Risk Factors

Belief Index levels are informational measurements with real, documented limitations. This page covers them: what the published data can and cannot tell you, and where the methodology’s simplifying assumptions bind.

​

Index & Data Limitations

​

Underlying Market Volatility

The constituents are prediction market contracts – binary instruments that settle at $1 or $0. Levels computed from them can move sharply:

• Event-driven gaps: A single news event (a court ruling, an economic data release, a political development) can move a constituent from 80% to 20% within minutes
• Binary resolution: At settlement, every constituent resolves to exactly $1 or $0 – there is no middle ground and no partial outcome
• Sentiment-driven pricing: Prices reflect crowd expectations, which can shift rapidly based on information, rumor, or market dynamics

​

Jump Risk and Correlation

Even in a diversified basket, multiple correlated events can resolve unfavorably simultaneously. For example:

• A series focused on interest rate expectations could see several markets move sharply in the same direction following a single Federal Reserve announcement
• A political series could see correlated movement across multiple races if a national political event shifts overall sentiment
• An economic data series could see cascading effects if a single data release (e.g., GDP) changes expectations for related indicators

Diversification reduces but does not eliminate jump risk. The methodology treats each market as independent and does not discount for information overlap. This means the smoothing benefit may be overstated for correlated markets.

Worst-case scenario

If every market in a series resolves against the tracked outcome, the series level goes to zero. With an equal-weighted 7-market basket, this requires all 7 events to resolve unfavorably. While unlikely, it is not impossible – especially for series tracking correlated events.Even partial adverse resolution is significant: if 5 of 7 markets resolve unfavorably, the level could fall approximately 70% from those resolutions alone.

​

Thin Markets and Midprice Reliability

Some prediction markets have limited trading activity, which affects the reliability of the midprice signal the index is computed from: Markets within a single series can have vastly different liquidity profiles. A market with $30 million in depth may sit alongside one with $400 in depth – and a midprice computed from a thin book is a weaker probability estimate.

​

Theoretical vs. Executable Values

Published levels are based on midpoint pricing – theoretical values that do not reflect the cost of actually executing trades. They do not account for:

• Trading fees on the underlying markets (typically 0.5-2%)
• Slippage from market-sized orders
• Market impact of replication trades

Nothing tracks these indices with real assets. Anyone attempting to replicate a series would realize values below the theoretical level, particularly at sizes that are large relative to market depth. Treat index levels as measurements, not attainable prices.

​

Static Weighting

Weights are fixed at composition time and do not dynamically adjust for:

• Changes in relative market liquidity
• Shifts in the importance or relevance of individual events
• Emerging correlations between markets

This means a market that becomes illiquid after inclusion continues to receive its full weight in the index, potentially reducing the reliability of the published level.

​

Single Data Source

Index computation depends entirely on order book data from a single prediction market venue. There is no secondary source for price verification. Venue outages, data errors, or API changes could affect the accuracy or availability of published levels. Venue disruptions could also suspend publication entirely, and the venue’s continued operation is not guaranteed.

​

Oracle and Resolution Risk

Constituent outcomes are determined by resolution oracles. Risks include:

• Disputed resolutions: Ambiguous event outcomes may lead to contested settlements
• Oracle failures: Technical failures in the resolution mechanism could produce incorrect settlements that flow into published levels
• Resolution delays: Events may take longer to resolve than expected, extending a series’ timeline

​

Concentration

Each series contains a relatively small number of markets (typically 5-20). This means:

• A single resolution can move the series level significantly – roughly 7-20% in an equal-weighted basket, and under the published weighting scheme a constituent may carry up to the 50% concentration cap, so a single adverse resolution can remove up to half the level
• The level of diversification is substantially less than a broad equity index with hundreds of constituents
• The series is more sensitive to individual event outcomes than a highly diversified aggregate

​

Reconstitution (Perpetual Series)

Perpetual series change composition through formal reconstitution events, which introduces methodology-specific caveats:

• Composition drift: The basket measured today is not the basket that will be measured indefinitely. Reconstitution is rules-based and forward-looking, but the theme’s constituents evolve.
• Replication cost is not reflected: The level is chain-linked across a composition change with no deduction for what replicating the change would cost. Anyone tracking a series would pay spreads and slippage at each reconstitution that the published level does not show.
• Publication gaps: During a reconstitution gate, computation is suspended and no new level is published – typically for less than 48 hours. The last pre-gate value remains authoritative during the gap.
• Aborted events: If a reconstitution aborts mid-flight, the prior composition is restored and publication resumes against it.

​

Publication and Technology Risk

The computation and publication pipeline depends on software and infrastructure that may experience bugs, outages, data corruption, or integration failures. If current prices cannot be fetched, the system falls back to stale data and flags the computation as stale rather than hiding it – but a stale level may not reflect current market conditions. Data-integrity halts may pause publication entirely while a discrepancy is investigated.

​

Regulatory Environment

Prediction markets operate in an evolving and uncertain regulatory environment. Regulatory change could restrict the underlying markets, affect the availability or legality of the data sources the indices depend on, or require changes to how the indices are produced and published.