# trending-1005 results

Run 2026-10-05 on the hunt2 harness, exactly as pre-registered in `prereg.md` (commit 48ba0874). 12 new cells ledgered once each, 4 earlier runs reused. Generated from `results.json`; charts data in `charts.json`.

**Verdict: none of the five techniques held up. 0 of 16 cells passed the registry gate.** After this batch the published family is 719 tests, 0 passed, lowest q 0.144. Every cell here has global q 1.00.

Columns: CAGR after costs, CAGR before costs (same run at zero cost), equal-weight benchmark CAGR (no costs), share of 200 random-date copies beaten, max drawdown rule / benchmark, excess CAGR in each half, trades, failed gates.

## Opening range breakout

| cell | role | universe | window | CAGR net | CAGR gross | bench | random beaten | maxDD rule/bench | halves (excess) | trades | fails |
|---|---|---|---|---|---|---|---|---|---|---|---|
| `tr5-orb-utc-1h-2r-cutoff` | best | crypto 1h x8 | 2017-08-17 to 2025-03-14 | -58.3% | -31.5% | 84.7% | 0.5% | 99.9% / 86.0% | -249.5 / -76.9 | 12141 | q, random, halves, net<=0 |
| `tr5-orb-utc-1h-eod` | variant | crypto 1h x8 | 2017-08-17 to 2025-03-14 | -65.1% | -22.3% | 84.7% | 11.5% | 100.0% / 86.0% | -245.6 / -90.8 | 19765 | q, random, halves, net<=0 |
| `s29-h-utc-orb` (reused) | variant | crypto 1h x8 | 2017-08-17 to 2025-03-14 | -50.2% | 5.8% | 84.7% | 87.5% | 99.5% / 86.0% | -223.2 / -81.1 | 18574 | q, random, halves, net<=0 |

## ICT Asia session sweep

| cell | role | universe | window | CAGR net | CAGR gross | bench | random beaten | maxDD rule/bench | halves (excess) | trades | fails |
|---|---|---|---|---|---|---|---|---|---|---|---|
| `tr5-ict-asia-sweep-be` | best | crypto 1h x8 | 2017-08-17 to 2025-03-14 | -48.1% | -21.6% | 84.7% | 0.0% | 99.3% / 86.0% | -254.4 / -47.9 | 10188 | q, random, halves, net<=0 |
| `tr5-ict-asia-sweep-2r` | variant | crypto 1h x8 | 2017-08-17 to 2025-03-14 | -48.5% | -22.1% | 84.7% | 0.0% | 99.4% / 86.0% | -254.4 / -48.9 | 10188 | q, random, halves, net<=0 |
| `tr5-ict-asia-sweep-hold` | variant | crypto 1h x8 | 2017-08-17 to 2025-03-14 | -47.7% | -21.0% | 84.7% | 0.5% | 99.3% / 86.0% | -253.4 / -48.6 | 10188 | q, random, halves, net<=0 |
| `s29-etf-turtle-soup` (reused) | variant | etf 1d x25 | 2005-01-04 to 2025-03-14 | -0.4% | -0.1% | 7.0% | 12.0% | 14.1% / 25.8% | -8.5 / -6.3 | 687 | q, random, halves, net<=0 |

## VWAP plus 9 EMA

| cell | role | universe | window | CAGR net | CAGR gross | bench | random beaten | maxDD rule/bench | halves (excess) | trades | fails |
|---|---|---|---|---|---|---|---|---|---|---|---|
| `tr5-vwap-ema9-trend` | best | crypto 1h x8 | 2017-08-17 to 2025-03-14 | -99.1% | -48.6% | 84.7% | 0.0% | 100.0% / 86.0% | -293.2 / -114.8 | 100047 | q, random, halves, net<=0 |
| `tr5-vwap-reclaim-ema9-exit` | variant | crypto 1h x8 | 2017-08-17 to 2025-03-14 | -73.5% | 32.3% | 84.7% | 20.0% | 100.0% / 86.0% | -264.7 / -92.0 | 39677 | q, random, halves, net<=0 |
| `social-ta-vwap-bounce` (reused) | variant | crypto 1h x8 | 2017-08-17 to 2025-03-14 | -49.8% | -1.1% | 84.7% | 1.5% | 99.5% / 86.0% | -245.6 / -63.9 | 16833 | q, random, halves, net<=0 |

## 9/21 EMA crossover

| cell | role | universe | window | CAGR net | CAGR gross | bench | random beaten | maxDD rule/bench | halves (excess) | trades | fails |
|---|---|---|---|---|---|---|---|---|---|---|---|
| `tr5-ema921-1h-both` | best | crypto 1h x8 | 2017-08-17 to 2025-03-14 | -51.3% | 15.0% | 84.7% | 19.5% | 99.7% / 86.0% | -232.4 / -77.5 | 21714 | q, random, halves, net<=0 |
| `h3-hourly-ema-9-21-cross` (reused) | variant | crypto 1h x8 | 2017-08-17 to 2025-03-14 | 21.8% | 80.9% | 84.7% | 100.0% | 70.6% / 86.0% | -112.9 / -34.1 | 9911 | q, halves |
| `tr5-ema921-daily-stocks` | variant | us_stock 1d x40 | 2016-09-15 to 2025-03-14 | 9.1% | 9.7% | 18.0% | 4.0% | 13.8% / 32.2% | -11.4 / -6.6 | 1862 | q, random, halves |

## Bitcoin 200-day line, golden cross, Uptober

| cell | role | universe | window | CAGR net | CAGR gross | bench | random beaten | maxDD rule/bench | halves (excess) | trades | fails |
|---|---|---|---|---|---|---|---|---|---|---|---|
| `tr5-crypto-200dma-30d` | best | crypto 1d x18 | 2017-08-18 to 2025-03-14 | 14.2% | 14.5% | 62.8% | 24.5% | 63.4% / 85.9% | -113.0 / -13.1 | 127 | q, random, halves |
| `tr5-crypto-golden-cross` | variant | crypto 1d x18 | 2017-08-18 to 2025-03-14 | 14.8% | 15.1% | 62.8% | 3.5% | 69.9% / 85.9% | -116.0 / -10.2 | 133 | q, random, halves |
| `tr5-crypto-uptober` | variant | crypto 1d x18 | 2017-08-18 to 2025-03-14 | 9.2% | 9.4% | 62.8% | 79.0% | 15.9% / 85.9% | -151.1 / 7.8 | 111 | q, random, halves |

## Reading notes

- Hourly crypto rules sit in cash much of the day while the benchmark (8 coins, 84.7% a year) is always long, so the benchmark gap mostly measures time out of the market. The random-copy percentile (same trades, same exposure, random dates) is the fairer read: the best-faith ORB, ICT and VWAP versions beat 0 to 1 of 200.
- Most hourly rules lost money **before** costs too (ORB best-faith -31.5%, ICT best-faith -21.6%). The VWAP reclaim and the always-in 9/21 cross were positive before costs and negative after.
- Closest calls: `h3-hourly-ema-9-21-cross` (earlier run) beat all 200 random copies but trailed the benchmark in both halves; `tr5-crypto-uptober` beat 79% of copies and was ahead in the second half only. Neither meets the gate; neither is an edge claim.
- Slow crypto rules (200-day, golden cross) cut max drawdown (63-70% vs 86%) and gave up most of the return: a risk change, not a win.
- ORB and ICT futures/stock evidence comes from earlier studies, quoted on the pages from their own files: `docs/research/orb-1003/PAGE-NUMBERS.json` and `results.json` (52 pass/fail cells, 0 passed), `research/liquidity-1001/results/cells.jsonl` (1057 cells, 0 survived BH).
