Gold short-term trading tactics:a complete playbook for 60 seconds to 5 minutes

Gold scalping and intraday momentum look simple: read the chart, click buy, take a few points, close. Anyone who has actually done it knows the hard part is not being right on direction — it is finding a process you can repeat across cost, volatility and emotion.

No mystical patterns here, only the five things that actually decide P&L on the 60-second-to-5-minute horizon: when to trade, where to trade, how to confirm it is not a fake breakout, how big to trade, and when you must stop. Every price in this article comes from real quotes on the XAUAD market API. The same snapshot is visible on today's gold price page.

Snapshot 2026-10-23 16:36 (Beijing time, London session; this page refreshes every 20 seconds while open)

1. Start with cost: short-term trading earns volatility minus the spread

On a 60-second horizon the thing most easily overlooked is not direction but the fixed cost of every entry and exit. The XAUUSD spread is quoted in dollars per ounce and one lot is 100 ounces, so the spread must be multiplied by 100 to get your real cost. To see what a given trade actually pays, drop the quotes into the spread and P&L calculator.

The arithmetic here is brutal: a 0.662 spread is 40% of the average 1-minute range (USD 1.66). In other words, if you take a 1-minute trade targeting only USD 1, the cost eats nearly half your profit. To survive at 60 seconds your target range must be at least 2–3x the spread, or you must trade only in high-liquidity sessions. If you think in grams, convert with the gram / ounce converter to compare CNY per gram.

2. Sessions: save your bullets for 21:00-24:00 Beijing time

Gold trades 24 hours, but volatility is not evenly spread. The three hours when London and New York overlap (21:00-24:00 Beijing time in summer time; one hour later in winter) offer the day's best liquidity, tightest spread and the highest chance of a trending move. The session clock shows in real time which market is open right now.

The comparison is direct: between 21:00 and 24:00 Beijing time on 20 October 2026, 5-minute candles averaged USD 5.97, median 6.12, maximum 10.44, with a three-hour range of 4113.49-4155.55 — 42.06 in total; the same day's Asian and European hours (11:00-21:00 Beijing) averaged only USD 3.75. You can reproduce this day-by-day comparison in the historical gold price lookup using the same data source. Same 0.662 spread, 59% more volatility in the overlap, and cost as a share of range falls from 17.5% to 11% — that is the entire arithmetic case for "only trade these three hours". For contrast on the following day: on 21 October the Asian session (07:00-13:55) had 84 five-minute candles averaging just USD 3.50 (median 3.06, max 8.95), only 59% of the previous day's overlap. Later that same day the European open (15:00-17:45) was quieter still — 34 five-minute candles averaging USD 2.94 (median 2.77) with a range of only 14.17 — the "opening impulse" in the first half hour of Europe is often fake; do not treat it as a trend. Asia is not untradeable, but you cannot demand overlap-session targets from it. Now the most recent full overlap (21:00-24:00 on 21 October 2026): 36 five-minute candles averaged USD 4.21 (median 3.85, max 8.04) over a 4121.64-4141.38 range of 19.74; the same day's European session (15:00-21:00, 72 candles) averaged only USD 3.22 and Asia (07:00-14:00, 84 candles) 3.50. The overlap was still the most active, but 30% narrower than the day before — session edge is a probability, not a guarantee.

Asia on 22 October, live (data through 11:50 Beijing): today's 00:00-11:50 produced 143 five-minute candles averaging USD 3.30 with a median of 2.69, of which 83 were under USD 3 and 42 under USD 2 — yet the range slid from 4142.23 all the way to 4090.99, 51.24 in total. Volatility clearly came in two blocks: the 84 candles between 00:00 and 07:00 averaged only USD 2.75 (median 2.19), with the three biggest bunched at 02:45-03:00 (10.78 / 14.15 / 10.52), dumping from 4140.19 to 4112.53 in half an hour, −27.66; while after 09:00 it was a different market entirely — 35 candles from 09:00 to 11:50 averaged USD 4.86 (median 4.39) with only 5 under USD 3, one candle at 11:20 printing 12.09 and breaking 4100 down to 4097.86 before grinding to 4090.99. Both are "Asia", yet the morning block carried 1.8x the overnight volatility: size by the ATR the day has actually printed, not by the label "Asia = quiet". Subtract today's measured USD 0.658 spread and the overnight median range leaves only 1.53 to earn, against 3.73 in the morning block — the same 60-second playbook is two different businesses at 2am and at 10am. This section is updated every trading day on the same basis as historical gold prices, so you can check it yourself.

Late Asia and the European open on 22 October (data through 15:15 Beijing): after 11:50 the second half of the Asian session delivered the day's most extreme stretch — 13:45 dumped to 4066.885 (the intraday low at the time, 75.35 below 4142.23 at 04:45), and the very next candle at 13:55 opened 4073.725, closed 4090.225 for a range of 17.53, swallowing the entire bodies of the three candles before it. The bounce to 4102.36 at 14:15 went nowhere, and the 12 candles from 14:00 to 15:00 averaged USD 5.11 (median 5.02), 2.3x the same measure overnight (median 2.19) — late-Asia volatility was already approaching overlap levels, and by then "Asia should be light size" is advice you needed three hours earlier.

The first three candles of the 15:00 European open played out the whole "opening impulse" script: the first had a 3.67 range (closing 4091.885, essentially flat), the second 10.79 (4092.155 → high 4100.755 → close 4098.105), 16.3x the current 0.662 spread; the third pushed to 4100.48 but closed only at 4098.995 — two attacks on 4100 and neither held. Anyone who entered long on the "4100 breakout" then had to sit through a full retracement: the 19:20 candle plunged from 4071.875 to 4058.83 (range 13.05), 19:55 probed the day's low at 4057.58 (and 22:05 went further to 4044.915), a total of −43.18 from 4100.755. This is the live version of the line repeated throughout this article: the first move after an open is a liquidity gap, not a trend.

The 21:00 overlap, live (data through 23:50 Beijing): a late-European bounce had carried price back to around 4087 (the 20:45 candle ran 10.39, 20:55 closed 4087.98), and the first overlap candle answered immediately — open 4087.37, high 4087.82, then straight down to 4078.115 closing 4080.57 for a 9.71 range; 21:30 ran 10.51, 21:35 printed 11.72 (4072.77 → 4061.045), and 21:45's 11.93 (4067.265 → 4055.335) was the session's largest, with 22:05 taking the intraday low down to 4044.915 before 22:40 bounced back to 4059.695 — a maximum 42.91 decline from 4087.82 inside 170 minutes, closing 23:50 at 4054.68. Those 35 candles averaged USD 6.327 (median 5.605) — 1.28x the same day's European session 15:00-21:00 (4.94), 2.3x the overnight block 00:00-07:00 (2.75), and 50% above the whole 10-21 overlap (4.21). Same "overlap" label, half again as volatile as yesterday: size down to today's ATR, not up to yesterday's memory. A single 11.93 candle is 18.1x the 0.658 spread, and that is the reason this window is worth waiting for — not because it must rise or fall, but because the range on offer covers the cost. By the 23:50 close the day's 287 five-minute candles averaged USD 4.24 (median 3.46) across a 4142.23-4044.915 range totalling 97.31; the five largest candles (17.53 / 14.15 / 13.05 / 12.09 / 11.93) printed at 13:55, 02:55, 19:20, 11:20 and 21:45 — four of the five inside an "open or data window", only one in the Asian afternoon. Source: XAUAD market API / Swissquote, same basis as historical gold prices, verifiable yourself.

Full day on 23 October (data through 22:25 Beijing): 270 five-minute candles from 00:00 to 22:25 averaged USD 3.90 with a median of 3.47, of which 95 were under USD 3 and 38 under USD 2; open 4052.735, a dump to 3986.855 at 10:00, then a fresh intraday low of 3982.15 at 22:25 with the last print at 3983.925 — a 4053.165-3982.15 range of 71.02. Split into four blocks: 84 candles from 00:00 to 07:00 averaged USD 2.70 (median 2.68), 84 from 07:00 to 14:00 averaged 4.67 (median 3.93), 84 from 14:00 to 21:00 came in at 3.75 (median 3.47), and the 18 after 21:00 averaged 6.64 (median 5.71) — the overlap ran 1.7x the daily average and 2.5x the overnight block, essentially level with the same window the day before (6.33), which says this is not an accident but the norm for this window. The 1-minute chart is no kinder: 400 valid candles between 15:49 and 22:28 averaged USD 1.90 (median 1.73), and the 0.662 spread eats 35% of that.

The stretch most worth repeating ran from 09:55 to 10:05: three consecutive five-minute candles printed ranges of 10.25 / 9.82 / 17.21. The 09:55 candle plunged from 4001.095 to 3991.465, 10:00 broke the intraday low at 3986.855 (sweeping both the 3990 round number and the prior low), and then 10:05 opened 3994.755, ran to 4011.795 and closed 4010.135 — down 14.24 then up 24.94 in fifteen minutes, one candle swallowing the whole decline of the previous two. Anyone short at 3986 with a stop above 3995 was taken out by that 17.21 candle (26x the 0.662 spread); anyone who waited above 3990 for the "sweep and reclaim" got the long body in their direction. By 22:25 the day's three biggest candles were 17.21 (10:05), 15.08 (22:15) and 14.40 (22:25) — none of them inside the ten hours from 11:00 to 21:00 (the largest there was 12.56 at 13:30). Volatility is not evenly distributed; your job is to wait for it to be delivered, not to nibble with small size through the ten hours in between.

The 21:00 overlap, live (data through 22:25 Beijing): the first candle opened 4026.255 and the first twenty minutes were a textbook "three attacks, none held" — 21:10 pushed to 4032.675, 21:15 touched 4033.13, 21:20 made a marginal new high of 4033.41 and closed 4030.965, the close straight back below the prior high; the next candle at 21:25 delivered a 6.91 down candle back to 4025.675 — that is exactly signal one from section 5 (close back on the other side of the breakout level) plus "no follow-through on the next candle", decidable within two candles with no moving averages required. A 60-minute staircase followed: 21:45 probed 4016.265, 22:00 broke 4017.215, 22:10 reached 4007.875, and then it accelerated — 22:15 a single 15.08 candle (4007.955 → 3994.035), 22:25 a single 14.40 candle (3996.555 → 3982.15), rewriting the intraday low from 3986.855 to 3982.15. From 4033.41 that is 51.26 of decline in 65 minutes; the 18 candles averaged 6.64 (median 5.71), 1.7x the day's 3.90 average and 2.5x the overnight 00:00-07:00 block (2.70); over the same span the 1-minute chart's 89 candles averaged USD 2.86 (median 2.61), 1.8x the 14:00-21:00 figure (1.62). A single 15.08 candle is 22.8x the 0.662 spread — the only thing worth waiting for in the overlap is not direction but a range large enough to cover cost. Look at the other side: the 4033.41 high at 21:20 was only 0.28 above the prior high, less than one spread; anyone chasing long there with a stop below 4029 lost both stop and position to that 15-dollar candle at 22:15, while anyone who waited below 4032 for a reclaim-and-confirm to go the other way collected a slice of the 51. Source: XAUAD market API / Swissquote, same basis as historical gold prices, verifiable yourself.

One more detail you must know: for 70 minutes from 04:55 to 06:04, the open, high, low and close of every 1-minute candle equalled 4024.03 — not a single move. This is not a frozen market but a quote freeze inside the platform's daily maintenance window (20:00-24:00 UTC, i.e. 04:00-08:00 Beijing) — candles in that period are drawn, not traded. Trade short timeframes inside a freeze and every support and resistance you drew, every "breakout" you saw and every moving average you computed is void; worse, the first minute after quotes resume often gaps straight to the other side of your stop. So today adds one more hard rule to section 2: no new positions between 04:00 and 08:00 Beijing, and any overnight position must be dealt with before 04:45. Source: XAUAD market API / Swissquote, same basis as historical gold prices, verifiable yourself.

3. Chart setup: 5 minutes for structure, 1 minute for execution

Most people who struggle with 60-second trades do not have a technique problem, they have the wrong division of labour between charts — staring at a 1-minute chart for direction is looking for a trend inside noise.

• 5-minute chart (structure): mark the previous day's high and low, the Asian range high and low, and the slope of EMA21. This decides whether you are long or short today.
• 1-minute chart (execution): used only to find entries, place stops and check whether a key level was reclaimed. No direction calls here.
• ATR(14): the 5-minute ATR is your stop ruler. Stop = 1x ATR — no gut feel.
• Never stack more than three indicators: EMA9, EMA21 and ATR are enough. The more indicators, the more hesitation.

Choosing your gold minute chart: three timeframes, three jobs

The same market is a different market on a different timeframe. Below are three real statistics from 22 October (XAUAD market API, spread USD 0.662), each with the cost of using that timeframe for the wrong job:

• 15-minute (direction): sets PDH/PDL, the day's range and the EMA21 slope — 89 candles today averaging USD 7.48 (median 6.53). Use it to pick entries and your stop is forced out beyond 7 dollars.
• 5-minute (primary): support and resistance, ATR, fake-breakout judgement — every case in this article uses this basis. Today 267 candles averaging USD 4.16 (median 3.36, max 17.53). Look only at the 5-minute and never drop to the 1-minute, and your entries will typically be 1–2 dollars off.
• 1-minute (execution): entries and stops only, no direction — today 1,000 candles (05:33-22:12) averaging USD 1.82 (median 1.51), of which 94 had a full range smaller than the 0.662 spread.

Divide each of those three by the spread and you get a more useful metric than range itself: range ÷ spread. The 15-minute is 11.3x, the 5-minute 6.3x, the 1-minute only 2.7x. Break the 1-minute down by session and the gap is starker: the 74 one-minute candles from 21:00 to 22:12 today averaged USD 3.16 (4.8x the spread), the 360 Asian candles from 09:00 to 15:00 averaged 1.91 (2.9x), the 360 European candles from 15:00 to 21:00 1.97 (3.0x), and the 86 candles from 00:00 to 07:00 only USD 0.53 — less than one spread. That is the physical boundary of gold intraday scalping: on that overnight 1-minute chart, long or short, you are theoretically losing on every single trade. Never take a 60-second trade in a window where the multiple is under 3 — and that is the entire reason the 60-second playbook has to crowd into 21:00-24:00: not because it "rises more easily" there, but because only those candles can pay the cost.

Those multiples determine the outcome far more than any entry pattern — the first question in gold short-term trading is never "what signal do I enter on" but "which session am I entering in". While we are here, let us pin down the term gold scalping: it means holding 30 seconds to 3 minutes, many trades a day, accumulating small ranges — a different business from intraday swing trading. The account requirements come first: the spread must be tight (mainstream XAUUSD is USD 0.2–0.7, 0.662 measured today), slippage small (under USD 0.1), and size tiny (0.01 lots = 1 ounce, where a 0.01 move is USD 0.01 of P&L and one full round trip costs about USD 0.66 in spread). If your account's spread is above USD 1, the expectancy of a 60-second trade is negative by construction — change accounts first, then talk technique. That is arithmetic, not mindset. Work out lots and point value on the margin and point value calculator before you order.

4. How to draw support and resistance (4 steps, only 3 lines)

Ninety percent of short-term entries come from "the second reaction of price at a level". Draw the level wrong and everything downstream is wrong.

1. Mark the previous day's high and low (PDH / PDL): the two most important lines of the day, where institutional stops pile up.
2. Mark the Asian range high and low (07:00-14:00 Beijing): the box edges most often tested at the European open.
3. Find levels touched three times or more: the more touches the better. Note you are drawing zones, not lines — on a 5-minute gold chart a "line" is usually 0.5–1 dollar thick.
4. Keep only three: one above, one below, one in the middle. More than three and you are hunting for a reason to trade.

5. Spotting fake breakouts: three signals plus eight real cases

The most common pattern in gold during the overlap is: sweep the stops on one side first, then go the other way. So "chase every breakout" is the fastest way to lose money on a 5-minute chart. Three signals identify a fake breakout:

1. The close returns to the other side of the breakout level: a wick through it is not a breakout — the close failing to hold is the point. Reclaimed within 1–2 candles of the break and it is a fake.
2. Small body, long wick: when the breakout candle's body is smaller than its upper wick, the push was sold straight back — no acceptance.
3. Against the slope of the 5-minute EMA21: chasing a breakout against the medium-term average slope cuts your success rate significantly.

6. Three playbooks: 60 seconds / 3 minutes / 5 minutes

The same session behaves completely differently depending on holding period. Do not mix them.

Beginners should start with the third. The reversal playbook (fake breakout) has a clean stop, a defined target and needs no direction call, which makes it the most quantifiable of the three. 60-second scalping looks fastest but demands the most in execution speed and emotional control — practise it last.

7. Position formula: risk no more than 5% of capital per trade

This is the most important section in the article. Get the size wrong and every technique above is meaningless. Run the formula in the position size calculator rather than by hand. The XAUUSD contract is 1 lot = 100 ounces, so every 1 dollar move is 100 dollars of P&L per lot.

Example: a 10,000 account with a 5% per-trade cap (= 500) and a 4-dollar stop on the 5-minute chart:

Quick reference: how many lots at 2% risk

If the formula will not stick, one glance at this table before the session is enough. It is computed at 2% risk per trade (the recommended live figure, not 5%), with stop distance down the side and balance across the top, rounded down to 0.01 lots. Remember one point value first: 0.01 lots = 1 ounce, so a 1-dollar move is 1 dollar of P&L on 0.01 lots.

How to use it: measure your stop distance first, then read across from your balance. If it comes out below 0.01 lots, that is not a rounding problem — the trade should not be taken at all: either the stop is too wide or the balance cannot support the playbook.

Now bring cost into the same picture. At a 0.66 spread, one round trip costs lots × 100 × 0.66: about 33 dollars on 0.5 lots, 66 on 1 lot, 132 on 2. Compare: a 5,000 account with a 3-dollar stop trading 0.33 lots risks 100 and pays 22 in round-trip cost — 22% of risk, acceptable; the same account squeezing the stop to 1 dollar must trade 1 lot, paying 66, which eats 66% of the risk and lets the spread swallow the expectancy. Rule of thumb: if round-trip spread cost exceeds 25% of your per-trade risk, the trade is not worth taking.

8. Stop-trading rules after consecutive losses

The biggest enemy in short-term trading is not the market, it is the urge to win it all back immediately after a loss. Write the rules before you start, because you will not be rational while losing.

1. Two losses in a row halve your size, and check whether you traded outside your plan.
2. Three losses in a row stop for 30 minutes, leave the screen, look at no charts.
3. A 5% drawdown on the day done for the day, no new positions of any kind.
4. An 8% gain on the day also stop, lock it in and avoid giving it back.
5. Three losing days in a week take a day off, review only, no orders.
6. Two stop violations a week on demo, back to simulated trading.

9. Review: six fields per trade

Short-term trading is built on samples, not inspiration. The simpler the journal, the more likely you keep it. Six items per trade are enough:

After 30 trades you will see something very concrete: your losses are most likely concentrated in one or two sessions and one playbook. Cut those out and what remains is your system. Every settlement quote and result can be checked line by line in settlement disclosures.

10. Three of the most common mistakes

• Mistake one: believing more trades means more money. The spread is a fixed cost, so ten trades cost ten times one trade, and your win rate does not improve with volume.
• Mistake two: using a 15-minute chart or higher to trade 60 seconds. The structure chart and the execution chart must match, or your stop placement is never sensible.
• Mistake three: increasing size after a loss. It is the fastest route to a zero balance, without exception.

After opening an account, deposits are USDT only; for your first transfer see the USDT deposit guide so you do not send on the wrong chain.

11. Frequently asked questions

Q. How many gold short-term trades a day is reasonable?
A. No more than 3 for a beginner, no more than 5 once experienced. At a 0.66 spread a round trip costs 66 per lot, so ten trades is 662. If there is no clean level, there is no trade.

Q. For 60-second scalping, should I watch the 1-minute or the 5-minute?
A. Both: the 5-minute sets direction and key levels, the 1-minute only finds the entry and the stop. Watching the 1-minute alone turns noise into a trend.

Q. What is the best Beijing time for gold scalping?
A. The 21:00-24:00 London-New York overlap. Measured five-minute ranges in that window averaged USD 5.97, about 59% above the same day's Asian and European 3.75.

Q. How wide should a gold short-term stop be?
A. By volatility: 1x ATR(14), or one dollar beyond the structural high or low. On a 5-minute chart that is usually between 1.5 and 4 dollars.

Q. Can I add to a loser to average down?
A. No. It pushes per-trade risk through the 5% ceiling and doubles the spread cost. Two losses in a row halves size, three stops you for 30 minutes, a 5% daily loss ends the day.