Is notional bias or equity-adjusted bias the better signal?

One /segments/{cohort}/history call with include=leverage returns both bias (long minus short over notional) and perp_bias (long minus short over equity) per hour since 2026-08-11; score each against the forward return from /ohlcv.

GET /v1/hl-cohorts/segments/{cohort}/history with include=leverage answers it in one call: it puts the two biases on the same hourly row. bias is (long - short) / (long + short), from -1 to +1, and ignores how much capital stands behind the positions; perp_bias is (long - short) / equity, which grows when the cohort leans harder on its own account. Pull one cohort from 2026-08-11, when the equity ratios begin, and correlate each series with the next hours' return. The full parameter list is on Cohort history.

Quick answer

The top 1% by realized PnL, from the first hour with equity ratios.

cURL

# the 3x smart cohort's hourly bias and perp_bias since the equity ratios begin
curl -H "X-API-Key: $VIKE_API_KEY" \
  "https://data.vike.io/v1/hl-cohorts/segments/3x_smart/history?axis=pnl&include=leverage&start=2026-08-11T00:00:00Z"

Example response

Captured from the live API on 2026-10-24 at 17:19 UTC, trimmed to 3 rows of 1,793: the newest with values and the two oldest. label_computed is left out. The top 1% of wallets by realized PnL were net short in both measures, but by different amounts: at 14:00 on 2026-10-24 bias was -0.108 of notional while perp_bias was -0.415 of equity. Since 2026-08-11 the notional lean barely moved (-0.126 to -0.108) and the equity lean shrank from -0.646 to -0.415. That gap is what the test asks about: does the equity-scaled number predict better?

Example response

{
  "axis": "pnl",
  "cohort": "3x smart",
  "label_basis": "point_in_time",
  "data_from": "2026-04-07T00:00:00+00:00",
  "series": [
    {
      "ts": "2026-10-24T14:00:00+00:00",
      "long_usd": 3562896704.0,
      "short_usd": 4423874705.0,
      "bias": -0.107801,
      "position_count": 21587,
      "perp_bias": -0.4154909975,
      "equity_exposure": 4.727841144,
      "equity_covered_share": 0.9932140373
    },
    {
      "ts": "2026-08-11T01:00:00+00:00",
      "long_usd": 2649775268.0,
      "short_usd": 3422563643.0,
      "bias": -0.127264,
      "position_count": 17269,
      "perp_bias": -0.6412524901,
      "equity_exposure": 4.30967204,
      "equity_covered_share": 0.9978203859
    },
    {
      "ts": "2026-08-11T00:00:00+00:00",
      "long_usd": 2638926840.0,
      "short_usd": 3399694652.0,
      "bias": -0.125984,
      "position_count": 17236,
      "perp_bias": -0.646397416,
      "equity_exposure": 4.30734456,
      "equity_covered_share": 0.9996344548
    }
  ],
  "has_more": false,
  "leverage": {
    "data_from": "2026-08-11T00:00:00+00:00",
    "scope": "native perp account only; main-dex positions of wallets with a known positive native equity",
    "note": "hours rebuilt from fills are null"
  }
}

Variations

The returns to score against. The series sums every coin, so score it against a broad move; BTC's hourly candles are the simplest proxy.

cURL — 1h candles

# BTC hourly candles over the same span
curl -H "X-API-Key: $VIKE_API_KEY" \
  "https://data.vike.io/v1/hyperliquid/ohlcv?symbol=BTC&interval=1h&start=2026-08-11T00:00:00Z&limit=5000"

Another cohort. The whale rung on the account-size ladder, the same two fields.

cURL — whale

# whales' hourly bias and perp_bias since 2026-08-11
curl -H "X-API-Key: $VIKE_API_KEY" \
  "https://data.vike.io/v1/hl-cohorts/segments/whale/history?axis=size&include=leverage&start=2026-08-11T00:00:00Z"

Before you rely on it

The two biases are built from different positions

bias is computed over the cohort's whole long_usd and short_usd, while perp_bias counts only main-dex positions of wallets with a known positive native equity, so perp_bias is not bias times equity_exposure (at 14:00: -0.510 against -0.415). The equity series starts on 2026-08-11, which leaves about 1,800 hours, and some of them are null: drop those hours from both series so the comparison runs on the same rows.

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_segments tool answers (view history), with hl_ohlcv for the candles, at 1 credit a call each.

Ask your agent

Give me the 3x smart cohort's hourly bias and perp_bias since 2026-08-11 so I can compare them with BTC's next-hour return

Is notional bias or equity-adjusted bias the better signal?

Quick answer

Example response

Variations

Before you rely on it

Ask your agent

Related