forex-scalping.com

A trader backtests a gold scalp, then goes live and the losses are twice what was expected. Often the strategy isn't the problem. The lot math is.

Common sources:

- Contract size differs by broker: 100 oz per lot is standard, but some use 10 oz or 1 oz. A 0.10 lot on one account can be ten times the exposure of 0.10 on another.
- Tick value confusion. On a 100 oz contract a $1 move in XAUUSD is $100 per lot. Traders who think in "pips" sometimes count 10 cents as a pip and size off by a factor of ten.
- Copying lot size from a forum post or signal without checking the contract.
- Prop accounts with different symbol specs from the broker you tested on.

A quick sanity check before any new gold account goes live: open the symbol specification, write down contract size and tick value, then place 0.01 and note the $ value of a 50-cent move. Compare with what you expect.

Example: risk $50 on a $3.00 stop. On 100 oz contracts that's 0.16 lots. On 10 oz it's 1.6 lots. Both are "correct" for their account and completely wrong for the other.

Has anyone been burned by this switching between brokers?

This is the exact structural leak that blows up mechanical scalpers transitioning from demo environments to live prop evaluations. The math failure you identified isn't just common; it is a systemic trap built into the fragmented retail CFD infrastructure. When transitioning a gold strategy between environments, the underlying alpha is rarely the culprit for sudden drawdowns—it is almost always an execution or specification mismatch.

To answer your question directly: yes, professional desks and serious retail traders see this daily. The classic failure pipeline involves a trader optimizing a gold breakout system on a standard 100-ounce historical feed with raw spreads, then deploying that exact strategy (or Expert Advisor) to an offshore broker or prop firm using a 10-ounce contract and marked-up spreads. The hardcoded sizing fails, and a mathematically proven strategy bleeds out in days.

Beyond contract size, the broker-switching trap usually triggers three other hidden execution variables.

The Decimal Quote Trap

Brokers quote XAUUSD in either 2 or 3 decimal places. On a 2-digit broker, a tick is 0.01. On a 3-digit broker, a tick is 0.001. Algorithmic traders who hardcode their position sizers for "points" instead of programming dynamic server queries will instantly over-leverage or under-leverage by a factor of 10 the moment they change feeds.

Margin and Leverage Cascades

A strategy's risk math might dictate a 0.16 lot size, but margin requirements for metals vary wildly. If your testing environment offered 1:200 leverage on XAUUSD, but your live prop account caps metals at 1:20, executing that 0.16 lot might consume your entire available free margin. This chokes the account, leading to premature margin stop-outs on minor pullbacks long before your actual structural stop is hit.

Synthetic Spread Amplification

A 50-cent stop loss is highly vulnerable to spread manipulation. Backtesting often assumes static or historically average spreads. Live accounts, particularly B-book prop accounts, synthetically thicken spreads during high-volatility windows (NY Open) or rollover (5:00 PM EST). The stop triggers not because the price action breached the level, but because the bid/ask spread expanded to swallow the stop order.

The institutional solution is to never trust manual entry for lot math. Professional execution scripts dynamically query the server's exact specifications (Tick Value, Contract Size, and Margin Currency) at the millisecond of execution, calculating the lot size relative to a strict percentage of the live account balance. If the broker changes the specs over the weekend, the math automatically adapts on Monday.

FTtrader, the 2 versus 3 decimal trap is a good addition, and it hits automated sizing harder than manual trading because nobody looks at the numbers. A defensive habit for EA code is to never hardcode a point size at all: query tick size and tick value for the symbol from the broker at startup and print them in the log. If a migration to a new broker changes either of those, the log shows it on day one rather than after the first oversized loss.