# BTC-USD perp trade decision (PaperTrade, 750–1000x)

**Read this first:** I could not fetch the Hyperliquid data this decision is supposed to use. The Hyperliquid `/info` API only accepts POST requests. In this run, every shell command failed before it started (`apply-seccomp: write /proc/self/setgroups ... Permission denied`), so I could not use `curl` or Python. The only HTTP tool that worked sends GET requests, and Hyperliquid answered those with **HTTP 405 Method Not Allowed**. Binance's API returned **HTTP 451**. So every number below comes from the **Coinbase Exchange public API** (BTC-USD spot) as a stand-in. It is **not** the Hyperliquid mid that PaperTrade fills at. Before you place the order, fetch the live Hyperliquid `l2Book` mid, then recompute `entry_mid`, `take_profit_price` and `bust_price_estimate` with the same bp distances. The usual spread between Hyperliquid perp and Coinbase spot is a few bp, which is the same size as the bust distance.

## Decision

```json
{"direction":"long","leverage":750,"take_profit_bp":10,"take_profit_price":81673.46,"bust_distance_bp":8.53,"bust_price_estimate":81522.24,"expected_net_return_on_margin_pct":54.3,"confidence":0.38,"regime":"trending_up","key_levels":{"support":81542,"resistance":81682},"entry_mid":81591.865,"data_time_utc":"2026-10-16T20:58:03Z","reasoning":"Since the 24h low near 80694 (around 17:15 UTC), the 15m candles have made nine straight higher lows (80797 to 81030 to 81197 to 81216 to 81230 to 81399 to 81465 to 81542 to 81542). The last 60 1m candles closed about +32 bp. At the snapshot, price had pulled back about 11 bp from the 81682 high toward support at 81542 that has held twice. 750x gives the widest survivable adverse move (8.53 bp), which puts the bust at 81522, below that support. A 10 bp target sits under the 81682 resistance and keeps about 72% of the gross gain after the haircut (54% return on margin).","risks":"The likeliest cause of liquidation is a break of the 81542 support during thin Friday-evening, post-US-close liquidity: single 1m candles in the last hour fell as much as 13 bp from their open. A second risk is a gap between the Hyperliquid mid and the Coinbase proxy price I used."}
```

## Evidence (facts — what the sources returned)

All of this was fetched on 2026-10-16 at about 20:53–20:58 UTC. Note that WebFetch can return cached responses up to 15 minutes old.

| Item | Value | Source |
|---|---|---|
| Last trade, bid / ask | 81591.86, 81591.86 / 81591.87 at 20:58:03 UTC | https://api.exchange.coinbase.com/products/BTC-USD/ticker |
| 24h volume (Coinbase) | about 4,910 BTC | same |
| Proxy mid | (81591.86 + 81591.87) / 2 = 81591.865 | derived |
| 1m candles, last 60 (19:54–20:53 UTC) | close 81392.33 → 81653.92 (**+32 bp**); high 81681.80, low 81393.73 (**35 bp range**) | https://api.exchange.coinbase.com/products/BTC-USD/candles?granularity=60 |
| 1m candles, last 15 | high 81681.80, low 81542.42 (17 bp range) | same |
| 15m candles, 24h | close 96 bars earlier 82022.8 → now about 81612 (**−51 bp over 24h**); 24h high 82831.01 (about 12:15 UTC); 24h low 80693.69 (about 17:15 UTC) | https://api.exchange.coinbase.com/products/BTC-USD/candles?granularity=900 |
| 15m lows since the 24h low | 80693.69, 80789.92, 80796.96, 80889.51, 81197.16, 81215.93, 81230.12, 81399.21, 81465.03, 81542.31, 81542.42 | same |
| Bids within 0.10% of best bid | 31.6 BTC (10.7 BTC within 0.05%); best bid 81592.3 | https://api.exchange.coinbase.com/products/BTC-USD/book?level=2 (ask side was outside the part of the page I read, so no imbalance figure) |

## Calculations

- 1 bp ≈ $8.16 at 81,592.
- Bust distance at 750x = 10000/750 − 4.8 = **8.53 bp** (about $69.6). Bust price ≈ 81591.865 × (1 − 0.000853) = **81522.24**.
- Take-profit price = 81591.865 × 1.0010 = **81673.46**.
- Haircut at a 10 bp target: move after the 0.2 bp deadband = 9.8 bp. scale = 0.90 / (1 + 1/14.7 + 100000/(1e6 × 0.00098 × 814.598)) = 0.90 / 1.1933 = 0.7542. Net = 0.00098 × 750 × 0.7542 × 0.98 = **54.3% of margin**.
- Alternatives I considered (750x, using a driftless random-walk hit probability p ≈ bust / (bust + TP)):
  - 5 bp: p ≈ 0.63, about +22% if it hits.
  - 10 bp: p ≈ 0.46, +54%.
  - 12 bp: p ≈ 0.42, +67%, but the target would be above the 81682 resistance.
  - 20 bp: p ≈ 0.30, about +119%.
  - **With no directional edge, every one of these has negative expected value**, roughly −0.23 to −0.34 of margin. The trade only pays if the uptrend drift is real. I picked 10 bp because it is the largest target that still sits below visible resistance.
- 1000x would leave only 5.2 bp to the bust. Several of the last 60 1m candles moved more than 5.2 bp against a long from their open within a single minute: 81607.71 → 81542.42 (8.0 bp), 81466.30 → 81399.21 (8.2 bp) and 81540.09 → 81432.01 (13.2 bp). That is why I used the minimum leverage. Of those 60 candles, 2 had an open-to-low drop larger than the 8.53 bp bust distance at 750x.

## Inferences (judgement, not fact)

- **Regime: trending_up on the 15m and 1m charts**, recovering inside a 24h downtrend. The higher-lows structure since about 17:15 UTC is the main evidence for going long.
- The snapshot price (81592) is a pullback of about 11 bp from the high. Buying here puts the entry near support rather than chasing the move.
- It is Friday 21:00 UTC, after the US equity close, heading into the weekend. Liquidity is usually thinner, which makes wick risk worse in both directions.

## Uncertainty and unanswered questions

- **Not observed:** Hyperliquid mid, mark/oracle premium, funding, open interest, Hyperliquid book imbalance, recentTrades CVD, and liquidation heatmaps. All the brief's Hyperliquid-specific metrics are missing because of the tool failure described at the top.
- I did not check the macro calendar for the next hour (major US releases are not normally scheduled for Friday evening, but I did not confirm that).
- Coinbase candles may lag by up to about 5 minutes because of response caching. The 1m series ends at 20:53 UTC while the ticker is from 20:58 UTC.
- Confidence of 0.38 reflects all of this. With an 8.53 bp bust, an ordinary 1m wick can liquidate the position.