What is UPFI?

UPFI provides a near real-time measure of anticipated Republican versus Democratic political control across major US electoral offices. It aggregates prediction market data to create a forward-looking gauge of political sentiment and expected power distribution.

How do I read the UPFI value?

  • 100 = Perfect political balance (50/50 Republican/Democratic probability)
  • Above 100 = Republican advantage (e.g., 105 = 55% Republican probability)
  • Below 100 = Democratic advantage (e.g., 95 = 45% Republican probability)
  • Formula: UPFI = 100 + (Republican Probability - 50)

The choice of which party appears "above" or "below" 100 is arbitrary and reflects no editorial bias. The index could equally be constructed as 100 + (Democratic Probability - 50).

What does UPFI NOT measure?

UPFI is not a polling index, approval rating, or forecast of specific election results. It measures market expectations of future political control. Consumers of the index can in turn use expectations of future political control to infer broader political sentiment across the US.

Index Composition

UPFI covers five office categories, weighted to the office's relative impact

  • Presidential elections (40% weight), Executive orders, Supreme Court appointments, foreign policy control
  • Senate races (30% weight), Judicial confirmations, treaty ratification, filibuster power
  • House contests (20% weight), Budget authority, impeachment power, frequent elections
  • Gubernatorial elections (8% weight), State policy implementation, redistricting influence
  • Mayoral races (2% weight), Local governance with limited federal impact

Where does the data come from?

UPFI exclusively uses Kalshi prediction market data, ensuring regulatory compliance through CFTC oversight and maintaining data consistency across all markets.

Methodology

Each market receives a weight based on four factors (scored 1-10):

Volume Sensitivity (7/10): Higher trading volume = greater market confidence

  • Rationale: Higher trading volume indicates greater market confidence and information aggregation
  • Calculation: Normalized volume relative to maximum across all markets in the index
  • Formula: volumeScore = marketVolume / maxVolumeAcrossAllMarkets
  • Weight Application: volumeScore × (volumeSensitivity / 10)

Time to Expiration (6/10): Closer elections = more reliable probabilities

  • Rationale: Markets closer to resolution provide more reliable probability estimates
  • Calculation: Inverse relationship with days to expiration, normalized to 4-year maximum
  • Formula: timeScore = max(0, 1 - daysToExpiration / 1460)
  • Weight Application: timeScore × (timeSensitivity / 10)

Population Sensitivity (5/10): Larger jurisdictions = greater significance (House/Gubernatorial/Mayoral only)

  • Rationale: Larger population jurisdictions have greater political and economic significance
  • Calculation: Normalized population relative to maximum across all markets in the index
  • Formula: populationScore = marketPopulation / maxPopulationAcrossAllMarkets
  • Application: Used for House, Gubernatorial, and Mayoral categories only
  • Exclusion: Presidential and Senate categories excluded due to national/statewide scope
  • Weight Application: populationScore × (populationSensitivity / 10)

Impact Sensitivity (8/10): Inherent political importance of the office type

  • Rationale: Different office types have varying inherent political significance
  • Calculation: Fixed impact scores by category (Presidential: 10, Senate: 8, House: 6, Gubernatorial: 7, Mayoral: 4)
  • Formula: impactScore = categoryImpactScore / 10
  • Weight Application: impactScore × (impactSensitivity / 10)

Individual market weights combine all applicable factors:

Standard Formula (House, Gubernatorial, Mayoral): marketWeight = (volumeScore × volumeSensitivity + timeScore × timeSensitivity + populationScore × populationSensitivity + impactScore × impactSensitivity) / 4

Modified Formula (Presidential, Senate): marketWeight = (volumeScore × volumeSensitivity + timeScore × timeSensitivity + impactScore × impactSensitivity) / 3

Normalization Process:

  • Calculate weighted probability: marketProbability × marketWeight
  • Sum across all markets in category: Σ(weightedProbabilities)
  • Sum total weights: Σ(marketWeights)
  • Category index: Σ(weightedProbabilities) / Σ(marketWeights) × 100

Volume weighting is used instead of excluding low-volume markets due to

  • Many politically significant local contests naturally have limited liquidity
  • Emerging markets need time to develop trading activity
  • Volume thresholds could exclude important niche electoral contests
  • Low-volume markets contribute but receive proportionally reduced weight

UPFI updates continuously 24/7 with a maximum 5-minute delay from market price changes. This captures rapid political momentum shifts that traditional polling might miss for days or weeks.

Calculation Details

Step 1: Calculate Category Probabilities Within each category, individual markets are weighted based on volume, time to expiration, population (where applicable), and impact factors. For example, if the Senate category contains:

  • Texas Senate Race: 79% Republican probability, weight 0.6
  • Maine Senate Race: 44% Republican probability, weight 0.4

Senate Category = (79% × 0.6 + 44% × 0.4) ÷ (0.6 + 0.4) = 64.8%

Step 2: Combine Categories Using Fixed Weights Apply the fixed category weights:

  • Presidential: 56.8% × 40% = 22.72%
  • Senate: 48.5% × 30% = 14.55%
  • House: 53.2% × 20% = 10.64%
  • Gubernatorial: 51.8% × 8% = 4.14%
  • Mayoral: 49.1% × 2% = 0.98%

Overall Republican Probability = 22.72% + 14.55% + 10.64% + 4.14% + 0.98% = 53.03%

Step 3: Apply Core Formula UPFI = 100 + (53.03 - 50) = 103.03

This indicates a Republican advantage across all office categories combined.

The index uses a 1 hour simple moving average (SMA) to smooth out large spikes across markets. Weights are calculated daily.

Index Operations

Resolved markets are removed to maintain UPFI's forward-looking focus. The index automatically reweights remaining markets within each category.

How does UPFI handle market disruptions?

  • Kalshi outages: Index freezes until platform resumes (all underlying markets are on Kalshi)
  • Outlier detection: Extreme probabilities (>95% or <5%) receive additional scrutiny
  • Volume adjustments: Natural weighting system reduces impact of suspicious low-volume movements

What about special elections or new markets?

  • Standard inclusion: 48-hour evaluation period for new markets
  • Special circumstances: Accelerated inclusion for major markets with ≥30 days until resolution
  • Weekly screening: Regular process for newly listed markets

Can the methodology change?

UPFI is currently in beta and the methodology may be refined as the index matures and market conditions evolve. As the index becomes more established, changes will require:

  • Formal written proposal with rationale and impact assessment
  • Technical review for feasibility
  • Full Index Committee approval
  • 30-day public notice period
  • Updated methodology documentation

Limitations and Risks

  • Participant bias: Prediction market users may not represent broader population demographics
  • Liquidity issues: Some politically significant contests have thin trading volumes
  • Coverage gaps: Limited to markets that Kalshi has listed
  • Resolution timing: Markets excluded immediately upon resolution, potentially creating temporary category underweighting

How reliable are prediction markets?

Prediction markets generally outperform traditional polling due to financial incentives for accuracy, but they can be affected by:

  • Information asymmetries among participants
  • Behavioral biases and partisan preferences
  • Potential manipulation through coordinated trading
  • Greater uncertainty for longer-term contests

For additional technical details, see the complete UPFI Methodology.