Skip to main content

Narrative Categories

MAGMA supports 24 narrative categories across two oracle modes:

  • Standard Oracle — resolved by weighted data-source consensus over empirical or evidential outcomes.
  • Consensus Oracle — resolved by multi-signal aggregation over collective social perception. Requires a Falsifiability Declaration and at least one measurable proxy metric at submission.

Exact per-source weights are intentionally not published; see Appendix B: Oracle Weights. The approximate weights shown below mirror the public values in Oracle Architecture.

Standard Oracle Categories

On-Chain and Financial

CategoryScopePrimary Sources
MARKETCrypto asset prices, DeFi metrics, on-chain statistics. The only category eligible for FAST (5-minute) narratives.Pyth Network price feeds (primary)
ECOSYSTEMProtocol TVL, user growth, on-chain activity metrics. Resolution extracts the protocol name from the thesis and compares current TVL against the stated target; supports directional theses.DeFiLlama (40%), Tavily (30%), Grok (30%)
ECONOMICSMacro indicators, central-bank decisions, trade data, inflation.FRED API, BLS API, Grok, Perplexity
BUSINESSCorporate events, M&A outcomes, earnings, executive changes.SEC filings, Reuters via Tavily, Grok, Perplexity
COMMODITIESCommodity price thresholds and supply data. Supported: WTI Crude, Brent Crude, Natural Gas, Copper, Wheat, Corn. Theses must state units.Alpha Vantage (75%), Grok reasoning (25%)
MACROMacroeconomic thresholds beyond ECONOMICS: Fed rate decisions, GDP growth, central-bank policy.FRED/BLS APIs (45%), Grok (30%), Perplexity (25%)
EQUITIESStock price thresholds, index levels (S&P 500, NASDAQ, Dow Jones), earnings beats/misses. Geo-blocked where securities prediction markets are restricted.Alpha Vantage (50%), Grok (30%), Tavily (20%)
FOREXFX rate thresholds: EUR/USD, GBP/USD, JPY/USD, and other major pairs. Theses must specify pair, direction, and target rate.Alpha Vantage (85%), Grok (15%)

Commodity unit conventions: Wheat and Corn prices are in cents per bushel; Oil prices are in USD per barrel. Theses must specify units explicitly.

Sports and Competition

CategoryScopePrimary Sources
SPORTSMatch outcomes, championships, player statistics, tournament results. Supports football/soccer, NBA, NFL, MLB, NHL, tennis, cricket.The Odds API (50%), Tavily (30%), Grok (20%)
ESPORTSTournament results, league standings, competitive gaming outcomes. Supports CS:GO, League of Legends, Dota 2, Valorant, Overwatch, Rainbow 6, Rocket League, Call of Duty.PandaScore (60%), Grok reasoning (40%)

The Odds API is rate-limited to 500 requests/month and aggressively cached.

Government and Policy

CategoryScopePrimary Sources
POLITICALElectoral results, policy implementations, official announcements.Government sources, AP/Reuters via Tavily
LEGALCourt rulings, regulatory decisions, legislative outcomes.Official legal databases via Tavily, Grok analysis
GEOPOLITICALInternational agreements, diplomatic outcomes, treaty status.Official government sources via Tavily, Grok analysis

Science and Technology

CategoryScopePrimary Sources
SCIENTIFICResearch findings, clinical trial results, academic publications.Published journals via Tavily, Grok analysis
TECHNOLOGYProduct launches, AI milestones, platform adoption.Tavily web search, Grok analysis, Perplexity
HEALTHClinical trial outcomes, drug approvals, public health events.WHO, FDA, PubMed via Tavily, Grok
CLIMATEEnvironmental data, emissions targets, policy outcomes.NOAA, government agencies via Tavily, Grok

Community and Ecosystem

CategoryScopePrimary Sources
ONCHAINProtocol-specific outcomes, MAGMA ecosystem milestones, on-chain metric thresholds. Resolution extracts a metric keyword (e.g. "TPS", "active wallets", "DEX volume") and compares against the stated target.Dune Analytics (70%), Grok (30%)
BUILDERDeveloper-ecosystem events, hackathon outcomes, project launches.Tavily (40%), Grok (35%), Perplexity (25%)
CASTFarcaster-native events and community milestones.Neynar (70%), Grok (30%)

Consensus Oracle Categories

MAGMA's consensus categories handle cultural and ideological narratives at scale. Resolution aggregates four signals — discourse, community (conviction-weighted backing distribution), research, and time-horizon — and requires supermajority weighted agreement across all of them.

CategoryScopePrimary Sources
SOCIALSocial-movement trajectories, platform growth, community developments.Neynar (50%), Twitter/X (30%), Grok reasoning (20%)
CULTURALMusic, art, entertainment, meme adoption, cultural-shift narratives.Grok (30%), Perplexity (35%), community signal (35%)
CONVICTIONIdeological claims, technology-adoption theses, societal-change predictions.Grok (30%), Perplexity (35%), community signal (35%)
SATIREClearly satirical market narratives. The satire flag must be set at creation time and is immutable.Grok (50%), Tavily (25%), Perplexity (25%)

Twitter/X is rate-limited to 100 searches/month and aggressively cached.

Consensus Oracle Resolution Requirements

Narratives in SOCIAL, CULTURAL, and CONVICTION must include:

  1. A Falsifiability Declaration (minimum 50 characters) specifying conditions under which the claim would be false.
  2. At least one measurable proxy metric evaluable at resolution time.

A Resolvability Score (0–100) is computed at submission. Narratives scoring below 40 are rejected.

Oracle Mode by Category

ModeCategoriesResolution Basis
StandardMARKET, ECOSYSTEM, ECONOMICS, BUSINESS, COMMODITIES, MACRO, EQUITIES, FOREX, SPORTS, ESPORTS, POLITICAL, LEGAL, GEOPOLITICAL, SCIENTIFIC, TECHNOLOGY, HEALTH, CLIMATE, ONCHAIN, BUILDER, CASTWeighted data-source consensus
ConsensusSOCIAL, CULTURAL, CONVICTION, SATIREMulti-signal aggregation with falsifiability and proxy-metric requirements

FAST Narrative Restriction

FAST narratives (5-minute resolution) are restricted to the MARKET category only — the only category resolvable via Pyth price feeds without AI reasoning or research verification. FAST narratives use a Pyth-primary, Switchboard-fallback path and require a minimum backing of 10 USDC equivalent.

ONCHAIN Category Guide

ONCHAIN narratives resolve using real on-chain data from Dune Analytics, surfaced through the MAGMA Intelligence feed. The oracle extracts a metric keyword from the thesis and queries the pre-cached Dune data.

Writing effective ONCHAIN narratives:

  1. State the metric explicitly — "Solana will reach 5,000 TPS", not "Solana will be fast".
  2. State the target value — include a numeric threshold the oracle can compare against.
  3. State the measurement window — "by end of month", "on any single day in June".
  4. Reference the chain — "on Solana", "across all L2s".

Supported ONCHAIN metrics:

MetricExample Thesis
TPS (transactions per second)"Solana will sustain above 3,000 TPS for 24 consecutive hours before June 30"
Active wallets / DAU"Solana active wallets will exceed 3 million in a single day by Q3 2026"
DEX volume"Solana DEX volume will exceed $10 billion in a single day before July 1"
TVL"A single Solana protocol will reach $5 billion TVL before mainnet launch"

Narratives that cannot be matched to a supported metric keyword receive a low Resolvability Score and are likely rejected.

  • Oracle Architecture — full source list, M-of-N consensus, and the challenge window.
  • Appendix B: Oracle Weights — why exact weights are not published.
  • MAGMA Intelligence — the Dune-powered feed behind ONCHAIN resolution.