{
  "market": "BTC",
  "direction": "short",
  "leverage": 800,
  "take_profit_bp": 15,
  "take_profit_price": 81425.17875,
  "bust_distance_bp": 7.7,
  "bust_price_estimate": 81610.291575,
  "expected_net_return_on_margin_pct": 92.579466,
  "confidence": 0.42,
  "regime": "ranging",
  "key_levels": {
    "support": 81447,
    "resistance": 81620
  },
  "entry_mid": 81547.5,
  "data_time_utc": "2026-10-20T11:13:47.798264Z",
  "reasoning": "Hyperliquid POST https://api.hyperliquid.xyz/info candleSnapshot, l2Book and recentTrades snapshots (2026-10-20 11:13 UTC; archived in research/) favor a BTC short over ETH: closed-candle 15-minute realized one-minute volatility was 4.00 versus 4.90 bp, and the initial 10-bp book imbalance was -28.1% versus -8.0%. BTC rejected 81620 and the refreshed mid fell to 81547.5 below its one-minute EMA9 of 81571.65, although refreshed book imbalance weakened to -5.6% and rising five- and fifteen-minute EMAs limit conviction. At 800x, the 7.7-bp bust buffer exceeds typical recent BTC one-minute ranges of 3.43 bp while a 15-bp target offers about 92.6% conditional net margin gain after the supplied haircut, requiring a break of 81447 support. This is a low-confidence directional inference: among 30 overlapping historical 30-minute candle paths at these distances, BTC shorts had 11 target-first, 14 bust-first and five unresolved outcomes; these trade-price paths are not calibrated mid-price win probabilities.",
  "risks": "A rebound through the estimated bust price before a support break liquidates the full margin; the recent 81620 high is beyond that bust price, and rising five- and fifteen-minute trends favor a squeeze. Book liquidity can disappear, volatility has expanded, and candle trade prices cannot resolve every intraminute mid-price crossing; the quoted return is conditional on take-profit, not an unconditional expected profit."
}