Your markets.
One data feed.
Read saved sentiment and evidence as JSON. Each record retains its source, timestamp, exact contract and model decisions.
/api/v1/signalsFilter a market or domain. Items are ordered by source publication time, newest first. API reads use saved history.
import requests
response = requests.get(
"https://YOUR-APP.replit.app/api/v1/signals",
params={"target_id": "politics-texas", "limit": 100},
timeout=30,
)
response.raise_for_status()
batch = response.json()
print(batch["items"])
# Use batch["next_cursor"] as cursor to read older items.Query parameters
| Parameter | Purpose |
|---|---|
domain | movies, politics, or companies |
target_id | A market ID from /api/v1/markets. |
source / status | Exact source name or evidence status. |
since | Items first received after an ISO timestamp with a timezone. |
limit / cursor | 1–500 items per page. Use the returned cursor for older items. |
Record fields
contract_id · direction_targetThe exact contract and outcome assessed. Waymo defaults to IPO confirmation before Jan 1, 2028; Texas Senate uses the configured party; Clayface uses its live critic-score rules.
market_implicationThe contract-specific mechanism selected by Jev, alongside the supporting evidence quote.
model_answers.directionBullish, bearish or neutral for the selected outcome, with class probabilities and model confidence. These are model judgments, not settlement odds.
relevance_probabilityNoul: probability from 0 to 1 that the source identifies the exact subject or race.
contract_connection_scoreScore from 0 to 100: unrelated, background, indirect driver or direct contract evidence.
evidence_strength_scoreScore from 0 to 100 for support in the supplied text. Linked articles and media are not assumed to have been read.
outcome_materiality_scoreScore from 0 to 100 for how consequential the reported development would be for this outcome if accurate.
information_novelty_scoreScore from 0 to 100 for added information compared with up to ten supplied earlier sources. Null when no comparison exists. Comparison IDs are retained.
aspects · model_answersValidated Choice, Score and Noul primitives, including each score's levels, probabilities and confidence.
contract_version · rubric_versionThe saved rules snapshot and grading definition. Read /api/v1/contracts/{contract_version} for the rules used.
signal_id · batch_idStable record and refresh IDs. Refresh adds up to ten distinct items per market; reanalysis preserves source IDs and first receipt times.
category · status · reasonThe evidence category and limits on using it as a signal. Low confidence or missing context remains visible.
published_at · first_seen_atSource publication and first receipt times. Original source text, URLs and optional analyst notes are retained.
sentiment_score · impact_score · summary_scoreLegacy research indices remain available for API compatibility. The website shows individual dimensions instead of a combined number.
Jev receives the source, the full live contract rules and comparison sources, then assigns the evidence dimensions independently. IPO preparation is distinct from IPO confirmation; audience enthusiasm is distinct from a critic score; polling is distinct from an election result.
Removed items are excluded from the feed. Feed access is public. Set FEED_API_KEY to require a bearer key.