Every automated trading failure story ends the same way: the system was right often, then wrong once in a way it could not afford. Risk management is the engineering that makes “wrong once” survivable — and in automated gold trading it can be enforced in code, which is precisely its advantage over human discipline.

Position sizing: risk a fixed fraction, always

The foundation is fixed fractional sizing: risk a small, pre-agreed fraction of the account on every idea, with position size scaled to equity and the distance to the planned exit. Wide stop, smaller position; tight stop, larger position; identical monetary risk either way. At a fraction of a percent per idea, it would take a run of losses so improbable as to indict the method itself before the account were meaningfully impaired.

Exits decided before entry

A stop-loss placed after a position moves against you is a negotiation; a stop placed at entry is a contract. Disciplined systems decide the exit before the entry — whether as a server-side stop order, or as an exit level enforced continuously by the algorithm itself. Either way it is honoured the first time, every time — no widening, no “one more candle.”

The daily loss lock

Per-trade exits cap each idea; the daily loss lock caps the day. If the account draws down a set amount intraday — for GOLD STRIKE clients, a limit agreed in the mandate — the system stops trading until the next session. This is the circuit breaker that prevents a bad morning from becoming a bad month, and it must be absolute: a lock the software can override is not a lock.

What a safe system never does

  • Martingale or grid recovery — doubling position size after losses. It converts a small edge into a guaranteed eventual ruin.
  • Unplanned averaging — improvising extra size to rescue a losing position. Deliberate, pre-mapped entry ladders around a level are a different discipline: sized as one idea in advance, never invented mid-loss.
  • Trading sudden news — spreads widen and slippage explodes precisely when models are least reliable.
  • Revenge trading — a loss changes nothing about the next setup. Code has no pride to restore, which is the point of using it.

The mandate model

The strongest arrangement is one where the risk limits are set by the account owner and enforced by construction — the agent can trade less than the mandate allows, never more. Ask any provider: can your system exceed my limits under any circumstances? The only acceptable answer is no. See how GOLD STRIKE’s risk mandate is structured.

Frequently asked questions

What is a good risk per trade?

Professional systems typically risk 0.25%–1% of the account per trade. Anything above 2% makes normal losing streaks account-threatening.

Are stop-losses guaranteed?

Standard stops can slip in fast markets. This is a second reason disciplined systems avoid news events, where slippage risk concentrates.

What drawdown is normal?

Every strategy draws down. What matters is that the observed drawdown stays within the bounds the sizing maths predicts — and that you were told the number honestly in advance.