Historical Analogs — What Happened Last Time?
The /historical-analogs endpoint finds past audited situations similar to the current forecast — same direction, similar confidence and upside_prob. It shows what actually happened in those analogous cases: accuracy, median change, outcome distribution, max adverse/favorable excursion, and percentiles. This is empirical evidence from analogous situations — ideal for backtesting and adding context to a live forecast.
Endpoint
GET /api/feeds/kronos/historical-analogs/:symbolKeyParameters
| Param | Type | Required | Description |
|---|---|---|---|
symbolKey | path | Yes | Asset symbol, e.g. btc_usdt, eth_usdt |
Pricing
$0.01 per request (payment after delivery — settled only if the response is successfully sent).
Example request
curl -H "X-PAYMENT: <x402-payment-header>" \
"https://kronos.seshat.markets/api/feeds/kronos/historical-analogs/btc_usdt"Example response (truncated)
{
"symbol": "btc_usdt",
"current_forecast": {
"direction": "up",
"upside_prob": 0.68,
"confidence": 0.72
},
"summary": {
"total_analogs": 23,
"accuracy": 0.74,
"median_change_pct": 1.8,
"outcome_distribution": {
"up": 17,
"down": 6
},
"max_adverse_pct": -2.1,
"max_favorable_pct": 5.4,
"percentiles": {
"p05": -1.8,
"p25": 0.3,
"p50": 1.8,
"p75": 3.2,
"p95": 4.9
}
},
"analogs": [
{
"decision_id": "c3d4e5f6-a7b8-9012-cdef-234567890123",
"predicted_at": "2025-04-12T08:00:00Z",
"direction": "up",
"upside_prob": 0.67,
"confidence": 0.71,
"actual_change_pct": 2.1,
"correct": true
},
{
"decision_id": "d4e5f6a7-b8c9-0123-defa-345678901234",
"predicted_at": "2025-03-28T16:00:00Z",
"direction": "up",
"upside_prob": 0.69,
"confidence": 0.73,
"actual_change_pct": -0.4,
"correct": false
}
]
}Use cases
Empirical Outcome Evidence
When the model says "up with 68% confidence", analogs show what actually happened in 23 similar past situations. If 17 of 23 went up, the 74% analog accuracy validates the confidence estimate.
Tail Risk Assessment
Check max_adverse_pct: -2.1% and p05: -1.8%. Even in analogous situations that resolved correctly, the worst drawdown was -2.1%. Set stop-losses based on empirical analog data, not assumptions.
Analog Accuracy vs Confidence
If the model says 68% confidence but analog accuracy is only 55%, the model is overconfident. Use analog accuracy as a reality check on the model's own confidence estimate.
Position Sizing from Percentiles
Use the p50 (median outcome) and p95 (best case) to size positions. If the median analog change is +1.8% and p95 is +4.9%, size for the median, not the best case.
Frequently asked questions
What is a historical analogs API for crypto predictions?
How do I find what happened in similar crypto market situations?
GET /api/feeds/kronos/historical-analogs/btc_usdt. If the model says "up with 68% confidence", analogs show what happened in 23 similar past situations. If 17 of 23 went up, the 74% analog accuracy validates the confidence estimate.Can I use historical analogs for crypto tail risk assessment?
max_adverse_pct and p05 in the response. Even in analogous situations that resolved correctly, the worst drawdown might be -2.1%. Set stop-losses based on empirical analog data, not assumptions.How much does the historical analogs API cost?
$0.01 per request (payment after delivery). Pay with USDC on Solana or Base via x402. Combine with /similar-markets at $0.005 for prediction market analogs.