Recipe 24 of 33
Smart Money Index versus price: does it lead?
Pull a coin's whole hourly SMI series from /indices?hours=8760 and its hourly candles from /ohlcv, then correlate SMI at hour t with the return over the next k hours.
GET /v1/hl-cohorts/symbols/{symbol}/indices with hours=8760 answers the index half in one call: the coin's
Smart Money Index for every settled hour, newest first, as a ratio from -1 to +1. Pull the same hours of 1-hour
candles from /v1/hyperliquid/ohlcv, then correlate smi at hour t with the
return from t+1 to t+k for k = 1 to 24; a lead shows up as a correlation that holds at positive k and fades as k
grows. The full parameter list is on Coin indices.
Quick answer
One page holds the whole series while it is under 5000 rows.
# every hourly SMI reading of BTC up to a year back, in one page while the series is under 5000 rows
curl -H "X-API-Key: $VIKE_API_KEY" \
"https://data.vike.io/v1/hl-cohorts/symbols/BTC/indices?hours=8760"Example response
Captured from the live API on 2026-10-24 at 17:19 UTC, trimmed to 3 rows; series held 4,909 rows, back to 2026-04-03.
series is the test input: one row per settled hour, its pnl rungs labelled as they stood that hour
(label_basis: point_in_time). has_more is false, so the 4,909 rows are the whole history and no second page is
needed. latest is the newest 5-minute snapshot and stays out of the test.
{
"symbol": "BTC",
"label_basis": "point_in_time",
"latest": {
"ts": "2026-10-24T17:12:00+00:00",
"smi": -0.109548,
"wsi": -0.330428,
"whale_bias": -0.195726,
"smart_bias": -0.105831,
"label_basis": "current"
},
"series": [
{
"ts": "2026-10-24T16:00:00+00:00",
"smi": -0.099091,
"wsi": -0.329866
},
{
"ts": "2026-10-24T15:00:00+00:00",
"smi": -0.111421,
"wsi": -0.377369
},
{
"ts": "2026-10-24T14:00:00+00:00",
"smi": -0.116197,
"wsi": -0.371108
}
],
"has_more": false
}Variations
The candles to correlate with. Hourly bars from 2026-04-07, the first day the series covers the whole book; limit reaches 5000, and ts is epoch
milliseconds, the start of the hour.
# BTC hourly candles from 2026-04-07
curl -H "X-API-Key: $VIKE_API_KEY" \
"https://data.vike.io/v1/hyperliquid/ohlcv?symbol=BTC&interval=1h&start=2026-04-07T00:00:00Z&limit=5000"Page back once the series passes 5000 rows. When has_more is true, ask again with end set to the oldest ts of the page minus 1 second; end is inclusive.
# END is the oldest ts of the previous page minus 1 second, e.g. 2026-04-03T04:59:59Z
curl -H "X-API-Key: $VIKE_API_KEY" \
"https://data.vike.io/v1/hl-cohorts/symbols/BTC/indices?hours=8760&end=$END"A pinned window. start and end pin the study period; both are ISO-8601 UTC.
# BTC's SMI from the start of the hourly labels to the end of September
curl -H "X-API-Key: $VIKE_API_KEY" \
"https://data.vike.io/v1/hl-cohorts/symbols/BTC/indices?start=2026-08-06T00:00:00Z&end=2026-09-30T23:00:00Z"Before you rely on it
Correlate the series, never latest, and skip the edges
latest.smi ranks wallets by today's grading (latest.label_basis is current), so it does not belong in a
historical test. Start the study on 2026-04-07: the oldest rows are thin or null (2026-04-03 00:00 is null on both
indices). Before 2026-08-06 the pnl labels change once a day, after it every hour, so a series crossing that date
changes resolution; and leave out the newest few hours, which can still be revised after they first appear.
Ask your agent
Connected to the Vike MCP server (https://vike.io/mcp) in Claude, ChatGPT or Cursor, ask it in plain words.
The hl_market tool answers (view indices), with hl_ohlcv for the candles, at 1 credit a call each.
Give me BTC's hourly Smart Money Index since April and BTC's hourly closes for the same hours