Every preset is a measurement.
Nothing here is a settings suggestion. Each row is a configuration tuned on real data and shipped only after it held its sign across two disjoint windows. In August 2026 every one of them was also run on the years BEFORE its own — the last column — and most did not reproduce. Both numbers stay on the page.
Gold
XAUUSD · GC · MGC charts run these nativelyAug 2025 – Aug 2026| Preset | Chart | Delivery | Net | Profit factor | Win rate | Max drawdown | Trades | Outside its window |
|---|---|---|---|---|---|---|---|---|
| GOLD 5m REGULARS | 5m · REGULARS | Built into PRO v1.5 | +21.5% | 2.24 | 43% | 9.2% | 30 | +37 · PF 1.001Jan 2022 – May 2025 · 67 trades |
| GOLD 5m SCALPSHot-regime only — Apr–Aug shown | 5m · SCALPS | Built into PRO v1.5 | +25.1% | 2.76 | 50% | 4.1% | 28 | −53,791 · PF 0.534Jan 2022 – May 2025 · 135 trades · halves 0.96 → 0.29 |
| GOLD 5m MAGIC | 5m · MAGIC | Built into PRO v1.5 | +43.9% | 1.81 | 46% | 6.9% | 93 | −41,482 · PF 0.809Jan 2022 – May 2025 · 311 trades |
| GOLD 15m REGULARS | 15m · REGULARS | Built into PRO v1.5 | +13.1% | 2.66 | 47% | 2.0% | 15 | −1,322 · PF 0.931Jan 2022 – May 2025 · 25 trades |
| GOLD 15m SCALPS | 15m · SCALPS | Built into PRO v1.5 | +29.7% | 2.22 | 44% | 4.1% | 43 | +47,090 · PF 1.371Jan 2022 – May 2025 · 183 trades · halves 1.15 → 1.54 |
Each lane measured independently on XAUUSD — fixed sizing, $100,000 account, 1% risk per trade, no compounding.
Micro Gold futures
MGC1!Aug 2025 – Aug 2026| Preset | Chart | Delivery | Net | Profit factor | Win rate | Max drawdown | Trades | Outside its window |
|---|---|---|---|---|---|---|---|---|
| MGC SCALP v1 | 5m · SCALPS | Built into PRO v1.5 | +56.6% | 1.87 | 39% | 6.9% | 105 | −12,932 · PF 0.844May 2022 – Aug 2024 · 115 trades |
| MGC MAGIC v123 trades is a sample, not a track record | 15m · MAGIC | Built into PRO v1.5 | +38.8% | 3.09 | 39% | 6.6% | 23 | −2,597 · PF 0.959Jan 2022 – Aug 2025 · 91 trades |
Tuned on the first half-year, validated on the second, disjoint half — adopted only because the sign held in both. Same sizing convention as gold.
Crypto perpetuals
BTCUSDT.P · ETHUSDT.P · BybitAug 2025 – Aug 2026| Preset | Chart | Delivery | Net | Profit factor | Win rate | Max drawdown | Trades | Outside its window |
|---|---|---|---|---|---|---|---|---|
| BTC 5m SCALPS v2 | 5m · SCALPS | Built into CRYPTO DIGGA | +40.6% | 1.47 | — | 13.8% | 128 | −99,482 · PF 0.828Mar 2020 – Jul 2025 · 733 trades |
| BTC 15m SCALPS v1 | 15m · SCALPS | Built into CRYPTO DIGGA | +21.1% | 1.38 | — | 8.3% | 81 | −13,099 · PF 0.953Mar 2020 – Jul 2025 · 362 trades |
| BTC 15m MAGIC v149 trades, and its edge rests on 3–4 winners per half | 15m · MAGIC | Built into CRYPTO DIGGA | +17.1% | 1.48 | — | 6.3% | 49 | −12,899 · PF 0.939Mar 2020 – Jul 2025 · 260 trades |
| ETH 15m SCALPS v1Zero adoptions — the inherited engine ships untouched | 15m · SCALPS | Built into CRYPTO DIGGA | +41.3% | 1.96 | — | 6.4% | 71 | +62,229 · PF 1.285Oct 2020 – Jul 2025 · 310 trades · halves 1.52 → 1.07 |
Twelve cells measured, four shipped. Tuned on 2025-07-28 → 2026-02-09 and validated on the disjoint 2026-02-10 → 2026-08-23; adopted only where the sign held in both. Same sizing convention as gold. These figures are venue-specific — the same configuration measured on another exchange can differ by multiples, so both columns are Bybit throughout.
Every figure left of it comes from the same twelve months the sleeve was tuned on. That is standard practice and it is also why such tables flatter. So each sleeve was run again on the years before that window — data no tuning decision ever touched.2 of 11 reproduced.Every measurement, ranked →
The rest did not, and we are not going to dress that up. What it means is genuinely open: they may be fitted to the recent year, or the market of 2022–2024 may simply have been a different game — this test cannot tell those apart. It is also why the ones that passed count for more than a high number measured only on its own year: a configuration that fails on old data is ambiguous, while one that works in both structures is not.
One pattern is worth more than any single row: the two thinnest samples in this catalogue — 15 and 23 trades — produced the two highest profit factors on their own year (2.66 and 3.09) and the two weakest afterwards (0.93 and 0.96). Nothing was done wrong in producing them; fifteen trades simply cannot support a number. Read the trade count before the profit factor, always.
Backtest results — past performance is not indicative of future results. Figures are per-lane on the stated account convention; they are not cumulative across lanes. Out-of-sample windows differ in length between instruments, so those results are given in account currency rather than percent. Crypto figures are venue-specific: the same configuration measured on a different exchange can differ by multiples.
No paste field required for what ships built in.
Anything marked “Built into PRO v1.5” runs the moment you open the right chart — gold and MGC charts recognise themselves and load their own measured lanes. The asset-presets table on the chart tells you exactly what is running and why.
ULTRA subscribers can request a preset for their instrument — one request every two months — or run the same bench we use and tune their own. A configuration that survives the validation bar can be minted as a one-line string and handed to any other subscriber.
Who built it, on what instrument and timeframe, and every threshold inside — decoded on screen without applying a thing. A preset either loads exactly as it was measured, or it does not load at all.