The short answer
Buying the dip in gold has paid off over long horizons, but a modest dip has not beaten simply buying on a random day, and nearly every dip got deeper before it got better.
We took every COMEX gold futures close from 30 August 2000 to 4 November 2026 (6,571 trading days) and marked the first close that was 5%, 10%, 15%, 20% or 25% below the record close at the time. Then we measured what gold did over the next 1, 3, 6 and 12 months, and how much further it fell first. The headline numbers:
| Rule: buy the first close this far below the record | Signals | Median after 12 months | 12 months later, up | Median further fall first | Worst further fall |
|---|---|---|---|---|---|
| 5% below the record | 26 | +21.0% | 20 of 25 | −3.1% | −25.4% |
| 10% below the record | 8 | +7.6% | 6 of 7 | −7.8% | −20.2% |
| 15% below the record | 6 | +9.2% | 5 of 5 | −5.6% | −17.0% |
| 20% below the record | 5 | +15.8% | 3 of 4 | −5.7% | −20.4% |
| Any day (baseline) | 6,319 | +11.5% | 75% |
The baseline row is what you would have got by buying on any day at all. That is the number a dip rule has to beat, and the 10% rule did not. The rest of this page explains why, shows every 10% signal, and turns the findings into rules a trader can use.
How we tested it
- Data: COMEX gold futures daily closes (GC=F via Yahoo Finance), as stored in our gold price history data, 30 August 2000 to 4 November 2026. Futures prices sit a little above spot XAUUSD (see spot vs futures), which does not change percentage results much.
- The record: the highest close so far. Using closes, the 2026 record is $5,318.40 on 29 January 2026. (Intraday spot prices went higher, which is why you will see bigger record numbers elsewhere.)
- Entry rule: the first close at least X% below that record. After a signal, the rule waits until gold makes a new record close before it can fire again. So one fall gives one signal per level.
- Results: the percentage change from the signal close to the close 21, 63, 126 and 252 trading days later (about 1, 3, 6 and 12 months).
- Further fall first: the lowest close in the 12 months after the signal, compared with the entry. This is the pain a dip buyer would have sat through.
- No costs, no leverage, no stops. This is a study of the price, not a trading system. A leveraged trader with a stop would have had very different results, as the example below shows.
Small samples. Gold has only had a handful of big falls since 2000, so the 10%, 15% and 20% rows rest on 5 to 8 events each. One unusual year can swing a median. Read the numbers as “what has happened”, not as odds.
Every 10% dip since 2000
| First close 10% below record | Close | Record close | After 1 month | After 3 months | After 6 months | After 12 months | Further fall first | Days to a new record |
|---|---|---|---|---|---|---|---|---|
| 13 Mar 2003 | $335.90 | $379.00 | −2.4% | +5.2% | +13.0% | +21.1% | −4.3% | 124 |
| 7 May 2004 | $378.70 | $427.80 | +3.2% | +5.8% | +14.5% | +11.0% | −1.0% | 117 |
| 24 May 2006 | $637.00 | $719.80 | −8.2% | −2.7% | −1.4% | +2.8% | −11.8% | 332 |
| 1 Apr 2008 | $882.90 | $1,003.20 | −2.3% | +4.9% | +0.6% | +4.5% | −20.2% | 366 |
| 21 Dec 2009 | $1,095.40 | $1,217.40 | −0.6% | −0.6% | +12.7% | +26.7% | −3.9% | 96 |
| 23 Sep 2011 | $1,637.50 | $1,888.70 | +0.9% | −1.7% | +2.9% | +7.6% | −6.2% | 2,218 |
| 9 Nov 2020 | $1,854.40 | $2,069.40 | −0.9% | −0.6% | −1.7% | −1.3% | −9.5% | 771 |
| 30 Jan 2026 | $4,745.10 | $5,318.40 | +8.0% | −2.1% | −13.8% | not yet | −17.8% | not yet |
“Days to a new record” counts trading days from the signal until gold closed above the old record. Three things stand out.
1. The first month was usually bad. Five of the eight 10% signals were lower one month later. A 10% fall is often the start of a bigger move, not the end of one.
2. The pain came before the gain. The median further fall was 7.8%. In 2008 the dip buyer at $882.90 watched gold fall another 20.2% during the financial crisis before it recovered.
3. Recovery times ranged from months to almost a decade. In 2009 gold made a new record in about 96 trading days. After the 2011 peak it took about 2,218 trading days, close to nine years, until 2020. A dip that “always comes back” can come back long after a trader has given up.
Why the 10% rule did worse than buying on any day
It sounds strange: buying cheaper should beat buying at random. The reason is that a 10% fall usually means something has changed. In 2006, 2008, 2011 and 2020 the drop came with a shift in rates, the dollar or risk appetite, and that shift kept working against gold for months. Buying at any random day, on the other hand, mostly means buying during gold’s long uptrends from 2001 to 2011 and from 2019 to 2025, when time did the work.
The 5% rule looks great (median +21.0% after 12 months) for the same reason: most 5% dips happened inside strong bull markets and were quickly bought back. That is a sign of the trend, not of the dip itself. If the trend had turned, as in 2011 to 2015, the same rule would have caught falling prices again and again.
The 15% and 20% rules did better than the 10% rule, but with five or six events each, a single good year (2003, 2006 or 2022) moves the median a lot. They also include painful cases: the first close 20% below the record in April 2013 was still 12.1% lower a year later, and gold took about 1,830 trading days to reach a new record.
For a different angle on the same question, our study of what gold did after big monthly drops looks at every 6% monthly fall since 2000, and gold in November after a bad October looks at the season.
Where gold stands now: a 25% drawdown
Gold’s closing record is $5,318.40 (29 January 2026). On 30 October 2026 it closed at $3,898.90, 26.7% below the record, the lowest close of the year in our data. Since then it has bounced: our spot feed showed about $4,137 early on 5 November after one of its biggest daily gains since August.
Falls of 25% or more from a record are rare. Before 2026 there were only two first closes at that level since 2000:
| First close 25% below record | Close | 12 months later | Further fall first | Days to a new record |
|---|---|---|---|---|
| 11 Sep 2008 | $741.30 | +35.6% | −4.9% | 252 |
| 15 Apr 2013 | $1,360.60 | −2.5% | −12.2% | 1,829 |
| 30 Oct 2026 | $3,898.90 | not yet | so far 0% | not yet |
Two cases, two opposite outcomes: a fast recovery after the 2008 crisis and a seven-year wait after 2013. History does not tell us which one 2026 will look like. What it does tell us is that both are possible, and a plan must survive either. Our explainer on why gold fell from $5,318 to $4,000 covers what drove this fall, and why gold is down in 2026 tracks it as it happens.
A worked example: the 2026 dip on a $1,000 account
Here is what the “buy the first 10% dip” rule did on real 2026 prices for a leveraged trader. Contract maths: on XAUUSD, 0.01 lot is 1 ounce, so a $1 move is worth $1. The account is hypothetical; the prices are real COMEX closes.
Account: $1,000. Entry: buy 0.01 lot on 30 January 2026 at the first 10% dip close, $4,745.10.
- One month later gold was up 8.0%: the trade showed about +$380. It felt like the dip rule worked.
- By 30 October gold had closed at $3,898.90: the same trade showed about −$846, or 85% of the account.
- With 0.05 lot (5 ounces), the loss would have been about $4,231, more than four times the account. The platform would have closed it long before (a stop-out).
The same idea, with a stop. Suppose the trader set a stop $100 below entry ($4,645.10) and risked $100 (10%, already too much). On 0.01 lot, the loss would have been capped near −$100, and the account would have lived to trade the next setup.
The lesson is not “dips are bad”. It is that a dip with no exit plan is a bet on time, and leverage takes away your time. A long-term holder of unleveraged gold can wait nine years; a trader on margin cannot.
Buying the dip vs averaging down
These two get confused, and the difference matters:
| Buying a dip (planned) | Averaging down (unplanned) |
|---|---|
| One entry at a level you chose in advance | Extra entries because the first one is losing |
| Stop-loss set before entry | Stop moved or removed “to give it room” |
| Size worked out from the stop | Size grows as the price falls |
| Small, known loss if wrong | Loss grows exactly when the idea is failing |
Averaging down on gold is how many accounts end. Our averaging down entry and the simulation in gold martingale strategy risks show how quickly it compounds.
How traders can use dips without guessing the bottom
These are general, educational rules of thumb drawn from the data, not signals:
- Treat the dip as a zone to watch, not a reason to buy. A 10% fall tells you to pay attention. It does not tell you the fall is over.
- Wait for the price to show a change. Many traders wait for a higher low, a break of a recent swing high, or a close back above a short moving average. Our best EMA for XAUUSD and ADX indicator tests show how trend filters behaved on gold.
- Size from the stop, not from the story. Decide where the idea is wrong, then work out the lot size with the position size calculator.
- Expect wider days after a big fall. In the 20 COMEX sessions to 4 November 2026, gold’s average daily range was about $76, with days up to about $159. A stop sized for a calm market will be hit by noise.
- Plan for both histories. Ask: “If this is 2009, what do I do? If this is 2013, what do I do?” If the answer to the second is “hold and hope”, the size is too big.
- Know your recovery maths. A 50% loss needs a 100% gain to get back. The drawdown recovery calculator shows how fast it gets hard.
- Watch your head. Dip buying often starts with fear of missing the bounce. Our trading psychology hub has a tilt check and a pre-trade checklist for exactly this.
After a deep fall, gold can rally hard in a day and then give it all back. In the week of 2 November 2026 it rose from a one-year low to above $4,100 in four sessions, partly on politics and oil. Sharp bounces inside a downtrend are common. Keep risk per trade small and do not add to losing positions.
Key takeaways
- Since 2000, buying the first close 10% below gold’s record gave a median +7.6% after 12 months, below the +11.5% for buying on any day.
- Dips usually got deeper first: a median further fall of 7.8% after a 10% signal, and up to 20%.
- Recovery to a new record took anywhere from about 96 to about 2,218 trading days.
- Gold’s 2026 fall reached 26.7% below its record close on 30 October, one of only three 25% drawdowns since 2000, with one past case recovering in a year and the other taking about seven.
- For leveraged traders, the dip itself is not the plan. The stop, the size and the patience are.
Next step: practise spotting the turn on past charts in the Gold Lab chart replay, then test your own rules in the Strategy Lab before trading them.
Sources
- XAUUSD Global calculation from COMEX gold futures daily closes (GC=F via Yahoo Finance, stored in
gold-daily.json), 30 August 2000 to 4 November 2026; 6,571 closes. Signals, forward returns, further falls and recovery times computed for this article on 5 November 2026 - XAUUSD Global spot gold feed (
spot-close.json), 5 November 2026, 05:07 UTC - The account sizes, lot sizes and stop in the example are hypothetical
Educational content only, not financial advice. Past results do not predict future returns. Trading gold with leverage carries a high risk of loss.
Frequently asked questions
Is it a good idea to buy the dip in gold?
It depends on the dip and your time frame. In our test of COMEX gold closes since 2000, buying the first close 10% below a record gave a median 12-month return of 7.6%, below the 11.5% median for buying on any day. Buying deeper dips of 15-20% did better on average, but the samples are tiny and the worst cases took years to recover. This is history, not a forecast.
How far did gold usually fall after a 10% dip?
After the eight first closes 10% below a record since 2000, gold’s lowest close in the next 12 months was a median 7.8% below the entry, and in the worst case 20.2% below (April 2008). A dip buyer at 10% usually had to sit through a further fall before any gain.
How long did gold take to recover from big drawdowns?
It varied hugely. After the 2009 dip gold regained its record close in about 96 trading days. After the 2011 peak it took about 2,218 trading days, close to nine years, from the first 10% dip in September 2011 to a new record close in 2020.
How far is gold below its record now?
Gold’s record COMEX close is $5,318.40 on 29 January 2026. The low close so far was $3,898.90 on 30 October 2026, 26.7% below the record. That makes the 2026 fall one of only three times since 2000 that gold closed 25% or more below its high.
Is buying the dip the same as averaging down?
No. Buying a dip is one planned entry with a stop. Averaging down means adding to a losing trade as it falls, which grows the position exactly when the idea is failing. On leveraged XAUUSD that can blow an account, which is why we treat them differently.
Comments
No comments yet.