💰 quantitative finance

What Happens When Institutional Liquidity Enters Prediction Markets: Identification, Measurement, and a Synthetic Proof of Concept

This paper proposes a research design to analyze the impact of institutional liquidity on prediction markets by defining a market-quality framework, addressing identification challenges in live data, and utilizing a synthetic microstructure laboratory to demonstrate that while institutional participation may improve overall liquidity, it does not guarantee equitable benefits for all traders, particularly slower ones during market shocks.

Original authors: Shaw Dalen

Published 2026-04-14

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Shaw Dalen

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a prediction market (like betting on who will win an election or what the weather will be) as a busy farmer's market.

For years, this market was run by regular folks: neighbors trading apples, farmers guessing the harvest, and locals sharing gossip. It was a bit messy, the prices fluctuated wildly, and if you wanted to buy a bushel of corn, you might have to haggle for a long time. But everyone had a fair shot at the information.

Now, Institutional Liquidity has arrived. Think of this as a massive, high-tech supermarket chain opening a stall right next to the farmers. They have robots, super-fast computers, and deep pockets. They promise to make the market "better" by offering tighter prices and more stock.

This paper asks a crucial question: When the supermarket opens, does everyone get a better deal, or do the regular farmers get squeezed out?

Here is the breakdown of the paper's findings using simple analogies:

1. The "Supermarket" Effect: It Looks Better, But Is It?

When the big institutions (the supermarket) arrive, the displayed prices look amazing. The gap between the buying price and selling price (the "spread") shrinks. It looks like the market is super efficient and cheap to trade.

The Paper's Insight: Just because the price tag looks cheaper doesn't mean you get that cheap price. If you are slow to react (like a farmer walking to the stall), the robots might have already bought the best apples the second the news broke. You end up paying more, even though the "official" price looks low.

2. The Three Ways the Supermarket Enters

The paper argues we shouldn't just say "Institutions are here." We need to look at how they get there, because they work differently:

Market Makers (The Official Stalls): These are people hired specifically to keep the stall stocked. They make the lines shorter and prices stable.
Liquidity Incentives (The Coupons): The market pays these big players to keep their shelves full.
Automation (The Robots): These are programs that react instantly to news.

The Finding: The robots (Automation) are great at keeping the shelves full, but they are also the ones who grab the best deals the millisecond news breaks, leaving slower traders with the scraps.

3. The "Shock" Test: When the Weather Changes

The paper uses a "Synthetic Laboratory" (a video game simulation) to test what happens when a surprise happens—like a sudden storm or a surprise election result.

In Calm Times: Everyone is happy. The robots keep prices tight, and the market feels smooth.
In Shock Times: This is where the magic trick fails. When a big news event hits, the robots react in milliseconds. The "slow" traders (regular people) arrive a second later. By then, the price has already moved against them.
The Result: The market looks more liquid on paper, but it becomes more dangerous for the slow trader. The "welfare" (how much money people actually keep) gets redistributed from the slow folks to the fast robots.

4. The "Pass-Through" Problem

The paper introduces a concept called Pass-Through.

Imagine the supermarket lowers its wholesale price by 50 cents.
Good Pass-Through: The supermarket lowers the price you pay by 50 cents.
Bad Pass-Through: The supermarket lowers the price, but only for the VIPs with the fast robots. You, the regular shopper, still pay the old price (or worse).

The paper finds that while the average price gets better, the slowest traders often see zero benefit or even get hurt when news breaks.

5. What This Means for the Future

The authors are building a "blueprint" for how to study this in the real world. They are saying:

Don't just look at whether the market predicts things correctly (the "forecast").
Look at who is winning and who is losing.
The biggest gains from these high-tech markets aren't necessarily better predictions; they are better execution for the fast players, often at the expense of the slow ones.

The Bottom Line

Think of the prediction market as a race.

Before: Everyone started at the same line, running at their own pace.
After: The "Institutions" have been given jetpacks. The finish line looks closer, and the track looks smoother. But if you don't have a jetpack, you might find that the finish line has moved further away for you the moment the race starts.

The paper's main lesson: We need to make sure that when we invite the "jetpacks" (institutions) into the market, we don't accidentally leave the regular runners behind. We need to measure not just how fast the market is, but who actually gets to run in it.

1. Problem Statement

Prediction markets are transitioning from "crowd polls" to sophisticated electronic markets driven by institutional liquidity (market makers, exchange incentives, and automation). While existing literature confirms that prediction markets aggregate information well, a critical gap remains: Who benefits when institutional liquidity enters?

The central hypothesis is that while institutional participation may improve average market quality (e.g., tighter quoted spreads), it may simultaneously degrade execution quality for slower or less-informed traders, particularly during "shock states" (e.g., macroeconomic data releases). The paper addresses the difficulty of identifying these heterogeneous effects using public data, where distinguishing between different channels of institutional entry and measuring trader-level welfare is complex.

2. Methodology

The paper proposes a rigorous research design that combines a live empirical agenda with a synthetic proof-of-concept laboratory.

A. Conceptual Framework: Decomposition of Liquidity Entry

Instead of treating "institutionalization" as a single binary variable, the authors decompose it into four distinct channels:

Designated Market-Maker (MM) Coverage: Formal quoting obligations or program enrollment.
Liquidity-Incentive Eligibility: Exchange subsidies for displayed orders.
Automation/API Intensity: The capacity for participants to react via infrastructure.
Participant-Composition Shifts: The resulting change in the mix of professional vs. retail flow (treated as a mediator, not a primitive treatment).

B. Empirical Identification Strategy (Live Data)

The paper outlines an 8-step agenda for analyzing real-world data (e.g., from Kalshi or Polymarket):

Staggered Adoption: Utilizing Difference-in-Differences (DiD) with interaction-weighted event studies to handle heterogeneous treatment timing.
Mediator Analysis: Measuring participant composition shifts (e.g., cancel-to-trade ratios, cross-market synchronization) as mediators rather than direct treatments.
Shock-State Classification: Distinguishing between scheduled news, semi-scheduled events, and breaking news to isolate adverse selection mechanics.
Welfare Measurement: Focusing on pass-through (the conversion of quoted spread improvements into effective spread improvements for specific trader groups) rather than just aggregate surplus.

C. Synthetic Proof-of-Concept

To validate the measurement pipeline before accessing private venue data, the authors built a simulator with:

Latent Belief Process: Modeled as a mean-reverting process with news shocks ( $J_{m,t}$ ).
Modular Channels: Separately activating MM, incentives, and automation to observe distinct causal effects.
Outcome Generation: Simulating order books, trade prints, and resolution outcomes to generate metrics like Brier scores, effective spreads, and no-arbitrage gaps.

3. Key Contributions

Joint Market-Quality Lens: Proposes evaluating prediction markets across five dimensions simultaneously: liquidity, information efficiency, forecast quality, internal consistency, and trader-level welfare.
Channel-Based Treatment Map: Replaces the "institutionalization dummy" with a granular map of entry channels (MM, Incentives, Automation) and their specific impacts.
Welfare Incidence Framework: Shifts the focus from "does the market work?" to "who gets the benefit?" by defining Pass-Through ( $PT_g$ ) and Shock Wedges ( $SW_g$ ) as primary estimands.
Measurement Blueprint: Provides a concrete table (Table 1) mapping raw data fields (quotes, trade signs, timestamps) to operational definitions and identifying specific risks (e.g., stale books, proxy contamination).

4. Key Results (Synthetic Laboratory Findings)

The synthetic results serve as a "sanity check" for the measurement pipeline, revealing several non-intuitive dynamics:

Divergent Effects on Execution vs. Forecasting:
- Execution Quality: Institutional channels significantly improve liquidity. High-bundle regimes saw quoted spreads drop ~14% and effective spreads drop ~19%. Market-maker coverage had the strongest impact on spread tightening.
- Forecast Quality: Gains were mixed. While internal consistency (no-arbitrage gaps) improved, the Brier score (forecast accuracy) actually worsened slightly in high-bundle regimes, suggesting execution gains do not automatically translate to better calibration.
The "Shock State" Divergence:
- In calm states, institutional liquidity benefits all participants.
- In shock states, the benefits are uneven. Automation and institutional presence intensify adverse selection for slower traders.
- Pass-Through Failure: Slow takers receive minimal pass-through of spread improvements during shocks. In the synthetic model, slow takers in high-bundle regimes actually faced worse execution costs during shocks compared to low-bundle regimes, while fast/informed traders continued to benefit.
Heterogeneity: The magnitude of liquidity gains varies significantly by information density, hedgeability, and resolution clarity, not just by broad event categories (e.g., politics vs. sports).

5. Significance and Implications

Regulatory & Design Insight: The paper argues that regulators and exchange designers cannot assume that tighter spreads equate to a fairer market. Institutional liquidity may create a "two-tier" system where fast, automated players extract rents from slower participants during information shocks.
Methodological Rigor: It establishes a standard for how to study prediction markets using public data, emphasizing the need for staggered-adoption identification and explicit mediator validation to avoid confounding venue mechanics with genuine liquidity effects.
Welfare Focus: The primary contribution is the shift in the research question from "Are prices accurate?" to "Is the market inclusive?" The findings suggest that while institutionalization improves market depth, it may redistribute welfare away from the slowest participants, necessitating specific protections or design adjustments for retail traders.

In conclusion, the paper provides a comprehensive blueprint for analyzing the microstructure of modern prediction markets, demonstrating that institutional liquidity is a double-edged sword: it enhances average market quality but risks exacerbating inequality in execution costs during critical information events.