Risk Management in Gold Trading
The part that decides whether a strategy survives long enough for its edge to matter.
Size from the stop, not from a feeling
Position size is not chosen; it is derived. Decide what fraction of the account a single idea may lose, mark where the idea is wrong, and the size follows arithmetically. Choosing a lot size first and then placing a stop where it "looks right" is the same decision made backwards, and it produces wildly different risk on different days without anyone intending it.
Because gold's volatility swings so widely, a fixed stop distance means a fixed dollar risk attached to a completely variable probability of being hit. Sizing from ATR keeps the meaning of the risk constant.
The minimum lot problem
On a small account the arithmetic can produce a size below the broker's minimum. The honest response is to refuse the trade. Rounding up to the minimum silently takes more risk than the policy allows, and the failure is invisible until a run of losses arrives. This system refuses rather than rounds, and records the refusal.
Layering, and the discipline it requires
Adding to a position can be legitimate: if the original thesis is intact and price offers a better location, a second entry may be a better trade than the first. It becomes ruinous when "the thesis is intact" is replaced by "the trade is losing".
Two rules keep the distinction real. The total risk of a basket is reserved up front, so later layers draw from a fixed budget rather than adding new risk. And each layer must earn its place on its own conditions — never merely because the previous one is underwater. A published record of layer depth against results makes this checkable: in the research behind this system, deep layers performed dramatically worse than first entries, and that finding is on the performance page rather than buried.
Protecting profit without strangling the trade
Moving a stop to break-even the moment a trade is slightly ahead feels prudent and is usually expensive: normal gold noise then closes trades that would have worked. The alternative used here is a staged ratchet — reduce risk at +0.6R, break even (past costs) at +1.0R, lock a third of the peak at +1.5R, then trail behind fresh structure — where each stage can only move the stop toward profit.
Measured across roughly eleven thousand development trades, this raised profit retention from 40% to 49%. Worth stating plainly: it did not turn losing strategies into winning ones. Management is a way of keeping more of whatever edge exists, not a source of edge.
What drawdown actually tells you
Maximum drawdown is the largest peak-to-trough fall in the record. It is the number that decides whether a strategy is survivable in practice, because it is the experience an operator has to sit through. A strategy with a good average and a drawdown deeper than the operator's tolerance will be abandoned at the worst possible moment, which makes its average irrelevant.
Drawdown and the longest losing streak are both published for this simulated record, alongside the results that look better.
Every figure on this page comes from a simulated account. Simulated results are a record of what a system did, not a promise of what it will do. Nothing here is investment advice, and no result is guaranteed.