{
  "market": "ETH",
  "direction": "short",
  "leverage": 750,
  "take_profit_bp": 15,
  "take_profit_price": 2413.52,
  "bust_distance_bp": 8.53,
  "bust_price_estimate": 2419.21,
  "expected_net_return_on_margin_pct": 82.6,
  "confidence": 0.35,
  "regime": "ranging",
  "key_levels": {
    "support": 2412.7,
    "resistance": 2421.0
  },
  "entry_mid": 2417.15,
  "data_time_utc": "2026-10-18T12:14:30Z",
  "reasoning": "Facts (Hyperliquid API, 12:14 UTC Sunday): ETH is -56 bp vs prevDayPx while BTC is +60 bp, ETH 15m highs have stepped down from 2428.7 to 2420.1 since the morning push, the last 10 recent ETH trades were ~90% sells by size, price sits just under the 60m VWAP and 1m EMA9/21, and funding is positive (longs pay) with mark 3.8 bp under oracle. ETH was chosen over BTC because BTC 1m realized vol is only ~1.9 bp/min, so a 15 bp BTC target hit within 30 minutes in just 2-4 of 60 replayed windows, while ETH at ~3.3 bp/min resolves inside the horizon (15-16 wins of 60 windows at 750x). Leverage is the 750x floor because the widest bust (8.5 bp, ~2.5 one-minute sigmas) is the only one that survives normal ETH noise, and the 15 bp target nets ~83% on margin after the haircut while stopping just above the 2412.7 hour low. Inference: this is a weak relative-weakness/lower-highs edge in a weekend range, not a strong trend, and the 60-minute replay suggests liquidation is more likely than the target (~32 of 60 windows busted first), which is why confidence is low.",
  "risks": "A routine 8.5 bp ETH bounce to ~2419.2 (the 2419.9-2421.0 highs from the last 15 minutes are above that) liquidates the position; a BTC push back toward 82,100+ or a short squeeze off the 2412.7-2413.8 support shelf would do it within minutes."
}