Skip to main content

CEX Leading Signals & On-Chain Data

Quant Flow enriches every trading decision with three external data sources that have academic backing as leading indicators for crypto price action.

Academic Basis​

SourcePaperKey Finding
CEX Funding RateMDPI Mathematics 2026CEX price discovery is 61% more efficient than DEX
MVRV / SOPRScienceDirect 2025MVRV and SOPR validated as strong directional signals
Fear & GreedContrarian indicator (widely replicated)Extreme readings predict reversals

Three Data Sources​

1. Binance CEX Funding Rate​

What it is: Binance perpetual futures funding rate for the same asset being traded on Hyperliquid.

Signal logic: When the Binance funding rate diverges significantly from the Hyperliquid funding rate, it's a leading indicator:

  • Binance funding spikes positive → CEX longs crowded → potential reversal signal
  • Binance funding negative while HL is neutral → short pressure building on CEX

API: Public Binance Futures API (no key required).

Template variable: {{ cex_funding_signal }}

2. Fear & Greed Index​

What it is: alternative.me composite sentiment index (0 = Extreme Fear, 100 = Extreme Greed).

Signal logic (contrarian):

ReadingSignal
0–20 (Extreme Fear)Contrarian buy signal
20–40 (Fear)Mild bullish lean
40–60 (Neutral)No signal
60–80 (Greed)Mild bearish lean
80–100 (Extreme Greed)Contrarian sell signal

Template variable: included in {{ onchain_summary }}

3. On-Chain: MVRV & SOPR​

MVRV (Market Value to Realized Value):

  • MVRV > 3.5: Market overheated, historically precedes major corrections
  • MVRV < 1.0: Market undervalued, historically precedes major rallies

SOPR (Spent Output Profit Ratio):

  • SOPR > 1: Coins are being spent at a profit — trend continuation or distribution
  • SOPR < 1: Coins spent at a loss — potential capitulation / reversal

API: blockchain.info and Glassnode public feeds.

Template variable: {{ onchain_summary }}

Enabling Data Enrichment​

# config.yaml
enhanced_analysis:
  enabled: true    # base switch — enables all external data collection

All three sources are collected automatically when enhanced_analysis.enabled: true.

Graceful Degradation​

Auto-Degradation

If any external API is unavailable (network error, rate limit, service down), the system:

  1. Logs a warning
  2. Omits that data from the prompt
  3. Continues trading — decisions are not blocked

Trading never halts due to missing enrichment data.

How Data Is Injected​

The MarketDataEnricher class (src/data/data_enricher.py) collects all external data and formats it for prompt injection:

enriched = await enricher.enrich(symbol, market_data)

# enriched contains:
# - cex_funding_signal: str  (narrative description)
# - onchain_summary: str     (MVRV + SOPR + F&G in one paragraph)
# - fear_greed_value: int    (raw 0-100 score)
# - mvrv: float | None
# - sopr: float | None

These are passed to PromptManager and rendered into the Jinja2 template.

Interaction with FinCoT​

CEX signals and on-chain data feed directly into Step 3 (Sentiment Check) of the FinCoT reasoning chain:

Step 3 - Sentiment Check:
  - Binance funding: +0.021% vs HL +0.008% → CEX premium building
  - Fear & Greed: 74 (Greed) → slight contrarian caution
  - MVRV: 2.1 → neutral (below 3.5 overheating threshold)
  - SOPR: 1.04 → coins spent at mild profit, continuation signal

Interaction with Debate Agent​

When Debate is enabled, both BullAgent and BearAgent also receive the enriched data. This allows the bears to argue MVRV overheating while bulls argue SOPR continuation — a more realistic debate.

Performance Impact​

  • Latency: External API calls add ~500ms–2s per decision cycle (async, parallel fetching)
  • Token cost: ~200–400 additional tokens per decision (the summaries are kept concise)
  • Rate limits: Binance and alternative.me public APIs have generous limits; blockchain.info may throttle under heavy use

Next Steps​