For each of N ≈ 500 firms in S&P 500, estimate firm-specific β̂_i from Spec 1. Then run cross-sectional regression:

spec 2 β̂_i = δ₀ + δ₁ · TrumpExposure_i + δ₂ · Z_i + u_i

where TrumpExposure_i is the composite pre-2024 measure from §3.5, and Z_i includes firm size, book-to-market, and other Fama-French controls. Hypothesis: δ₁ > 0. A 1-SD increase in pre-determined Trump exposure should generate a measurably larger asset-price response to PolyMarket moves.

5.3 FX cross-section (Spec 3)

Same as Spec 2 but with FX rates. Key prediction: MXN (Mexican peso) should have the largest negative β due to Trump tariff threats; CNY similar; EUR ambiguous; INR/BRL mildly negative.

5.4 Rates / term premium (Spec 4)

Decompose the 10Y yield response into:

Expected short-rate path (using Adrian-Crump-Moench term-premium decomposition)
Term premium
Inflation expectations (from TIPS)

Then run Spec 1 on each component. Expected: Trump probability ↑ → inflation expectations ↑ AND term premium ↑.

5.5 Volatility surface response (Spec 5)

Same as Spec 1 with VIX term structure as outcomes. Hypothesis: PolyMarket moves should shift the volatility surface, especially in the short end of the curve.

6. Mechanism Tests

Once we have β̂_i per firm, we decompose the cross-sectional variation into the five policy channels.

Channel	Test	Identifying intuition
Tax	Run Spec 2 with TaxExposure_i = (Statutory − Effective tax rate)_i	Firms with low ETR get most from corporate tax cuts
Regulation	RegulatoryBurden_i = federal-register mentions × lobbying spend	Highly-regulated firms (banks, energy, healthcare) gain most from deregulation
Trade	ImportShare_i,China + ImportShare_i,Mexico	Firms with high China/Mexico supply chains hurt by tariffs
Immigration	H-1B-dependency_i + UnauthorizedShare_i	Construction, agriculture, hospitality, tech firms hurt
Geopolitical	DefenseContractShare_i + Russia/Ukraine/Iran exposure	Defense, energy beneficiaries; renewables, China-supply chain losers

A horse-race regression of β̂_i on all five channels jointly identifies which dominates. Hypothesis: tax + regulation are largest; trade pulls in the opposite direction (Trump-exposed firms have offsetting tax-gain and trade-loss).

6.1 The Mexican peso as the canonical mechanism check

The MXN response is the canonical case study. Trump tariff threats are explicit and unambiguous → cleanest causal channel. We expect:

β_MXN ≈ -3% per 10pp Trump probability increase

This is a falsification test: if β_MXN ≠ negative-and-large, our identification has issues.

7. Robustness

Window length: Repeat all specs with 5min / 15min / 30min / 60min / 1-day windows. Coefficients should be stable.
Event leave-one-out: Drop each of the 9 major events one at a time. Coefficients should not be driven by any single event.
Kalshi as alternative PolyMarket: Re-run all specs using ΔP_Kalshi instead of ΔP_PM. Kalshi is regulated, no whale risk — robustness check on the whale-IV.
Median of three: Use median(ΔP_PolyMarket, ΔP_Kalshi, ΔP_IEM) — robust to platform-specific noise.
News controls: Add Bloomberg news count and RavenPack sentiment as Xe. Coefficients should be reduced if news is the omitted variable, stable if not.
Pre-event placebo trends: Run Spec 1 on windows BEFORE event timing. Coefficients should be ≈ 0.
Different industry exposure measures: Replace composite with each channel separately and with alternative constructions (e.g., 10-K text-mining vs lobbying-based).
SE clustering: Two-way cluster by event AND asset; bootstrap; HAC.

8. Falsification Tests

A skeptical reviewer's nightmare list. We pre-commit to all of these.

Falsification 1 — Null-exposure industries should NOT respond.

Utilities (regulated by state, indifferent to federal policy) and grocery retail (low tax / regulation exposure) should have β_i ≈ 0. If they don't, our cross-section identification is biased.

Falsification 2 — PolyMarket sports markets should NOT predict S&P 500 returns.

If PolyMarket prices on NBA games systematically predict equity returns, then our PolyMarket → asset price link is spurious correlation rather than causal information. Tests of placebo channel.

Falsification 3 — Pre-election (Jan 2023) placebo.

Run Spec 1 on Jan 2023 PolyMarket data (low liquidity, low salience). Effects should be much smaller in magnitude.

Falsification 4 — Outside-US-hours placebo.

Run Spec 1 on PolyMarket moves during overnight US hours when major US events shouldn't be driving asset prices. If we still see correlation, there's an unexplained common factor.

9. Heterogeneity Tests

Pre- vs post-Sep 10 debate (when Trump's PolyMarket odds permanently exceeded Harris's): coefficients may differ in regime where market becomes increasingly confident
Whale-led vs not-whale-led moves (using Sirolly-Sethi cluster data): whale-led moves should have weaker asset-price response if markets correctly identify them as noise
"Information" vs "liquidity" moves (using Kalshi-PolyMarket divergence): information moves should have full response, liquidity moves should have much smaller response
Large vs small PM moves: kernel regression of β on |ΔP_PM| — test for nonlinearity
US trading hours vs Asia/Europe hours: cross-platform leadership test (who leads whom?)
By industry liquidity: high-liquidity stocks may respond faster, low-liquidity stocks may show delayed response

10. Risks & Mitigations

Risk	Severity	Mitigation
Reverse causality: asset prices drive PolyMarket, not vice versa	High	Triangulation across three identification strategies. Granger-causality at 1-minute resolution shows PolyMarket actually lags major macro markets by ~30s — supports the "PM as response to information that asset prices also respond to" framing but rules out crude reverse causality.
Omitted variable: news event drives both	High	HF event window absorbs unobserved news via the conditional-orthogonality assumption (standard SWZ / Bauer-Swanson). Robustness with RavenPack news controls.
Bartlett-O'Hara already covered similar ground	Medium	Theirs is microstructure-within-Kalshi (Kyle's λ, Glosten-Harris). Ours is cross-asset propagation OUTSIDE the PM. Different empirical object. We cite as benchmark.
Flynn-Tarkom did DJT specifically	Medium	Theirs is single-asset VAR with daily data. Ours is cross-section with intraday identification. Our paper subsumes theirs.
Crowded field; 3-4 papers may be in progress	Medium	Speed: aim for first draft by month 6. Novelty in (a) cross-asset breadth, (b) the industry exposure measure construction, (c) the triangulation across 3 identification strategies.
Kalshi API access (3-week response time)	Medium	Apply on day 1. In the worst case, PolyMarket on-chain alone is sufficient for the main paper.
Industry exposure measure is subjective	Medium	Pre-register the measure with OSF. Use multiple alternatives. Validate on 2016 election as out-of-sample test.
Whale wallet identification noisy	Low	Use Chainalysis's published 11-wallet list as the conservative measure. Robustness with broader (top-1% Akey et al.) and narrower (manual top-3) definitions.
WRDS / Bloomberg subscription cost	Low	Standard academic infrastructure; ~$5K/year. Worth it for a top-5 paper.

11. Timeline & Resources

Phase	Month	Deliverable
Setup	1	Apply for Kalshi academic API. Pull Cong et al. Apr 2026 PolyMarket dataset. Set up WRDS / Bloomberg pipeline. Build Polygon-RPC backup.
Data pipeline	2	Minute-level PolyMarket Trump probability time series. Match with intraday asset prices. Verify against published Tsang-Yang figures.
Event identification	3	Construct 9 major + 30 minor event list with exact timestamps. Identify whale-trade events from Tsang-Yang and Chainalysis.
Industry exposure measure	3-4	Build the composite TrumpExposure_i measure with 5 sub-components. Pre-register at OSF.
Main specifications	4-5	Spec 1 (asset-class) + Spec 2 (industries) + Spec 3 (FX) + Spec 4 (rates) + Spec 5 (vol). All three identification strategies.
Mechanism + robustness	5-6	Channel decomposition. Full robustness battery. Falsification tests.
First draft	6-7	Working paper ready. Internal seminar.
Polish + submit	8-9	External seminar feedback. *Submit to Journal of Finance* (or QJE).**
R&R cycle	10-18	Typical 1-2 R&R rounds at top journal.

12. The Three-Paper Bundle

Paper A is the lead of a 3-paper program that shares data infrastructure, identification toolkit, and wallet-attribution methodology.

Paper A — "Political Risk Prices and Cross-Asset Returns: Evidence from $5B in 2024 Prediction Markets" (this proposal)

Submitted: Month 6-9. Target: J Finance / QJE / JFE.

Paper B — "Self-Fulfilling Prophecy in Prediction Markets: PolyMarket Pricing, Donor Flows, and Voter Turnout"

Reuses the wallet-clustering + event-identification infrastructure. New data: FEC daily donations, Vivvix TV ad spending, county-level turnout. Uses whale-induced PolyMarket shifts as IV for media-reported odds. Submitted: Month 12-15. Target: AER / QJE.

Paper C — "Does the Federal Reserve Listen to Kalshi? Textual and High-Frequency Evidence"

Reuses Kalshi data pipeline + Bauer-Swanson identification template. New data: FOMC minutes/transcripts (LDA + sentiment), Fed Funds Futures intraday. Possible coauthor: Diercks (Fed) or Wright (Hopkins/Fed). Submitted: Month 15-18. Target: AER / J Mon Econ.

Why bundle: ~80% of data infrastructure overlaps. The wallet-attribution + PM-shock methodology built for Paper A serves both B (PolyMarket-FEC channel) and C (Kalshi-Fed channel). One data investment → three top-5 shots → portfolio of papers that all reference each other.

13. Litigation Consulting Spinoffs

The methodological infrastructure built for Paper A produces several side-deliverables of direct value to government and private litigation:

Deliverable	Litigation use case
Wallet-attribution pipeline (clusters + entity resolution)	CFTC manipulation investigations; DOJ insider-trading cases (Van Dyke Maduro); SEC enforcement
PM-shock → asset-price impact methodology	Securities fraud cases involving Trump Media (DJT), prediction-market-linked ETFs, election-related corporate disclosure
Industry-exposure scoring (pre-determined Trump-policy exposure measure)	Tax / regulation / trade litigation; M&A risk assessment for politically sensitive deals
Event-timing identification (minute-level news → PM → asset prices)	Insider-trading prosecution; market-timing dispute resolution

The Griffin-Shams (Tether 2020 J Finance) playbook. Griffin and Shams published a rigorous identification paper documenting that Tether issuance manipulated Bitcoin prices. The paper became the canonical empirical foundation for dozens of subsequent CFTC, SEC, and private litigation actions. Both authors became expert witnesses across the crypto-litigation ecosystem. Combined consulting practice: estimated $1M+/year.

The PolyMarket-Kalshi space is currently more active than crypto was in 2020. With papers A + B + C as the methodological foundation, the consulting franchise that follows is plausibly larger than Griffin-Shams's was.

Concrete consulting spinoffs from Paper A alone (before B and C land):

CFTC Office of the Chief Economist: contracted methodology review for ongoing PolyMarket manipulation investigation
DOJ Public Corruption Division: methodology consultation on Van Dyke prediction-market trading case + future cases
SEC Division of Enforcement: PolyMarket-related security fraud cases
Private securities-fraud class actions involving DJT / event-contract-linked instruments
Election-integrity lawsuits (multiple plaintiffs in 2024 cycle)

Working draft research proposal · Generated May 19, 2026 · For internal planning purposes

References & bibliography in prediction-markets.html · §1 Modern Era 2024-2026 · §4 Foundations 1988-2020 · §5 Mechanism Design & Theory

← Prediction Markets ← All Readings