Documentation
Learn how Attena indices are calculated, what data sources we use, and how to interpret the values.
Overview
Attena indices are rolling benchmarks built from live prediction market prices. Each index blends multiple contracts into one continuously updating value that never expires. One price, always live, hedgeable with the underlying markets.
Weather indices are the first category available, tracking predicted daily high temperatures across multiple cities based on Kalshi prediction market prices.
Weather Indices
Weather indices track market-priced temperature predictions for major cities. Each index rolls forward automatically, comparing real-time market prices against NOAA seasonal normals. Go long on heat, short on cold.
How It Works
The Core Formulas
Blended Prediction
0.7 × Today + 0.3 × Tomorrow
Index Value
100 + (Predicted − Normal)
Each index point = 1°F deviation from seasonal normal
Fetch Market Data
We retrieve temperature prediction contracts from Kalshi's API for the target city. Each contract represents a temperature "bucket" (e.g., "High between 70-74°F") with an associated yes price reflecting the market's probability for that outcome.
Calculate Predicted Temperature
For each day, we compute a probability-weighted predicted high temperature. Here's the full process:
Example: Raw Market Prices
Step 2a: Normalize Probabilities
Prices sum to $1.00, so they're already normalized. If they don't sum to 1, we divide each by the total.
Step 2b: Calculate Weighted Average
Multiply each bucket's midpoint by its probability, then sum:
Blend Today and Tomorrow
To reduce volatility during the daily contract roll, we blend today's and tomorrow's predicted highs:
This 70/30 weighting emphasizes today's forecast while incorporating forward-looking information. It smooths the transition as today's contract settles and tomorrow's becomes the new "today."
Compare to Seasonal Normal
The blended predicted high is compared against the historical normal high temperature for the current date. These normals come from NOAA's 30-year climate data (1991-2020) for each city's local weather station. Each date has its own normal—January 15th has a different baseline than July 15th.
Calculate the Index Value
The final index value represents the deviation from the seasonal normal, centered at 100:
- • Index = 100: Predicted temperature equals seasonal normal
- • Index > 100: Warmer than usual (each point = +1°F)
- • Index < 100: Colder than usual (each point = −1°F)
Worked Example
Let's walk through a complete calculation for NYC on January 15th:
Given Data
Interpretation: The market expects NYC to be approximately 7°F warmer than the seasonal average for mid-January. An index of 106.9 signals a notably warm winter day.
Interpretation Guide
The index is centered at 100, representing the seasonal normal. Here's how to interpret different values:
| Index Value | Meaning | Example Context |
|---|---|---|
| 100 | Exactly normal | Market expects seasonal average temperature |
| 105 | 5°F above normal | Mildly warmer than typical |
| 110+ | 10°F+ above normal | Significant heat event, unseasonably warm |
| 95 | 5°F below normal | Mildly colder than typical |
| 90 or below | 10°F+ below normal | Significant cold snap, unseasonably cold |
Key insight: Each index point equals exactly 1°F deviation from normal. An index of 108 means the market expects it to be 8°F warmer than the historical average for that date.
Edge Cases
Tomorrow's market not available
If tomorrow's contract hasn't opened yet (e.g., late in the day), we use 100% of today's prediction instead of the 70/30 blend. The index remains valid but relies solely on today's market.
During maintenance windows
Kalshi markets operate 24/7 with one exception: weekly maintenance every Thursday from 3:00 AM to 5:00 AM ET. During this brief window, the index displays the last calculated value with a "Market Closed" indicator. The index resumes live updates immediately after maintenance ends.
After contract settlement
Once a day's temperature contract settles, that contract's contribution uses the actual recorded temperature rather than the probability-weighted expectation. Settlement occurs after the National Weather Service (NWS) releases the official Daily Climate Report—typically overnight but sometimes as late as 8:00–9:00 AM in the station's local timezone (Eastern for NYC/Miami/Philadelphia, Central for Chicago/Austin, Pacific for Los Angeles). This ensures the index reflects reality once outcomes are confirmed.
Low liquidity markets
In markets with low trading volume, prices may not sum to exactly $1.00 or may have wider spreads. We normalize probabilities to ensure they sum to 1 and use midpoint prices. Very illiquid markets may show stale prices until new trades occur.
Index Updates
The Weather Index is a rolling, continuously updating value:
- • Recalculates every 5 minutes during Kalshi market hours
- • All calculated values are stored for historical tracking and charting
- • This is a continuous rolling index with no settlement or expiration
- • Perfect for continuous hedging—no expiration dates, no positions to roll
- • During market closures, the index displays the last calculated value until markets reopen
Available Cities
NYC Weather Index
LiveA New York City rolling temperature index derived from live market signals. Values above 100 mean warmer than expected; below 100 means cooler.
- • Market: KXHIGHNY
- • Baseline: NOAA Central Park normals
- • Update frequency: Every 5 minutes
Chicago Weather Index
LiveA Chicago rolling temperature index derived from live market signals. Values above 100 mean warmer than expected; below 100 means cooler.
- • Market: KXHIGHCHI
- • Baseline: NOAA O'Hare International normals
- • Update frequency: Every 5 minutes
Miami Weather Index
LiveA Miami rolling temperature index derived from live market signals. Values above 100 mean warmer than expected; below 100 means cooler.
- • Market: KXHIGHMIA
- • Baseline: NOAA Miami International normals
- • Update frequency: Every 5 minutes
LA Weather Index
LiveA Los Angeles rolling temperature index derived from live market signals. Values above 100 mean warmer than expected; below 100 means cooler.
- • Market: KXHIGHLAX
- • Baseline: NOAA Los Angeles Downtown normals
- • Update frequency: Every 5 minutes
Austin Weather Index
LiveAn Austin rolling temperature index derived from live market signals. Values above 100 mean warmer than expected; below 100 means cooler.
- • Market: KXHIGHAUS
- • Baseline: NOAA Camp Mabry normals
- • Update frequency: Every 5 minutes
Philadelphia Weather Index
LiveA Philadelphia rolling temperature index derived from live market signals. Values above 100 mean warmer than expected; below 100 means cooler.
- • Market: KXHIGHPHIL
- • Baseline: NOAA Philadelphia International normals
- • Update frequency: Every 5 minutes
Data Sources
Kalshi Prediction Markets
Real-time prediction market prices for daily high temperature outcomes. Kalshi is a CFTC-regulated exchange where traders buy and sell contracts on future events.
Visit Kalshi →NOAA Climate Normals
Historical average temperatures from NOAA's 30-year climate normals (1991-2020). Each city uses data from its local weather station to establish the seasonal baseline.
View NOAA Climate Normals →Trading
Perpetual trading on Attena indices is planned but not yet live. Here's the current status and how it will work.
Current Status
Demo Mode
The Long/Short buttons on index pages currently link to our waitlist for early access. Trading functionality is in active development.
The indices themselves are fully live. All index values update in real-time based on live prediction market prices. Only the direct trading interface (perpetuals) is pending launch.
How Perpetuals Will Work
Continuous Tracking
Index perpetuals will track the rolling index value continuously—no expiration, no contracts to roll. Hold a position indefinitely as long as you maintain margin.
Funding Rate Mechanism
A funding rate balances long and short positions: when the perpetual trades above the index, longs pay shorts; when below, shorts pay longs. This keeps the perp price anchored to the underlying index.
Hedgeable via Underlying Markets
Every perp position can be hedged using the underlying prediction market contracts (e.g., Kalshi temperature buckets for weather indices). This creates tight arbitrage bounds and reliable price discovery.
Why This Works
Replicability
Each index value is mathematically derived from observable market prices. Any trader can replicate the index by holding the constituent contracts.
Arbitrage Enforcement
If the perp diverges from the index, arbitrageurs can trade the spread profitably. This natural mechanism keeps prices aligned without central intervention.
Real Money Stakes
The underlying prediction markets have real capital at risk. This ensures the prices feeding our indices reflect genuine conviction, not cheap speculation.
Market Infrastructure
Attena indices require standardized prediction markets. Here's our current approach and future plans.
Current Markets (Weather)
Weather indices are built on existing Kalshi prediction markets:
- • Standardized structure: Consistent temperature buckets (5°F ranges) across all cities
- • Daily resolution: Contracts settle based on official NWS temperature readings
- • Continuous availability: New contracts open automatically for upcoming days
- • CFTC-regulated: Trades occur on a regulated exchange with real money
Future Categories
For Politics, Culture, and Attention indices, Attena will deploy its own standardized prediction markets. Why?
The Problem with Existing Markets
Existing prediction markets for politics, culture, and attention lack the consistent structure needed for rolling indices. Contract formats vary, resolution criteria differ, and continuous coverage isn't guaranteed. You can't build a perpetual on a one-off event market.
By deploying our own standardized markets, we ensure every category has the infrastructure for reliable, continuous index calculation—same structure, same resolution timing, same bucket design across all contracts.
Standardization Requirements
For any market to power an Attena index, it must meet these criteria:
Consistent Contract Structure
Same buckets, same resolution timing, same payout structure across all instances
Verifiable Outcomes
Clear, objective resolution criteria tied to authoritative data sources
Continuous Availability
New contracts roll automatically—no gaps, no manual intervention
Sufficient Liquidity
Enough trading activity for reliable price discovery and tight spreads
NBA Edge Index
The NBA Edge Index is a market-derived performance rating system that measures how well each NBA team performs relative to Polymarket betting expectations. Unlike traditional power rankings based on win-loss records, this index rewards beating the odds and penalizes underperforming expectations.
How It Works
The Core Formula
Rating Change per Game
Rating Δ = (Result − Market Prob) × K
Where K is a scaling factor (typically 30-50)
Initialize at Baseline
Every team starts the season at 2000. This is the neutral baseline—above 2000 means outperforming expectations, below means underperforming.
Fetch Pre-Game Probabilities
Before each game, we capture the Polymarket win probability for each team. For example, if the Lakers are priced at 60% to win, their Market Prob = 0.60.
Calculate Rating Change After Game
After the game, we compute the rating change. Result = 1 if team won, Result = 0 if team lost.
Example: Lakers (60% favorite) win
Rating Δ = (1 − 0.60) × 40 = +16
Example: Lakers (60% favorite) lose
Rating Δ = (0 − 0.60) × 40 = −24
Update and Rank
Ratings accumulate over the season. Teams are ranked by their current rating, revealing which teams consistently beat or miss market expectations.
Interpretation Guide
Rating Scale
Above 2000
Outperforming expectations
Below 2000
Underperforming expectations
Why Underdog Wins Matter More
The formula naturally rewards unexpected outcomes more than expected ones:
- • A 20% underdog winning gains +32 (1 − 0.20 = 0.80)
- • An 80% favorite winning gains only +8 (1 − 0.80 = 0.20)
- • An 80% favorite losing costs −32 (0 − 0.80 = −0.80)
This creates a rating that captures performance quality, not just win quantity. A team can have a losing record but a high rating if they consistently beat the spread.
Data Source
Polymarket NBA Markets
Pre-game win probabilities are derived from Polymarket's NBA game outcome markets. These markets have real money at stake and update in real-time based on injury news, lineup changes, and betting activity.
Visit Polymarket →Update Frequency
Ratings update after each game resolves. During the NBA season, this typically means multiple updates per day as games finish. Historical ratings are preserved, allowing you to track performance trends over the season.
NHL Edge Index
The NHL Edge Index is a rolling performance rating for every NHL team. It measures how well teams perform relative to what Polymarket prediction markets expected—rewarding underdog wins and penalizing favorite losses. All teams start at 2000.
How It Works
The Core Formula
Rating Δ = (Result − Market Prob) × K
K = 40 (scaling factor) • Result = 1 (win) or 0 (loss)
Initialize at Baseline
Every team starts the season at 2000. This is the neutral baseline—above 2000 means outperforming expectations, below means underperforming.
Fetch Pre-Game Probabilities
Before each game, we capture the Polymarket win probability for each team. For example, if the Bruins are priced at 65% to win, their Market Prob = 0.65.
Calculate Rating Change After Game
After the game, we compute the rating change. Result = 1 if team won, Result = 0 if team lost.
Example: Bruins (65% favorite) win
Rating Δ = (1 − 0.65) × 40 = +14
Example: Bruins (65% favorite) lose
Rating Δ = (0 − 0.65) × 40 = −26
Update and Rank
Ratings accumulate over the season. Teams are ranked by their current rating, revealing which teams consistently beat or miss market expectations.
Interpretation Guide
Rating Scale
Above 2000
Outperforming expectations
Below 2000
Underperforming expectations
Why Underdog Wins Matter More
The formula naturally rewards unexpected outcomes more than expected ones:
- • A 20% underdog winning gains +32 (1 − 0.20 = 0.80)
- • An 80% favorite winning gains only +8 (1 − 0.80 = 0.20)
- • An 80% favorite losing costs −32 (0 − 0.80 = −0.80)
This creates a rating that captures performance quality, not just win quantity. A team can have a losing record but a high rating if they consistently beat the spread.
Data Source
Polymarket NHL Markets
Pre-game win probabilities are derived from Polymarket's NHL game outcome markets. These markets have real money at stake and update in real-time based on injury news, lineup changes, and betting activity.
Visit Polymarket →Update Frequency
Ratings update after each game resolves. During the NHL season, this typically means multiple updates per day as games finish. Historical ratings are preserved, allowing you to track performance trends over the season.
Geopolitics Indices
Geopolitics indices track global conflict risk by aggregating real-money prediction market probabilities. The Global Conflict Risk Index is the first in this category, blending escalation probabilities across major conflict zones into a single, tradeable value.
How It Works
The Core Formulas
Blended Probability
P = Σ (weighti × probi)
Index Value
100 + 100 × (P − Baseline)
The index is calculated in real-time by fetching escalation probabilities from 10 Polymarket contracts across 4 categories, applying time-horizon and category weights, and comparing to a fixed quarterly baseline.
Fetch Market Data
Retrieve YES prices from 10 active Polymarket contracts covering major conflict regions at different time horizons
Calculate Escalation Probability
For each market, determine escalation probability. Some markets are inverted (e.g., "ceasefire by X" → 1 minus YES price = conflict continuation probability)
Apply Time-Horizon Weights
Within each category, weight markets by time to expiration: Near-term (0-60 days) 45%, Mid-term (60-120 days) 35%, Long-term (120+ days) 20%
Apply Category Weights
Multiply each category's probability by its weight: Russia/Ukraine 30%, China/Taiwan 30%, Middle East 30%, Tail Risk 10%
Compare to Fixed Baseline
Compare current probability to a fixed quarterly baseline (recalculated from prior 90 days each quarter). Index = 100 + 100 × (current − baseline)
Why a Fixed Baseline?
Unlike weather indices that compare to fixed NOAA normals, conflict risk has no natural "normal" level. We use a fixed quarterly baseline to:
- •Prevent drift — A rolling EMA causes the index to trend toward 100 over time
- •Enable comparison — Index values are comparable across months
- •Provide stability — Removes self-referential feedback loops
- •Allow interpretation — "Index at 108" means 8% above the quarterly baseline
The baseline is recalculated automatically on the first day of each quarter using the prior 90 days of probability data.
Component Markets
| Category | Weight |
|---|---|
| Russia / Ukraine | 30% |
| China / Taiwan | 30% |
| Middle East | 30% |
| Tail Risk | 10% |
Time-Horizon Weighting
Within each category, markets at different time horizons are weighted to emphasize actionable, near-term signals:
45%
Near-term
(0-60 days)
35%
Mid-term
(60-120 days)
20%
Long-term
(120+ days)
Near-term markets are more liquid and reflect imminent risks. As contracts approach expiration, they're replaced with successor contracts.
Interpretation Guide
< 95
Below Baseline
Lower than average risk
95–105
Near Baseline
Risk at historical average
> 105
Elevated Risk
Above average escalation risk
Data Source: Polymarket
Polymarket is a decentralized prediction market platform where users trade on the outcomes of real-world events with real money (USDC). It has processed over $1B in cumulative trading volume.
Polymarket prices are used because they represent aggregated market expectations with financial stakes, providing more responsive signals than traditional forecasting methods.
Politics Indices
Political event indices tracking election outcomes, policy changes, and domestic political developments. Will be powered by Attena-deployed standardized prediction markets with consistent contract structures for reliable index calculation.
Culture Indices
Cultural trend indices measuring entertainment, media, and social phenomena. Will be powered by Attena-deployed standardized prediction markets with consistent contract structures for reliable index calculation.
Attention Indices
Attention economy indices tracking celebrities, streamers, actors, and musicians. Will be powered by Attena-deployed standardized prediction markets with consistent contract structures for reliable index calculation.
FAQ
How often does the index update?
Weather indices sync with Kalshi market data every 5 minutes during market hours. Real-time updates are pushed to all viewers automatically.
Can I trade this index?
Not yet—but soon. The Long/Short buttons currently link to our waitlist for early access to perpetual trading. The indices themselves are fully live; only direct trading is pending launch. In the meantime, you can hedge via the underlying Kalshi temperature contracts.
What cities are supported?
Currently live: New York City, Chicago, Miami, Los Angeles, Austin, Philadelphia.
Why use prediction markets instead of weather forecasts?
Prediction markets aggregate information from many sources and participants with real money at stake. This creates a financial incentive for accuracy and often produces forecasts that are more responsive to new information than traditional weather models.
Why is Attena deploying its own markets for new categories?
Rolling indices require standardized, continuously available prediction markets. Weather works because Kalshi already has this structure. For politics, culture, and attention, no existing markets meet these requirements—so we're building them.