{
  "direction": "long",
  "leverage": 950,
  "take_profit_bp": 10,
  "take_profit_price": 81715.1335,
  "bust_distance_bp": 5.726315789473685,
  "bust_price_estimate": 81586.75408,
  "expected_net_return_on_margin_pct": 68.8131394281212,
  "confidence": 0.53,
  "regime": "ranging",
  "key_levels": {
    "support": 81610,
    "resistance": 81639
  },
  "entry_mid": 81633.5,
  "data_time_utc": "2026-10-17T11:00:51.855629+00:00",
  "reasoning": "Hyperliquid public API snapshots (https://api.hyperliquid.xyz/info; requests and responses in evidence/book.json, evidence/1m.json, evidence/5m.json and evidence/15m.json) show a bid/ask mid of $81,633.50 above the closed-bar 1m EMA9/21 at $81,627.85/$81,624.67, with 63.91 BTC of visible bids versus 40.12 BTC of asks; I infer a modest long bias after the rebound, within a range. Closed 1m return volatility contracted from 1.47 bp over 60 minutes to 0.62 bp over 15 minutes, so 950x combines a 5.73 bp bust cushion below the $81,610 recent low with substantial leverage rather than the tightest 1000x cushion. The 10 bp target is below the six-hour high of $81,737 and keeps about 72.4% of the raw gain after the stated deadband, haircut and win fee, yielding 68.81% on margin conditional on winning. Offline calculations in evidence/metrics.json found 4 target hits, 1 bust and 25 unresolved outcomes in 30 overlapping historical 30-minute long windows at these parameters; this limited sample does not establish positive expected profit or a calibrated success probability.",
  "risks": "A rejection at $81,639 followed by a break of $81,610 could reach the estimated bust price and forfeit all margin; displayed liquidity can disappear. The target requires a local breakout, most historical windows remained unresolved after 30 minutes, and candle trade prices cannot establish exact mid-price barrier ordering; news and scheduled events were not independently checked."
}