# BTC-USD perpetual trade decision for PaperTrade (HyperEVM)

Data fetched 2026-10-16 20:58:46 to 20:58:49 UTC from the Hyperliquid public info API (`https://api.hyperliquid.xyz/info`). Live mid at the l2Book fetch: best bid 81715.0, best ask 81716.0, mid 81715.5.

## Decision

```json
{
  "direction": "long",
  "leverage": 750,
  "take_profit_bp": 12,
  "take_profit_price": 81813.5586,
  "bust_distance_bp": 8.5333,
  "bust_price_estimate": 81645.77,
  "expected_net_return_on_margin_pct": 67.3,
  "confidence": 0.5,
  "regime": "trending_up",
  "key_levels": { "support": 81624, "resistance": 81803 },
  "entry_mid": 81715.5,
  "data_time_utc": "2026-10-16T20:58:49Z",
  "reasoning": "BTC is in a 3-hour recovery from the 80772 session low (+55 bp over 3h, +16 bp over the last hour) with higher 1m lows at 81480, 81624 and 81676, price above the 1m/5m/15m EMAs, 5m EMA slopes rising, and mid 46 bp above session VWAP. Realized volatility is contracting (5m candle ranges 5-11 bp versus a 19 bp 6-hour average; 1m close-to-close stdev about 4.3 bp), so the 8.5 bp bust distance at 750x is roughly two 1-minute sigmas while a 12 bp target clears the 60-minute high (81765) and the 17:30 spike high (81803) and keeps 75% of the raw gain after the haircut. 750x was chosen over higher leverage because each extra 1 bp of bust distance matters more than the 7-23 point difference in return on margin when the per-minute sigma is 4.3 bp.",
  "risks": "Liquidation most likely comes from a single 1m flush like the 20:45 candle (8.2 bp open-to-low) or a rejection at 81765/81803 that retraces to the 81646 bust price; the perp still trades at a 3 bp discount to oracle and the top of book is ask-heavy, so a stall at resistance is the main danger."
}
```

The same object is saved as `decision.json` in the repository root.

## Facts (observed in the fetched data)

| Item | Value | Source |
|---|---|---|
| Best bid / best ask / mid | 81715.0 / 81716.0 / 81715.5 | l2Book, 20:58:49 UTC |
| markPx / oraclePx | 81712.0 / 81740.0 (mark 3.4 bp below oracle) | metaAndAssetCtxs |
| premium | −2.94 bp | metaAndAssetCtxs |
| funding (hourly) | +0.00125% | metaAndAssetCtxs |
| openInterest | 41181.9 BTC | metaAndAssetCtxs |
| 24h notional volume | 2.88 B USD | metaAndAssetCtxs |
| prevDayPx | 82100 (mid is −46.8 bp vs prior day) | metaAndAssetCtxs |
| 24h high / low | 82898 / 80772, mid at 44% of range | 15m candles |
| 6h high / low | 82072 / 80772 | 5m candles |
| 60m high / low | 81765 / 81480 | 1m candles |
| Rate of change | 5m −5.2 bp, 15m +5.3 bp, 30m +17.7 bp, 60m +15.9 bp, 3h +54.9 bp, 6h −30.8 bp, 24h −77.3 bp | candles vs mid |
| Mid vs EMAs | 1m EMA9 +2.2, 1m EMA21 +4.3, 5m EMA9 +7.6, 5m EMA21 +19.0, 15m EMA9 +21.3, 15m EMA21 +22.4 bp | candles |
| EMA slope (last vs 5 bars ago) | 1m EMA9 +0.1, 1m EMA21 +1.2, 5m EMA9 +9.8, 5m EMA21 +9.6, 15m EMA9 +25.4 bp | candles |
| Session VWAP (00:00 UTC, 5m data available from 14:55) | 81344, mid +45.6 bp above | 5m candles |
| 24h VWAP | 81759, mid −5.4 bp below | 15m candles |
| 1m close-to-close stdev | 15m 4.36 bp, 30m 3.66 bp, 60m 4.31 bp | 1m candles |
| Average 1m high-low range | last 15: 4.5 bp, last 30: 4.1 bp, last 60: 4.4 bp | 1m candles |
| Average 5m range | last 6: 7.6 bp, last 12: 9.3 bp, last 72 (6h): 19.5 bp | 5m candles |
| Average 15m range, 24h | 27.8 bp | 15m candles |
| Book within 10 bp of mid (top 20 levels returned) | bids 60.4 BTC, asks 77.1 BTC, imbalance −0.12 | l2Book |
| Last 10 trades | buy 2.85 BTC vs sell 0.05 BTC, average size 0.29 BTC, span 10 s | recentTrades |
| Funding history 15:00-20:00 | +0.0010% to +0.0013% hourly, premium −2.2 to −4.2 bp every hour | fundingHistory |
| 1m volume | last 5 avg 41 BTC vs 60m avg 21 BTC | 1m candles |

### Adverse excursion against the bust distance

Share of candles whose high−open or open−low exceeded the bust distance:

| Leverage | Bust distance | 1m candles (last 61) | 5m candles (last 73) |
|---|---|---|---|
| 750x | 8.53 bp | 4 (7%) | 56 (77%) |
| 800x | 7.70 bp | 7 (11%) | 57 (78%) |
| 900x | 6.31 bp | 8 (13%) | 64 (88%) |
| 1000x | 5.20 bp | 11 (18%) | 65 (89%) |

Within the last hour the only 1m candles that moved more than 8.5 bp against a long on an open-to-low basis were none; the largest open-to-low was 8.2 bp (20:45). The largest open-to-high was 8.0 bp (20:52).

### Path simulation, first touch of target vs bust, 30-minute horizon

Entering at each candle open and walking forward. A candle that touched both target and bust is counted as a loss.

| Data | Leverage / target | Long W / L / unresolved | Short W / L / unresolved |
|---|---|---|---|
| 1m, last 60 min | 750x / 12 bp | 33 / 16 / 11 | 7 / 44 / 9 |
| 1m, last 60 min | 750x / 15 bp | 18 / 16 / 26 | 2 / 46 / 12 |
| 1m, last 60 min | 1000x / 10 bp | 32 / 23 / 5 | 11 / 45 / 4 |
| 5m, last 6 h | 750x / 12 bp | 26 / 43 / 3 | 33 / 38 / 1 |
| 5m, last 6 h | 1000x / 10 bp | 24 / 48 / 0 | 28 / 44 / 0 |

### Haircut table (BTC parameters from the brief)

| Target | Share of raw gain kept | Return on margin at 750x | 800x | 900x | 1000x |
|---|---|---|---|---|---|
| 8 bp | 69.2% | 42% | 44% | 50% | 55% |
| 10 bp | 72.4% | 54% | 58% | 65% | 72% |
| 12 bp | 74.7% | 67% | 72% | 81% | 90% |
| 15 bp | 77.1% | 87% | 93% | 104% | 116% |
| 20 bp | 79.7% | 120% | 128% | 143% | 159% |

## Inferences

- **Direction long.** The 3-hour recovery with higher 1m lows, price above every EMA tested, rising 5m EMA slopes, and above-average 1m volume on the up-candles (20:44, 20:50, 20:52, 20:56) is the main evidence. The 1m path simulation over the last hour strongly favours long, but that sample is the uptrend itself and overstates the edge.
- **Against long.** The 5m simulation over 6 hours slightly favours short because that window contains the 17:00-17:30 selloff. The perp trades 3 bp under oracle with a negative premium every hour since 15:00 (persistent background state, not a fresh signal), the top of book is ask-heavy, and the day is still down 47 bp versus the prior day. The last 5 minutes are −5 bp.
- **Regime: trending_up on the 1-3 hour horizon, inside a down day.** Volatility is contracting from a very volatile afternoon (15m ranges of 40-117 bp between 15:00 and 19:30) toward 5-11 bp 5m ranges.
- **Leverage 750x.** 1m sigma is about 4.3 bp, so 8.5 bp of bust distance is about two sigmas per minute. At 1000x the 5.2 bp bust was exceeded by 18% of 1m candles in the last hour versus 7% at 750x, while the return on margin for a 12 bp target only rises from 67% to 90%. The survival difference dominates.
- **Target 12 bp (81813.56).** It sits above the 60-minute high (81765) and marginally above the 17:30 spike high (81803), so a fill requires a genuine breakout, but 12 bp keeps 75% of the raw gain versus 69% at 8 bp, and in the last hour 33 of 60 1m entry points reached +12 bp before −8.5 bp. A 15 bp target left 26 of 60 entries unresolved within 30 minutes, which is too slow for the brief.
- **Time of day.** 21:00 UTC Friday is after the US equity close. No US macro releases remain today (Industrial Production and Capacity Utilization were at 13:15 UTC per the TD Economics calendar linked below) and no Fed speakers were found for this evening, so no scheduled event should hit inside the 30-minute window.

## Uncertainty and unanswered questions

- No BTC-specific news for 2026-10-16 was found through web search; the cause of the 17:00-17:30 UTC drop to 80772 and the sharp rebound is unknown. A repeat of that headline-driven move would liquidate either direction instantly.
- The l2Book with nSigFigs 5 returns only 20 levels per side, all within about 2 bp of mid, so resting liquidity between entry and the 12 bp target or the 8.5 bp bust could not be measured.
- recentTrades returns only the last 10 prints (10 seconds), so the cumulative volume delta figure is not meaningful beyond the immediate tape.
- Open interest change versus price change could not be computed because only one OI snapshot is available from the API without historical polling.
- No liquidation heatmap or other-exchange order flow was consulted; those sources were out of reach within the time budget.
- The 0.2 bp deadband, scale formula, and 2% win fee were taken from the brief as given and not verified against PaperTrade's contract.
- The decision is a 50% confidence call. The expected value is roughly neutral given a 4.8 bp bust trigger offset; the trade is a bet that the hour-long upward drift continues for one more leg before the next flush.

## Sources

- Hyperliquid info API, POST `https://api.hyperliquid.xyz/info` with the candleSnapshot (1m/5m/15m), metaAndAssetCtxs, l2Book (nSigFigs 5), recentTrades and fundingHistory requests described in the brief. Raw responses were fetched 2026-10-16 20:58:46-20:58:49 UTC.
- TD Economics weekly calendar for the week of 2026-10-09 listing the 2026-10-16 US releases: https://economics.td.com/domains/economics.td.com/documents/reports/bottom-line/tables/2026-10-09/upcoming.htm
- Web search for BTC news on 2026-10-16 returned no same-day reporting (results were 2022-2025 articles and are not cited).