Gold testing
GOLD JOURNAL METRICS
Win rate is the most quoted and least useful number in gold trading, because it means nothing without the size of the wins and losses next to it. This page sets out the metrics that actually change a decision, in the order they become reliable as your sample grows.
The metrics that matter on gold are expectancy in R, win rate paired with average R (never alone), MAE and MFE distributions, R by session hour, trading costs as a share of gross profit, and drawdown measured in R rather than currency. Below roughly 30–50 trades, most of these are still noise.
WHY R, NOT DOLLARS
Every metric on this page is expressed in R — the multiple of initial risk — rather than in dollars, because dollar figures are contaminated by position size changes, account growth and the specific lot size used on any given trade. A trade risking $50 that made $150 and a trade risking $200 that made $600 are the same result, 3R, and only R lets you compare them honestly across a growing account.
On XAUUSD this matters more than on lower-volatility instruments because position sizing genuinely varies trade to trade — a scalp with a $2 stop and a swing with a $15 stop on the same account will use very different lot sizes, and a journal measured in dollars will make the swing trades look far more important than they are simply because the numbers are bigger.
Convert every closed trade to R using the initial planned stop, not the trailing or adjusted one. Grading against a stop that moved in the trader's favour inflates the R figure and hides how much risk was actually on the table at entry.
GRADE EVERY TRADE IN R
- GOLD R MULTIPLE CALCULATOR
Grade every closed trade in R so the record survives changes in account size.
- GOLD RISK AND POSITION SIZE CALCULATOR
Turn balance, risk percentage and a dollar-per-ounce stop into a lot size on the 100-ounce contract.
THE METRIC-DEFINITION TABLE
These are the six metrics worth tracking on a gold journal, what each one is calculated from, and what it actually tells you that the others do not.
| Metric | Formula | What it tells you |
|---|---|---|
| Expectancy (R) | (Win% × Avg Win R) − (Loss% × Avg Loss R) | The average result you should expect per trade, in units of planned risk |
| Win rate vs Avg R | Wins ÷ Total trades, read alongside Avg Win R ÷ Avg Loss R | Whether a high win rate is hiding a poor risk-reward, or a low one is hiding a strong one |
| MAE (adverse excursion) | Furthest distance price moved against the position before it closed | Whether your stop is wider than winning trades ever actually need |
| MFE (favourable excursion) | Furthest distance price moved in your favour before the position closed | Whether your target or exit is leaving profit on the table |
| R by session hour | Expectancy in R grouped by UTC hour of entry | Whether the edge lives in a specific window or is spread evenly (it rarely is) |
| Cost ratio | (Spread + commission + swap) ÷ Gross profit | How much of the strategy's raw edge is being consumed by trading costs |
| Drawdown in R | Peak-to-trough equity decline, measured in cumulative R rather than currency | The real size of the worst run relative to your own edge, independent of account size |
EXPECTANCY IN R — THE HEADLINE NUMBER
Expectancy answers the only question that ultimately matters: on average, does this strategy make money once costs are included, and by how much per trade. Anything consistently above 0.15R per trade after real spread and commission is a workable intraday gold edge. Above roughly 0.3R sustained over hundreds of trades is exceptional, and a number that high over a small sample is more likely a data error or an unrepresentative run than a genuine edge.
Expectancy must be calculated after realistic XAUUSD costs, not before them. A backtested or paper expectancy of 0.25R can collapse to 0.05R once a $0.25 average spread and a $5 round-turn commission are subtracted from trades averaging a $6 stop — the cost is not a footnote, it is often a third of the theoretical edge on a scalp-sized stop.
Track expectancy on a rolling basis, not lifetime. A strategy that ran at 0.3R for its first 60 trades and 0.02R for its last 60 has a lifetime average around 0.16R that looks healthy while hiding a strategy that has, in practical terms, already stopped working.
WIN RATE VERSUS AVERAGE R — READING THEM TOGETHER
A 78% win rate on gold usually means tiny targets and a stop wide enough to erase a month of gains in two trades. Win rate in isolation tells you nothing about the health of a strategy; it must always be read next to the average win-to-loss ratio, because the two trade off against each other by construction in almost every rule set.
A 40% win rate at an average 2.5R win against a 1R loss produces a healthier expectancy, and a far smoother equity curve, than an 80% win rate at 0.25R wins against 1R losses. The second system needs to be right four times out of five just to break even before costs, and a single bad week of consecutive losses — entirely normal variance at that win rate — can undo a month of grinding gains.
The practical use of this pairing is diagnostic: if win rate is falling while average R per win is rising, the strategy may be shifting toward fewer, larger, more selective trades — not necessarily a problem. If both are falling together, that is unambiguous decay.
| Win rate | Avg win : avg loss | Expectancy | Verdict |
|---|---|---|---|
| 80% | 0.25R : 1R | (0.8×0.25) − (0.2×1) = 0.00R | Breakeven before costs — one bad week is fatal |
| 40% | 2.5R : 1R | (0.4×2.5) − (0.6×1) = 0.40R | Strong edge, survives a losing streak comfortably |
| 55% | 1R : 1R | (0.55×1) − (0.45×1) = 0.10R | Workable but thin — costs matter enormously here |
| 30% | 4R : 1R | (0.3×4) − (0.7×1) = 0.50R | High variance; needs a large sample to trust |
MAE AND MFE — LETTING THE DATA SIZE YOUR STOP AND TARGET
Maximum adverse excursion (MAE) is how far a trade moved against you before it eventually worked or was stopped out. Plotting MAE for winning trades only usually reveals that a stop is wider than it needs to be — if no winner on the whole sample ever moved more than $2.20 against entry, a $4.50 stop is donating roughly $2.30 of unnecessary risk on every single trade, win or lose.
Maximum favourable excursion (MFE) is the mirror image: how far a trade moved in your favour before it closed, win or lose. Plotting MFE across the full sample, including losers that briefly went positive, shows whether the target is capturing the available move. If the median trade reaches $6 in your favour and you routinely bank at $3, the exit rule — not the entry — is the binding constraint on results.
On XAUUSD, MAE and MFE distributions should be tracked separately by session, because a $3 stop that is generous in the compressed Asian range can be far too tight during a London expansion, and the excursion data will show this clearly within thirty to fifty trades per session bucket.
SIZE STOPS TO ACTUAL VOLATILITY
- GOLD ATR STOP CALCULATOR
Size the stop to gold volatility rather than to a round number you like the look of.
- GOLD SESSION VOLATILITY HEATMAP
Average XAUUSD hourly range by UTC hour and weekday, so stop distances match the hour you trade.
R BY SESSION HOUR
Grouping every closed trade's R result by the UTC hour of entry is one of the highest-yield exercises in a gold journal, because XAUUSD edge is rarely distributed evenly across the day. A setup that shows a flat 0.1R lifetime expectancy will frequently reveal, once split by hour, that it earns 0.4R between 07:00 and 09:00 UTC and loses money for the rest of the day.
This breakdown requires a reasonable sample per hour bucket — under ten trades in any single hour is not enough to act on. The fix for a low per-hour sample is not to abandon the analysis but to widen the buckets to session-level (Asian, London, overlap, rollover) until enough trades accumulate to narrow them further.
Once a genuine session or hour bias is found, the correct response is usually to restrict the strategy to its productive window rather than to abandon it. A strategy that only trades three hours a day at 0.4R is a better business than the same rule set spread across sixteen hours at a blended 0.05R.
COST AS A SHARE OF GROSS
The cost ratio — spread plus commission plus any swap, divided by gross profit before costs — answers a question expectancy alone cannot: how much of the strategy's raw edge is being handed to the broker rather than kept. On a scalp-sized gold strategy with average wins in the $8–15 range, a $0.25 average spread and a $5 round-turn commission can consume 20–35% of gross profit, which is a materially different business than the same strategy on a $50 average win where the same costs are under 5%.
This ratio should be recalculated whenever the account changes broker, account type, or lot size regime, because spread and commission structures vary enough between ECN, standard and raw accounts to change a marginal strategy into a losing one or vice versa.
A cost ratio above roughly 30% on an intraday gold strategy is a signal to either widen the average target, reduce trade frequency, or move to a lower-cost account type — the strategy's market edge may be intact while its net-of-cost edge is not.
PRICE YOUR OWN COSTS
- GOLD SPREAD COST CALCULATOR
Price what your gold spread costs per trade, per day and per year at your trade frequency.
- GOLD SWAP CALCULATOR
Overnight carry on a gold position across a holding period, including triple-charge Wednesdays.
DRAWDOWN IN R, NOT IN CURRENCY
Measuring drawdown in dollars ties the number to a specific account size and position sizing regime, making it useless for comparing periods where risk per trade changed. Measuring drawdown in cumulative R — the sum of R results from equity peak to trough — makes the number comparable across time and across account sizes, and lets you compare the worst run in your live data against what your backtest or demo sample implied was plausible.
A drawdown of 12R is a very different event depending on the strategy's typical trade-to-trade variance: for a system averaging 0.3R expectancy with wins around 2R, a 12R drawdown might be six consecutive losing trades and well within normal variance. For a system with a 0.05R expectancy and 1:1 win-loss, the same 12R drawdown is a much larger number of trades and a much stronger signal that something has changed.
Track the largest historical drawdown in R from backtesting or demo alongside the live drawdown in R. If live drawdown materially exceeds the worst modelled drawdown, that is a direct signal to reduce size and investigate rather than to wait for the equity curve to recover on its own.
SIZE FOR THE DRAWDOWN YOU ACTUALLY GET
- DRAWDOWN RECOVERY CALCULATOR
The gain a drawdown demands, in cash, in R and in trades at your current edge.
- RISK OF RUIN CALCULATOR
See what your risk percentage and win rate imply about the probability of hitting your ruin threshold.
WHICH METRICS MISLEAD AT LOW SAMPLE SIZE
Every metric on this page requires a minimum sample before it is more signal than noise, and treating an early reading as settled fact is the single most common way traders abandon a working edge or double down on a broken one. The table below is a rough, conservative guide — real minimums depend on the underlying variance of the strategy, and high-variance setups need larger samples than the numbers below suggest.
The general rule: any metric derived from a subset of trades (by session, by setup tag, by hour) needs its own minimum sample within that subset, not just a minimum for the overall journal. A strategy with 200 total trades but only 8 of them in the London-open hour has 8 data points for that hour, not 200.
| Metric | Unreliable below | Reasonably stable above |
|---|---|---|
| Win rate | ~50 trades | 100+ trades |
| Expectancy in R | ~30 trades | 100+ trades, and check it is not one outlier driving it |
| MAE/MFE distribution | ~30 trades per bucket | 50+ trades per session or setup bucket |
| R by session hour | ~10 trades per hour | 30+ trades per hour |
| Cost ratio | A single week | A full month across varied volatility conditions |
| Drawdown in R | One losing streak | Multiple independent drawdown events across regimes |
A single 8R winning outlier sitting inside a 40-trade sample can single-handedly move expectancy from negative to comfortably positive. Before trusting any metric on a small sample, remove the single best and single worst trade and recompute — if the conclusion changes materially, the sample is not yet large enough to act on.
PUTTING THE METRICS TOGETHER
No single metric on this page should drive a decision on its own. Expectancy tells you the average outcome; win rate and average R together tell you the shape of that outcome; MAE and MFE tell you whether the stop and target match the trade's actual behaviour; session-hour R tells you where the edge lives; cost ratio tells you how much of it survives execution; and drawdown in R tells you what the worst plausible run looks like relative to your own edge.
The practical monthly review reads all six together: has expectancy held, is the win-rate-to-R relationship stable, are MAE/MFE distributions consistent with the stop and target in use, is the session-hour bias still present, has the cost ratio changed with account type or volatility, and does live drawdown remain inside the range the backtest or demo implied. A change in any one metric in isolation is worth noting; a change in two or more together is worth acting on.
RISK NOTE
Leveraged gold carries a high risk of rapid loss. Everything on this page is a testing method, not financial advice, a signal, or a projection of returns.
GOLD TRADING FAQ
- What is a good expectancy for a gold strategy?
- Anything consistently above 0.15R per trade after real spread and commission is a workable intraday gold edge. Above roughly 0.3R sustained over hundreds of trades is exceptional and worth double-checking for a data error or an unrepresentative sample.
- Is a high win rate a good sign on gold?
- Only when read next to the average win-to-loss ratio. An 80% win rate at 0.25R wins against 1R losses is breakeven or worse before costs, while a 40% win rate at 2.5R wins is a far more durable edge.
- How do MAE and MFE help size a gold stop?
- MAE across winning trades shows the furthest price ever moved against you before working — a stop noticeably wider than that figure is donating unnecessary risk. MFE across all trades shows how far the move typically travelled, which tells you whether your target is leaving profit on the table.
- Why measure drawdown in R instead of dollars?
- A dollar drawdown is tied to a specific account size and position sizing regime and cannot be compared across time. Drawdown measured in cumulative R is comparable across account sizes and against what a backtest or demo sample implied was plausible.
- How many trades before session-hour R is trustworthy?
- Aim for at least 30 trades within a given hour bucket before trusting the number, and treat anything under 10 as anecdotal. Widen to session-level buckets (Asian, London, overlap, rollover) if hourly samples are too thin.
- What is the cost ratio and why does it matter on gold?
- It is spread plus commission plus swap divided by gross profit before costs. On a scalp-sized gold strategy this can consume 20-35% of gross profit, turning a strategy with a real market edge into one with little or no net edge after execution.
JOURNAL YOUR GOLD TEST FREE
Live XAUUSD signals on a public ledger, a free gold trade journal that syncs with MetaTrader 5, and the macro pulse before the London open.
EXPLORE GOLDHUNTS
- FREE GOLD SIGNALS
Live XAUUSD calls with entry, invalidation and target — tracked to outcome.
- FREE GOLD TRADE JOURNAL
Connect MetaTrader 5 and every XAU trade is recorded and analysed.
- GOLD MACRO PULSE
Real yields, the dollar, central bank demand and the data that reprices gold.
- LEARN GOLD TRADING
The fundamentals pillar: sessions, volatility, lot sizes, spreads and costs.
- XAU PAIRS
XAUUSD, XAUAUD, XAUEUR, XAUGBP, XAUJPY, plus futures and physical gold.
- GOLD STRATEGIES
Session breakouts, liquidity sweeps, mean reversion and news momentum on gold.
- TESTING GOLD STRATEGIES
Backtesting, demo-to-live, sample size and why gold edges decay.
- GOLD INTELLIGENCE
How gold signals are generated, validated and kept under human oversight.
- GOLD CALCULATORS
Pip value, lot size from your stop, margin, and a live UTC session clock.
- GOLD DATA API
Free JSON endpoints, OpenAPI and MCP tools so agents can call the gold desk.
- GOLD BROKERS & PLATFORMS
XAU symbol names on MT5, gold on Trading 212, and spread comparisons.
THE FULL GOLD TESTING CLUSTER
RELATED ON GOLDHUNTS
- Testing a Gold Strategy: XAUUSD MethodCluster hub
- Moving From Demo to Live on Gold: The Execution Gap That Costs MoneySame cluster
- Gold Paper Trading vs Demo Account: What Each Actually MeasuresSame cluster
- Gold Testing Mistakes That Void the SampleSame cluster
- Free Gold Trading CalculatorsCalculator
- Best Gold Trade Journal 2026: 10 Tools Ranked for XAUUSDComparison