Intraday, weekly and quarterly — one metal, three time scales, three different answers.
"Where is gold going" is not one question. It is three. Sixty seconds asks who is more urgent in the next minute. A week asks where real rates and the dollar are heading. A quarter asks who keeps buying, and at what price.
The inputs, failure conditions and payout structures differ completely. Trading a quarterly thesis on a 60-second expiry is the most common — and most expensive — retail mistake: the direction can be right while the time scale is guaranteed wrong.
Most losses are not wrong calls. They are the wrong ruler. The same candle is noise intraday and meaningless quarterly. Start by matching the question to the horizon:
| Framework | Question it answers | Valid holding period | Main inputs | Failure signal |
|---|---|---|---|---|
| Intraday | Order-flow imbalance over the next minute to an hour | 60s – 60min | Liquidity, event windows, key levels | Breakout on no volume; the moment data prints |
| Weekly | Direction over one week to three months | Days – weeks | Real rates, dollar, rate-cut odds, ETF flows | Real rates and gold moving together for weeks |
| Quarterly | Centre of gravity over one quarter to two years | Months – 2 years | Central-bank buying, deficits, easing cycle | Central banks turn sellers; ETF outflows persist |
Intraday price is almost entirely order flow. It has nothing to do with what gold is "worth" long term. So the intraday framework needs three variables: session, event, level.
| Session | Beijing time | Character | How to use it |
|---|---|---|---|
| Asia | 07:00–15:00 | Thinnest liquidity, range-bound, many false breaks | Fade the range edges only; no chasing mid-range |
| Europe | 15:00–20:30 | Volatility lifts, the day's trend starts to form | Wait for the first pullback to confirm |
| US open | 20:30–22:00 | Data plus equity open: best odds of a one-way move, worst noise | Let the first impulse finish, trade the retrace |
| US late | 22:00–02:00 | Trend continuation or give-back | Trade with the day's direction, never hold overnight |
NFP (first Friday, 20:30 or 21:30 depending on daylight saving), CPI and PCE (mostly 20:30), FOMC (decision 02:00, press conference 02:30). For roughly 10–15 minutes around these prints the tape jumps hardest, directional accuracy collapses toward a coin flip, and the payout does not compensate you for it. A 60-second trade into a data print is a bet on a tick you cannot forecast.
Previous day's high and low, the Asia range edges, round numbers every $10–20, the intraday VWAP. A 60-second trade is paid for the microscopic reaction at a level, not for a trend. The same call entered mid-range and entered at the edge are not the same trade — not even close.
Good intraday trading is few and specific: clear level, right session, no data. Not more trades — only the handful you can explain out loud.
From one week to three months the dominant variable shifts from order flow to opportunity cost. Gold pays no yield, so the real rate (nominal minus inflation expectations, usually read as the 10-year TIPS yield) is the price of holding it. Real rates up, gold usually under pressure. Real rates down, gold benefits.
One caveat: that negative link has clearly weakened since 2022, because the marginal buyer changed — from rate-sensitive ETFs to central banks that do not read charts. Real rates are still a major variable; they are no longer the only one.
| Variable | How to read it | For gold | Weight |
|---|---|---|---|
| 10-year TIPS real yield | Weekly trend, not single days | Inverse | High |
| Dollar index (DXY) | Where the weekly candle closes | Inverse | High |
| Rate-cut odds (rate futures) | The change over a week | Rising odds → bullish | High |
| Gold ETF holdings | Four consecutive weeks in one direction | Inflows → bullish | Medium |
| CFTC speculative net long | Whether it sits at a historical extreme | Extreme → contrarian | Medium |
Beyond a quarter, pricing power sits with price-insensitive buyers. World Gold Council data shows central banks net-buying at a historically high pace of roughly 1,000 tonnes a year since 2022 — about double the prior decade's average. These buyers do not read charts. They buy on strategic allocation, and they keep buying into weakness.
| Structural force | Direction | Why it matters |
|---|---|---|
| Central-bank net buying (~1,000 t/yr in recent years) | Bullish | Price-insensitive bid under the market |
| Fiscal deficits and debt monetisation | Bullish | Erodes fiat credibility, revives gold's monetary bid |
| Real rates entering an easing cycle | Bullish | Lower cost of holding |
| A structurally strong dollar | Bearish | Raises the cost for non-US buyers |
| Central banks slow down + ETF outflows for months | Turning bearish | The single most important regime signal to watch |
The classic quarterly mistake is treating a structural bull case as a short-term guarantee. Structure says deep dips get absorbed. It does not say this month cannot fall. Gold has produced multiple 10%–20% drawdowns inside structural bull markets — and intraday volatility expands during those, exactly when short-expiry trades get stopped out hardest.
One rule: holding period and evidence must share the same scale.
| What you are doing | Which framework | Never this one |
|---|---|---|
| 60-second trade | Intraday (session + level) | Weekly, quarterly |
| 5–60 minute trade | Intraday, with weekly for directional bias | Quarterly |
| Overnight to several weeks | Weekly | Intraday noise |
| Multi-month exposure | Quarterly | Intraday |
Downgrade to the shortest one. Weekly bullish, intraday bearish: either stand aside, or take the short and never hold it overnight. The reverse — "I'm long-term bullish so I'll buy the 60-second too" — collapses three frameworks into one, and a grand rationale never improves a payout.
XAUAD pays 68% at 60 seconds up to 90% at 60 minutes. Break-even win rate = 1 ÷ (1 + payout). That is arithmetic, not opinion:
| Expiry | Payout | Break-even | What it means |
|---|---|---|---|
| 60s | 68% | 59.5% | You need close to six in ten to break even |
| 5 min | 80% | 55.6% | — |
| 10 min | 82% | 54.9% | — |
| 15 min | 85% | 54.1% | — |
| 30 min | 88% | 53.2% | — |
| 60 min | 90% | 52.6% | Just over half is a positive expectancy |
Two things follow. First, the low payout on the shortest expiry is fair: inside one minute price is almost pure noise, and the payout has to match that noise share. Second, the same skill turns into positive expectancy more easily at longer expiries, because signal takes up more of the move.
Do not invert it, though. A longer expiry ties up capital longer while per-trade risk stays identical. Stretching a 60-second read out to 60 minutes for the 90% payout only puts you on a scale you cannot read.
| Mistake | Why it loses | Fix |
|---|---|---|
| Quarterly logic on a 60-second trade | Scale mismatch — right call, no capture | 60s reads session and level only |
| Entering on the data print | The jump is unforecastable; payout does not pay for it | Flat 10 minutes either side |
| Doubling after a loss | At a 68% payout, six losses in a row is ~0.4% — it will happen | Fixed 1%–2% risk per trade |
| Reading candles, ignoring sessions | Asia's false breaks stop you out repeatedly | Session first, level second |
| "Long-term bullish" read as "up today" | Structure gives no timing | Medium-term view sets bias only |
Last updated: October 2026 · Figures are order-of-magnitude descriptions from public sources and change over time; always check the latest release.
English
简体中文