#3: The numbers. Six years of gold, two risk profiles Part 01

#1: When average is not enough. A new approach to gold night trading
#2: The DNA of a night system. Philosophy before performance - Trading Systems - 15 June 2026 - Traders' Blogs

Part 3 of the launch series — July release. Previous: #1 When average is not enough · #2 The DNA of a night system.

The first two posts made a promise. #1 set the direction. #2 explained the philosophy and the character behind it — and both ended on the same line: no assumptions, no promises, only data.

This is where I keep that promise.

No story this time. No build-up. Just the test, the conditions it ran under, and the results — including the parts most vendors prefer to leave out.

The test conditions

A result is only worth as much as the conditions it was produced under. So those come first, not the profit number.

• System: NightScalper Project #1
• Instrument: Gold (XAUUSD)
• Period: 1 January 2020 – 6 June 2026 (about 6.4 years)
• Broker / account: RoboForex-ECN, EUR account, 1:100 leverage
• Position size: fixed micro lot (0.01)

The period matters. It contains the 2020 COVID crash, the 2022 inflation shock and rate cycle, and the 2024–2026 gold rally. This was not tested on a quiet stretch chosen to flatter the curve.

Two risk profiles, same engine

I ran the identical system at two risk settings, because "how much does it make" is the wrong question on its own. The right question is how much, against how much risk.

Same logic. Same trades, give or take. The only thing that changed was the risk dial — and the drawdown moved far more than the return did. That trade-off is the whole point of showing both.

The number that actually matters

There are two drawdown figures in that table, and the gap between them is the most honest thing I can show you.

The balance drawdown — based only on closed trades — looks almost too clean: 1.34% and 6.14%. That is the number marketing departments love.

The equity drawdown — what the account looked like at its worst while positions were still open — is the real one: 5.73% at standard risk, and 25.05% at the higher-risk setting.

I lead with the equity figure on purpose. An open position that is temporarily down is real risk, even before it closes. A quarter of the account drawn down at peak stress, at the higher setting, is not a footnote — it is the price of the higher return next to it. If you only ever see the closed-trade drawdown, someone is showing you the flattering half of the picture.

This is exactly what #2 meant by focusing on how it loses. The system recovered from these drawdowns across the full six years and ended in profit on both settings, with no full-period stop-out. But the path was not a straight line, and pretending otherwise would be dishonest.

What these numbers do not say

Equally important: the limits of this data.

• It is one broker, one symbol, one historical window. Different spread and execution conditions will produce different results.
• The system was developed against this history. Strong past behaviour is evidence, not a guarantee — markets do not repeat on command.
• A backtest is a model of the past, not a forecast of the future.

Forward, briefly

Live and demo observation has started. Two early, deliberately small samples:

• RoboForex demo, ~15 days: net positive, drawdown under 1%.
• A live account, ~2 days: net positive, drawdown under 2%.

These are far too short to prove anything, and I am not presenting them as proof. They are here because transparency means showing the early forward data too — not only the parts that have had years to settle.

The point

The philosophy in #2 was: a system you can trust is one whose behaviour you can predict. Numbers are how you check whether the behaviour matched the philosophy.

On six years of real-tick gold data, it did — at a return, and at a cost, both of which are now on the table in full.

No system is risk-free. A backtest is not a promise. Past performance does not guarantee future results.

More forward data will follow before the July release.