Learn
Gold trading guides – backed by real backtest data
Each article answers one question new traders keep asking, and ends with results run on many years of gold data. No hypothetical examples, no entries picked after the move already happened.
Money management
3 min read
A 95% win rate – and the account still blows up
Why win rate alone tells you nothing, and 3 questions to ask before following any signal group.
Risk management
3 min read
A 20-pip stop on gold: why it almost always gets hit
Gold now moves ~1,050 pips a day on average. Set the stop by structure first, then size the lot.
Trend
3 min read
Catching tops and bottoms: what the data says about trading against the trend
Same method, but selling into a rising market dragged results from break-even into the red.
Coming soon: Pending orders or wait for a confirmation candle? · Close everything at the first take-profit or let it run? · How much should you risk per trade? · Do SMC and FVG still work on gold? · Support and resistance: why you shouldn't chase price · MACD, RSI, Bollinger: which indicator still has an edge?
Money management · 3 min read
A 95% win rate – and the account still blows up
Every signal group brags about "80%, 90% wins". But win rate is only half the story. The other half – how much you make when you win and how much you lose when you don't – is what decides whether you end up in profit.
The formula to remember: average profit per trade = (win rate × average win) − (loss rate × average loss).
From that you get the minimum win rate needed to break even (before costs). In the table, take-profit (TP) and stop-loss (SL) are in pips – 1 pip is a $0.1 move in the gold price.
| Trade setup | Reward : risk | Minimum win rate |
|---|---|---|
| TP 30 pips, SL 70 pips | 1 : 2.33 (small wins, big losses) | 70% |
| TP = SL | 1 : 1 | 50% |
| TP twice the SL | 2 : 1 | 33.3% |
| TP three times the SL | 3 : 1 | 25% |
CashbackEA backtest · winning often, still losing money
- 6,000 gold trades with TP 30 / SL 70: 68% winners, yet −13,800 pips after spread.
- Grid / averaging down: 95% of baskets win, but one losing basket wipes out 120–400 winning ones.
- Closing everything at the first TP: a prettier win rate, but most of the profit is gone.
The CashbackEA bot is no exception: the free channel suggests closing half early, so it wins ~6/10 trades; the full signal group holds trades to target and wins only ~4–5/10, yet makes ~2x the profit.
CashbackEA bot strategies · gold 2015–2026 · after spread · fewer wins, still profitable
| Strategy | Win rate | Avg. profit per trade |
|---|---|---|
| System 23 | 21% | +0.16R |
| System 12 | 17% | +0.13R |
| System 24 | 41% | +0.27R |
| System 15 | 53% | +0.24R |
R = one unit of risk: +2R means a profit of twice the amount you were willing to lose. +0.16R = on average, each trade makes 16% of the amount you risked.
System 12 wins fewer than 1 in 5 trades and is still profitable, because its winners run very far. The price you pay is psychological: over 11 years of backtesting (2015–2026), the longest losing streaks across these strategies were 6–14 trades in a row.
Before you follow any signal group, ask 3 questions:
- How big are the average win and the average loss?
- Do they publish their losing trades too?
- What is the longest losing streak?
See the numbers for 25 popular methods in our backtest library.
Next: A 20-pip stop on gold – why it almost always gets hitRisk management · 3 min read
A 20-pip stop on gold: why it almost always gets hit
"Keep the stop-loss (SL) tight to stay safe" sounds sensible. On gold, though, it's often the reason traders lose steadily: the trade is in the right direction, but SL gets tagged before price takes off.
Why: gold now swings far more each day than it used to. The average daily range (1 pip = a $0.1 move) has grown almost 4x in 3 years:
| Year | Avg. daily range | A 20-pip SL equals |
|---|---|---|
| 2024 | ~286 pips | 7% of the daily range |
| 2025 | ~528 pips | 4% of the daily range |
| 2026 (through 10/2026) | ~1,050 pips | 2% of the daily range |
CashbackEA backtest
- Fixed 20-pip SL, every TP level tested: negative across the board.
- QML pattern on M15, SL ~25 pips, 1,113 trades 2020–2026: 24% winners, −0.39R per trade, negative in all 7 of 7 years. Spread alone ate 0.1R per trade.
- The profitable CashbackEA bot strategies typically use stops of 400–700 pips.
R = one unit of risk: −0.39R means each trade loses, on average, 39% of the amount you risked.
Doesn't a wider stop mean losing more money? Not if you size the lot to the SL. Money at risk = lot size × SL distance. Make the SL 10x wider and the lot 10x smaller.
Example: a $2,000 account risking 0.3% = $6 per trade.
- SL 30 pips → ~0.02 lot, but it gets stopped out easily.
- SL 500 pips → 0.0012 lot. A standard account can't go that small (0.01 lot already means $50 at risk).
- On a Cent account (denominated in cents, 100x smaller) it's 0.12 cent lot – exactly $6.
That's why accounts under $3,000 are better off on a Cent account: with a 400–700 pip SL, 0.01 lot usually loses $40–70 per losing trade, while on Cent it's only $0.4–0.7.
What to do:
- Place the SL by structure, clearly beyond the nearest swing high/low – not a fixed number of pips.
- Work out the lot size after the SL is set – try it now with the lot size calculator.
- When gold is swinging wide, widen the SL and shrink the lot. Don't bump up the lot to make it "worth it".
Trend · 3 min read
Catching tops and bottoms: what the data says about trading against the trend
"It's gone up too much, it has to turn soon" – that line has burned more gold accounts than any news release. Buying the bottom feels like getting a bargain, but the data shows it's one of the fastest ways to lose.
CashbackEA backtest · gold 2024–2026 · 3-pip cost per trade
- Picking tops/bottoms against the H4 trend (waiting for an indicator to print 3 peaks/troughs): 22% winners, −0.84R per trade in 2026 – the worst in the whole library.
- System 25 of the CashbackEA bot, trading with the trend: +0.36R per trade in 2026 (data not used when building the strategy) – the best of the 8 methods compared.
R = one unit of risk: +2R means a profit of twice the amount you were willing to lose.
The same method, run on gold from 01/2024 to 10/2026 (gold rose from ~$2,050 to over $4,000), split by trade direction:
| Direction | Trades | Total result |
|---|---|---|
| BUY (with the uptrend) | 613 | +11.0R |
| SELL (against it) | 453 | −75.4R |
Selling into a rising market, on its own, dragged the whole method from break-even into the red.
But don't chase price into a key level either: on gold 2015–2026, out of roughly 3,400 touches of a support/resistance zone, price pulled back 69% of the time before continuing.
Why is bottom-picking so tempting – and why does it lose?
- A "cheap" bottom can get a lot cheaper while the bigger trend is still running.
- After a loss, traders often flip direction right as price pulls back – and lose twice.
- Those perfect bottom-catching screenshots are usually picked after the move already happened.
What to do:
- Check the daily trend first: higher highs and higher lows means look for Buys only, and vice versa.
- Wait for the pullback; don't chase price at a key level.
- Only switch direction once the daily chart confirms a trend change – not because "price has gone too far". See today's bias in the gold outlook.
See the bot apply all this to live trades
Free Telegram channel: part of the bot's gold trades, losing trades included. No account needed.