market-data-cron

name: market-data-cron description: Wrapper untuk 3 fetcher (ccxt generic, Hyperliquid native, Bybit V5 native) — output JSON schema seragam untuk konsumsi agent. version: 1.1.0 metadata: hermes: tags: [trading, market-data, cron, ccxt, hyperliquid, bybit] category: trading requires_toolsets: [terminal] config: - key: market-data-cron.exchange description: "Default fetcher: ccxt | hyperliquid | bybit" default: "ccxt" prompt: "Fetcher backend" - key: market-data-cron.symbols description: "Comma-separated symbol list (format depends on backend)" default: "BTC/USDT:USDT,ETH/USDT:USDT,SOL/USDT:USDT" prompt: "Symbols" - key: market-data-cron.timeframe description: "Default timeframe (1h untuk bias scan, 15m untuk execution)" default: "1h" prompt: "Timeframe" required_environment_variables:

name: EXCHANGE prompt: ccxt exchange id required_for: data fetch
name: SYMBOLS prompt: ccxt symbol list (comma-separated) required_for: data fetch
name: TF prompt: timeframe (1m, 5m, 15m, 1h, 4h, ...) required_for: data fetch

Market Data Cron — fetcher wrapper (3 backend)

Skill ini menjalankan salah satu dari 3 fetcher (di-bundle di panduan-openclaw/hyperliquid/):

Script	Backend	Symbol format	Funding scale	Catatan
`fetch_market_data.py`	ccxt generic (Binance, OKX, Bybit, dst)	`BTC/USDT:USDT`	per-8h	Multi-exchange via single script. Butuh `pip install ccxt`.
`fetch_hyperliquid.py`	Hyperliquid native `/info` API	`BTC` (bare coin)	per-1h (HL convention)	No deps — stdlib only. Output dapat field `open_interest`.
`fetch_bybit.py`	Bybit V5 public API	`BTCUSDT` (no slash)	per-8h	No deps — stdlib only. Field `open_interest` + `category` (linear/inverse/spot).

Pilih berdasar venue eksekusi:

Trade di Bybit → fetch_bybit.py (CEX, deep liquidity, semua pair major)
Trade di Hyperliquid → fetch_hyperliquid.py (on-chain perp, HYPE native)
Multi-exchange → fetch_market_data.py (ccxt, swap exchange via env)

Output schema identik untuk 3-tiganya (lihat di bawah) — scout & sniper SOUL.md bisa interchange tanpa perubahan kode.

Output schema sama untuk dua-duanya:

{
  "timestamp": ...,
  "timeframe": "15m",
  "symbols": [
    {
      "symbol": "BTC/USDT:USDT",
      "price": 67420.0,
      "funding_rate": 0.0123,
      "funding_label": "LONG_BIAS (0)",
      "ema21": 66980.5,
      "macro_bias": "BULLISH",
      "nearest_fvg": {
        "type": "bull",
        "top": 66950, "bottom": 66890, "mid": 66920,
        "gap_pct": 0.089, "dist_pct": 0.74,
        "status": "fresh"
      },
      "active_fvg_count": 3
    }
  ]
}

When to Use

Scout profile: cron-driven 15m scan
Sniper profile: on-demand 5m fetch sebelum confirm setup
Journal profile: end-of-session price snapshot untuk review

Procedure

ccxt generic (`fetch_market_data.py`)

EXCHANGE=binanceusdm \
SYMBOLS="BTC/USDT:USDT,ETH/USDT:USDT,SOL/USDT:USDT" \
TF=1h \
python3 fetch_market_data.py

Hyperliquid native (`fetch_hyperliquid.py`)

COINS=BTC,ETH,SOL,HYPE \
TF=1h \
python3 fetch_hyperliquid.py

Bybit V5 native (`fetch_bybit.py`)

SYMBOLS=BTCUSDT,ETHUSDT,SOLUSDT \
CATEGORY=linear \
TF=1h \
python3 fetch_bybit.py

Lalu parse stdout JSON, pakai field macro_bias, funding_label, nearest_fvg.dist_pct untuk klasifikasi GO-LONG / GO-SHORT / SKIP.

Funding rate label → bias signal

Label	Crowdedness	Action
EXTREME_SHORT	Short overcrowded	Contrarian long candidate
SHORT_BIAS	Slight short	OK long
NEUTRAL	Balanced	Either side OK
LONG_BIAS	Slight long	OK short
HEAVY_LONG	Long crowded	Contrarian short candidate
EXTREME_LONG	Long overcrowded	SKIP long, contrarian short

Pitfalls

Rate limit — semua script sleep antar symbol (0.15-0.5s), tapi kalau >10 pair, raise limit atau pisah ke 2 cron run.
FVG dist_pct > 3% di-skip otomatis (terlalu jauh untuk scalping).
nearest_fvg=null artinya tidak ada FVG fresh dalam range — biasanya skip pair.
Bybit funding = per-8h, Hyperliquid funding = per-1h. Threshold di fr_label() sudah disesuaikan per script — jangan tukar fungsi label antar script. Kalau bandingin angka raw funding_rate, scale ke unit yang sama dulu.
Bybit kline list = newest first (V5 spec). Script auto-reverse jadi oldest first sesuai schema. Kalau pakai endpoint lain, perhatikan order.

Trading execution (separate dari market data)

Skill ini cuma untuk read market data publik. Untuk eksekusi trade ada skill terpisah:

Bybit: install official skill dari https://github.com/bybit-exchange/skills — supports HMAC + RSA signing, auto-update, dedicated sub-account workflow.
Hyperliquid: TODO (lihat CATATAN.md) — pakai HL Python SDK dengan agent wallet sub-account.

Untuk saat ini, semua eksekusi tetap manual — sniper kasih level (entry/SL/TP), user place order di exchange UI atau API langsung.

Verification

Cek output ada error field per symbol — kalau ada (ccxt error, exchange down), skip pair tsb dan log warning. Jangan halusinasi data.