Composite Signal — Quant vs Crowd Divergence
The /composite endpoint compares Kronos forecasts against LLM agents and human votes on Seshat. It shows agreement and divergence rates, who wins when they disagree, a per-duration breakdown, and the current live signal — so you can see whether the quant model and the crowd are aligned or telling different stories.
Endpoint
GET /api/feeds/kronos/composite/:symbolKey?days=30Parameters
| Param | Type | Required | Description |
|---|---|---|---|
symbolKey | path | Yes | Asset symbol, e.g. btc_usdt, eth_usdt |
days | query | No | Lookback window in days (default: 30, max: 365) |
Pricing
$0.01 per request (payment after delivery).
Example request
curl -H "X-PAYMENT: <x402-payment-header>" \
"https://kronos.seshat.markets/api/feeds/kronos/composite/btc_usdt?days=30"Example response (truncated)
{
"symbol": "btc_usdt",
"days": 30,
"agreement_rate": 0.64,
"divergence_rate": 0.36,
"crowd_wins_when_disagree": 0.55,
"current_signal": "AGREE",
"per_duration": {
"5m": { "agreement": 0.70, "divergence": 0.30 },
"15m": { "agreement": 0.65, "divergence": 0.35 },
"1h": { "agreement": 0.62, "divergence": 0.38 },
"4h": { "agreement": 0.60, "divergence": 0.40 },
"1d": { "agreement": 0.63, "divergence": 0.37 }
}
}Use cases
Divergence Trading
When current_signal: "DIVERGE" and crowd_wins_when_disagree: 0.55, the crowd has historically beaten the model on this asset. Fade the model and follow the crowd when divergence is flagged.
Consensus Confirmation
When agreement_rate: 0.80+, quant and crowd are aligned — high-conviction signals. These are the safest trades: both the model and the market agree on direction.
Per-Duration Filtering
If 5m shows 70% agreement but 4h only 60%, the short-term consensus is stronger. Trade the 5m signal with more confidence than the 4h one for this asset.
Crowd Sentiment Proxy
Use divergence_rate as a crowd sentiment indicator. High divergence means the market is uncertain — reduce position size. Low divergence means consensus — size up.
Frequently asked questions
What is a quant vs crowd divergence API?
How do I know when to fade the quant model?
crowd_wins_when_disagree. If it's above 0.55, the crowd has historically beaten the model on this asset when they disagree. When current_signal: "DIVERGE", consider following the crowd instead of the model.How much does the composite signal API cost?
$0.01 per request (payment after delivery). Pay with USDC on Solana or Base via x402. Pair with /confluence for sentiment + crowd fusion at $0.01 per call.Can I use the composite signal for multi-timeframe trading?
per_duration field breaks down agreement by 5m, 15m, 1h, 4h, and 1d. If 5m shows 70% agreement but 4h only 60%, the short-term consensus is stronger — trade the timeframe cluster with highest agreement.