Short answer: after a month in which gold falls 6% or more, it has more often been higher three months later than lower. In COMEX gold futures data since August 2000, 12 of the 17 earlier cases were higher after three months, with a median gain of +4.3%. But when it went wrong it went badly wrong: after April 2013 gold was another 18.4% lower by December, and after March 2026 it was 13.1% lower three months later.
That matters now because gold has just had two of these months in a row. It fell 6.6% in September and 6.9% in October 2026, settling at $3,898.9 on 30 October. That is the first back-to-back pair of 6%+ monthly drops in our data, the worst October since 2008, and leaves gold about 30% below the 29 January record high of $5,586.2. Here is every comparable month, what followed, and how to use it without turning it into a prediction.
How we measured it
We used the daily COMEX gold futures settlement series behind our gold price history pages (Yahoo Finance GC=F, August 2000 to 30 October 2026). For each calendar month we took the last settlement and compared it with the previous month’s last settlement. Any month down 6% or more counts as a “big drop”. Then we measured the change from that month-end close to the close one, three, six and twelve months later.
These are futures settlements, so they differ slightly from the spot price you see on a trading platform. The live spot price is on our XAUUSD price page.
There are 19 such months since 2000. The two most recent (September and October 2026) do not have three months of follow-up yet, so most statistics below use the 17 completed cases.
Every 6%+ monthly drop in gold since 2000
| Month | Monthly change | Month-end close | +1 month | +3 months | +6 months | +12 months |
|---|---|---|---|---|---|---|
| Apr 2004 | −9.4% | $387.0 | +1.8% | +1.0% | +10.7% | +12.4% |
| Aug 2008 | −9.3% | $829.3 | +5.4% | −1.6% | +13.5% | +14.8% |
| Oct 2008 | −18.0% | $716.8 | +13.9% | +29.4% | +24.3% | +45.0% |
| Dec 2009 | −7.3% | $1,095.2 | −1.1% | +1.7% | +13.7% | +29.8% |
| Jan 2011 | −6.1% | $1,333.8 | +5.7% | +16.7% | +22.1% | +30.3% |
| Sep 2011 | −11.4% | $1,620.4 | +6.4% | −3.4% | +3.0% | +9.3% |
| Dec 2011 | −10.3% | $1,565.8 | +11.0% | +6.6% | +2.4% | +7.0% |
| May 2012 | −6.1% | $1,562.6 | +2.6% | +7.8% | +9.5% | −10.9% |
| Apr 2013 | −7.7% | $1,472.2 | −5.4% | −10.9% | −10.1% | −12.0% |
| Jun 2013 | −12.1% | $1,223.8 | +7.2% | +8.4% | −1.8% | +8.0% |
| Jul 2015 | −6.5% | $1,094.9 | +3.4% | +4.3% | +2.0% | +23.2% |
| Nov 2015 | −6.6% | $1,065.8 | −0.5% | +15.8% | +14.0% | +9.9% |
| Nov 2016 | −7.9% | $1,170.8 | −1.8% | +7.0% | +8.6% | +8.7% |
| Feb 2021 | −6.4% | $1,728.8 | −0.9% | +10.0% | +5.2% | +9.9% |
| Jun 2021 | −6.9% | $1,771.6 | +2.3% | −0.8% | +3.2% | +2.0% |
| Mar 2026 | −11.4% | $4,647.6 | −0.4% | −13.1% | −9.9% | n/a |
| Jun 2026 | −12.1% | $4,038.5 | +1.7% | +3.7% | n/a | n/a |
| Sep 2026 | −6.6% | $4,186.7 | −6.9% | n/a | n/a | n/a |
| Oct 2026 | −6.9% | $3,898.9 | n/a | n/a | n/a | n/a |
COMEX gold futures month-end settlements, our calculations from the site’s daily data. “n/a” = not enough time has passed.
The summary: a lean towards rebounds, not a guarantee
| Horizon | Cases | Higher | Median change after a big drop | Median change, all months |
|---|---|---|---|---|
| 1 month | 18 | 11 (61%) | +2.1% | +0.9% |
| 3 months | 17 | 12 (71%) | +4.3% | +3.1% |
| 6 months | 16 | 13 (81%) | +6.9% | +5.6% |
| 12 months | 15 | 13 (87%) | +9.9% | +11.1% |
Three things stand out.
1. The edge is modest. Gold rose in most three-month windows anyway, because it has been in a long uptrend since 2000: 62% of all three-month windows were positive. After big drops the figure was 71%. That is a lean, not a law, and with 17 cases a couple of different outcomes would change the percentages a lot.
2. The first month is a coin toss with a slight tilt. Only 11 of 18 next months were up, and September 2026 itself was followed by a further −6.9% in October. A big monthly drop does not reliably mark the low.
3. By twelve months the advantage disappears. The median one-year change after a big drop (+9.9%) was slightly below the median for all months (+11.1%). Whatever bounce tendency exists is a short-to-medium-term effect.
Why big drops have often been followed by bounces
There is no law that makes gold rebound, but there are plausible reasons the pattern shows up:
- Positioning gets cleaned out. A 6% month usually forces leveraged buyers to cut positions or get stopped out. Once that selling is done, there are fewer forced sellers left, so smaller amounts of buying can move the price.
- Volatility pulls in bargain hunters. Physical buyers, central banks and long-term investors who missed the rally often treat a large discount to recent prices as an opportunity, which can put a floor under the market for a while.
- Bad news gets priced quickly. A sharp month often reflects a sudden change in expectations, such as a more hawkish Fed. Once the market has adjusted, the next piece of news has to be even worse to push prices down at the same pace.
None of these works when the reason for the drop keeps getting stronger. That is the difference between the two templates below.
How much further did gold fall first?
A rebound three months later does not mean the low was in. We also measured the lowest settlement in the three months after each big drop:
- In 13 of 17 completed cases the lowest close of the next three months was less than 5% below the month-end close. The September 2011 case came within a whisker, at −5.0%.
- In three cases gold fell much further first: August 2008 (−15.0%), April 2013 (−17.7%) and March 2026 (−13.7%).
For a trader, that is the more useful number. Even in the “good” cases a further 3–4% dip was common. On gold near $3,900, 4% is about $156, which is roughly one of October’s biggest daily ranges. Our volatility tool shows how wide current daily ranges are.
After a big monthly drop, daily ranges are usually wider than normal. A stop sized for a $40-a-day market can be hit by noise in a $70–$150-a-day market. Size from the current range with the position size calculator and an ATR-based stop, and keep risk per trade small.
Two templates: October 2008 and April 2013
The averages hide two very different stories. Most big drops look like one of these.
Template 1: the forced-selling flush (October 2008)
October 2008 was gold’s worst month in our data, −18.0% to $716.8, during the global financial crisis. Investors sold what they could, including gold, to raise cash and meet margin calls. On 25 November 2008 the Federal Reserve announced it would buy up to $100 billion of debt issued by Fannie Mae, Freddie Mac and the Federal Home Loan Banks and up to $500 billion of mortgage-backed securities, the start of what became known as quantitative easing. Gold was +29.4% three months later and +45.0% a year later.
The features of a flush: the drop is driven by liquidation rather than a change in gold’s fundamentals, and a policy response then changes the backdrop. December 2011 and the 2015–2016 cases fit a milder version of this, and so does August 2008, though gold fell another 15% before the low.
Template 2: the start of a downtrend (April 2013)
In April 2013 gold fell 7.7%, including a two-day collapse on 12 and 15 April that Reuters described at the time as on course to be the biggest two-day fall since 1983. Reports that Cyprus might sell part of its gold reserves set off fears of wider central-bank selling, on top of outflows from gold ETFs. That was not a flush. The bull market that began in 2001 had already peaked in 2011, and the drop confirmed a new trend: gold was 10.9% lower three months later and 18.4% lower by the end of December, which was 24.6% below its end-March level. A further 12.1% drop followed in June 2013.
The features of a trend break: the large drop comes after a major top, not during an uptrend, and it is followed by more large drops rather than a quick recovery.
Which one does 2026 look like?
Honestly, it has features of both, which is why this page does not forecast. On the flush side: October ended with a $132.5 (3.3%) fall on its last day, settling near the day’s low, which looks like forced selling into a weekend. And October 2008 also came only two months after another big drop (August 2008), yet it was followed by the strongest rebound in the table. On the trend side: like 2013, 2026’s drops come after a record high (29 January), and 2026 now has four 6%+ monthly falls in eight months (March, June, September, October). Gold has not done that before in our data. March 2026, the first of them, was followed by more losses. Our explainers on why gold is down in 2026 and the 2026 drawdown leg by leg cover the drivers.
Back-to-back big drops: a new situation
September and October 2026 are the first consecutive months both down more than 6% since our data begins in August 2000. Loosening the filter to two consecutive months each down 3% or more gives nine earlier pairs:
| Pair of months | Changes | Next 3 months | Next 12 months |
|---|---|---|---|
| Jun–Jul 2002 | −4.0%, −3.3% | +4.9% | +16.8% |
| Feb–Mar 2003 | −4.9%, −4.1% | +3.0% | +27.2% |
| Dec 2004–Jan 2005 | −3.1%, −3.6% | +3.1% | +35.3% |
| Mar–Apr 2008 | −5.8%, −5.8% | +5.9% | +3.2% |
| Apr–May 2013 | −7.7%, −5.4% | +0.3% | −10.6% |
| May–Jun 2013 | −5.4%, −12.1% | +8.4% | +8.0% |
| Nov–Dec 2013 | −5.5%, −3.9% | +6.8% | −1.5% |
| Sep–Oct 2014 | −5.9%, −3.3% | +9.2% | −2.5% |
| Oct–Nov 2016 | −3.2%, −7.9% | +7.0% | +8.7% |
Measured from the close of the second month. The two 2013 rows overlap.
All nine were higher three months later, though the April–May 2013 case only just (+0.3%) and only after a 12.1% fall in June. Over a year the record was mixed: three of the nine were lower, all of them in the 2013–2015 bear market. Again: small sample, wide spread.
What this means for your trading
Historical base rates are useful for planning, not for picking a direction.
- Plan both paths. A 71% three-month rebound rate means roughly three cases in ten did not rebound. Write down in advance what a rebound would look like on your chart (for example, a reclaim of a broken level) and what continued weakness would look like (new lows on rising ranges). Our XAUUSD technical analysis page and the weekly gold forecast give current levels.
- Expect a further dip even in the good cases. In most rebounds gold still traded a few percent lower first. Do not size as if the month-end close was the low.
- Respect the calendar. The week ahead has the US jobs report on 6 November and the US midterm elections; see our week-ahead guide for times. Event days widen ranges further.
- Check the seasonal context, but lightly. In our data November has been up 15 times and down 11 times since 2000, a weak tendency. The monthly returns heatmap shows every month at a glance.
- Keep a journal of how you handle these months. Large-drop months are where traders make their biggest mistakes, usually by averaging down or by sizing up to “win it back”. Our October 2026 review lists the lessons from the month just gone.
If you want to work through months like this with other traders, the XAUUSD Global community reviews the market every week. The first month is free, then $19.99/month.
Key takeaways
- Since 2000, 12 of 17 completed 6%+ monthly drops in gold were higher three months later (median +4.3%).
- The edge over a normal month is small, and it fades by twelve months.
- Even in rebounds, gold usually dipped a few percent further first; in three cases it fell 13–18% further.
- October 2008 (flush, then +29.4% in three months) and April 2013 (trend break, then −18.4% by December) are the two templates.
- September–October 2026 is the first back-to-back 6%+ pair since 2000, and 2026 already has four 6%+ drops, so history gives context, not an answer.
Sources
- Gold prices: COMEX gold futures (GC=F) daily settlements via Yahoo Finance, August 2000 to 30 October 2026, as stored in the site’s daily data file; all monthly and forward returns are our calculations
- NPR: Fed to give credit markets $800 billion jump start (25 November 2008)
- Reuters via Business Standard: Gold slumps to 2-year low below $1,400 per ounce (15 April 2013)
- Reuters via Business Standard: Gold losses seen limited on Cyprus bullion sale plan (11 April 2013)
- Our related studies: Gold in October 2026, best month for gold and gold data studies
Educational content only, not financial advice. Trading gold with leverage carries a high risk of loss.
Frequently asked questions
What usually happens to gold after a 6% monthly drop?
In COMEX gold data since August 2000, 12 of the 17 earlier months with a fall of 6% or more were followed by a higher price three months later. The median three-month change was +4.3%, against +3.1% for all months. The sample is small and the losses in the other five cases were large, so it is a tendency, not a rule.
Has gold ever fallen more than 6% two months in a row before?
Not in our data, which starts in August 2000. September 2026 (−6.6%) and October 2026 (−6.9%) are the first back-to-back monthly falls of more than 6%. The nearest earlier case is May and June 2013 (−5.4% then −12.1%), which came straight after April 2013’s −7.7%.
What happened to gold after October 2008?
October 2008 was gold’s worst month since 2000, down 18.0% to $716.8. Three months later, at the end of January 2009, it was $927.3, up 29.4%, after the Fed announced large asset purchases in November 2008. A year later it was up 45.0%.
When did a big monthly drop in gold lead to more losses?
April 2013 is the clearest case. Gold lost 7.7% that month and was another 10.9% lower three months later and 18.4% lower by the end of December 2013, or 24.6% below the end-March level. March 2026 (−11.4%) was also followed by a 13.1% further fall over the next three months.
Does this mean gold will rise in November 2026?
No. The study describes what happened after earlier large drops; it does not forecast this one. Two of 2026’s own large drops were followed by more selling. Use the numbers to plan for both a rebound and further weakness, and size positions for the wider ranges that follow big months.