When the tide of leverage receded, the gold price closed October at $3,913 per ounce, down 6.3% month-on-month, with gold prices denominated in all major currencies posting declines.
Our short-term gold price performance attribution model (GRAM) shows that falling US Treasury yields provided support for gold in October.
Yet that month gold and Treasury yields moved in the same direction in a rare occurrence that the model cannot explain, likely driven by three factors: market repricing of the Federal Reserve's interest rate path expectations; a contraction in total speculative positioning; and a temporary shift of funds away from equities and gold into bonds as stocks fell and credit spreads widened.
Most of the selling was likely led by the derivatives market.
In October, managed fund net long positions on the COMEX fell by about $7 billion (53 tonnes), after an even larger decline in September.
The drop in total open interest was even more pronounced, with non-reportable net long positions and net spread trading positions both declining in the final week of October, which is one reason the model produced negative residuals.
Despite the gold price decline, global gold ETFs still saw net inflows of about $3 billion (19 tonnes).
Europe led the way, followed by Asia, while North America saw a small net outflow.
The shift of funds into bonds, combined with North American gold ETF outflows and reduced COMEX futures open interest, together pushed the gold price lower.
As the US midterm elections concluded with unexpected results, the market repriced policies on US fiscal matters, AI regulation and other areas, and gold staged a rebound in early November.
But market expectations for the Fed's future rate path continue to shift with the release of economic data, still exerting a significant influence on gold's short-term moves.
At the same time, the actual implementation and impact of policies after the midterm elections remain highly uncertain, and we will continue to monitor their effect on gold.
Where is the gold price bottom? Looking back at the patterns
Spot gold is down about 25% from its January record high, and investors keep asking: is the gold price close to a bottom? For many commodities, the economics of primary supply can provide part of the answer, but that is not the case for gold.
Fair value estimates may offer a useful reference, but as with stocks, fair value is not an effective timing tool.
So instead of trying to pinpoint an exact bottom, we posed this question: historically, what conditions are usually in place when gold recovers during a downtrend? We focused on drawdown phases over the past 20 years, meaning periods when gold's 50-day moving average stayed below its 200-day moving average for at least one month, and tested a series of candidate indicators, three of which stood out, each reflecting a different feature of a selloff.
For each indicator, a signal is triggered when the reading falls in the most extreme 30% range of the drawdown months: price deviation, where the gold price is far below its 200-day moving average, gauging how oversold the market is; dollar strength, where the US Dollar Index (DXY) has risen sharply relative to the past 52 weeks, since when a gold decline is driven by a surging dollar the weakness is often temporary and reflects currency moves rather than weakening gold demand; and positioning washout, where speculative futures positioning (including managed funds and small traders) falls to unusually low levels relative to total open interest, because gold's recovery needs fuel, and when less committed speculators finish selling, fewer potential sellers remain.
The single indicator of the gold price sitting well below trend has some value for predicting subsequent performance, but it is imperfect.
When a selloff is accompanied by a surging dollar and a washout in speculative positioning, future returns improve markedly.
Based on historical cases, when all three conditions were met, the gold price trended higher three months later.
However, the sample size is small, with only five instances.
Still, when at least two of the three conditions are met, that can also provide some confidence in positive future returns.
You be the judge of the gold price bottom: analyzing indicator performance since 2005
Has the gold price bottomed? Currently, two of the three conditions are met: the gold price is far below its 200-day moving average, and the dollar has strengthened into a high range.
However, speculative positioning remains elevated.
The speculative bubble has not fully dissipated, and consistent with past drawdowns, speculative futures positioning still needs to be cleared further before gold can enjoy a more sustained recovery.
Two out of three met, a fair situation: gold price, drawdown phases and indicator levels.
Indicators are useful, but context matters more
This framework only assesses the outlook for the next three months and cannot predict the path over a year or whether a sustained uptrend can restart, which depends on gold's broader strategic allocation logic.
In that respect, the macroeconomic and fundamental backdrop may matter more.
Some investors may draw a parallel with the 2011 peak.
But we believe that comparison is inappropriate.
The multi-year selloff after 2011 occurred in an environment far less favorable for gold.
Based on the fundamental backdrop, the current situation is more likely a correction than the start of a long-term downtrend.
Even so, our simple framework shows the relevant conditions are not yet fully in place.
While many factors still favor gold over the medium term, short-term developments could still work against the gold price, especially given the event risk that fourth-quarter policy decisions and geopolitical factors may trigger.
For investors still on the sidelines, the signal to watch is a further washout in speculative positioning.
If that signal appears while the other two conditions still hold, then the current situation would align with historically attractive entry points for gold.
Other technical indicators send a similar message.
From a technical perspective, the gold price has fallen to what we typically regard as an extreme downside level, 10% below its 200-day moving average, but remains above support at $3,857 per ounce.
That support corresponds to the 38.2% Fibonacci retracement of gold's entire rally since 2015.
This suggests that, given the price level and technical state, investors can begin watching for signals of a potentially important bottom.
However, with trend momentum still weakening, it seems premature to call a bottom now.
That means the gold price could still test $3,857 per ounce, but that level is likely to prove an attractive entry point for investors.
Approaching an attractive level: gold price and the 10% envelope below the 40-week moving average.
Summary
This framework is not a trading rule but a simple judgment method, based on a small number of historical cases and a three-month observation window, to assess whether a selloff is beginning to show oversold signs.
The framework shows that the conditions for a gold rebound are gradually accumulating but are not yet fully in place.
The broader market environment suggests the current situation is more likely a correction than a repeat of the post-2011 downtrend, to which some investors tend to compare it.
For the remainder of 2026, event risk could still drive gold prices up or down, and the policy outcomes of the US midterm elections are likely to be a key factor.
If equity and credit markets weaken further, US Treasuries could become more attractive, once again producing the rare positive correlation between gold and yields we have seen in recent weeks.
But that would require inflation pressures to subside, and whether inflation can fall back remains unclear.
For strategic investors, gold has delivered annual returns of nearly 9% since 1971, and historically, buying in tranches on dips has been a reasonable approach.
So the more meaningful question is not where the gold price bottom is, but whether weakening gold prices are gradually becoming a more attractive opportunity to add to positions.
[1] Total AI trading exposure was cut sharply: Morgan Stanley found that beneath the index surface, the market is falling apart.
[2] Data as of October 30, 2026.
Our short-term gold price performance attribution model (GRAM) is a multiple regression model of monthly gold returns that includes four key drivers of gold performance: economic expansion, risk and uncertainty, opportunity cost, and trend momentum.
The related themes of these drivers reflect the motivations behind gold demand and, more importantly, behind investment demand.
These are considered the marginal drivers of gold price returns in the short term.
The residual contains other factors not yet captured by the model.
The model is estimated using a five-year rolling window.
Source: Bloomberg, World Gold Council.
[3] Methodology: We focus on drawdown phases because these indicator signals behave differently across trends.
In an uptrend, a gold price falling below trend, a stronger dollar, or reduced positioning often signals a pause in the rally rather than an investment opportunity.
Only after a downtrend is established do these same conditions mark panic selling and a subsequently stronger recovery.
We use monthly data since 2005, forward three-month returns, non-overlapping observations, and set the threshold at the most extreme 30% range of monthly readings; results remain robust when the threshold is adjusted to 20% and 40%.
[4] Methodology: We focus on drawdown phases because these indicator signals behave differently across trends.
In an uptrend, a gold price falling below trend, a stronger dollar, or reduced positioning often signals a pause in the rally rather than an investment opportunity.
Only after a downtrend is established do these same conditions mark panic selling and a subsequently stronger recovery.
We use monthly data since 2005, forward three-month returns, non-overlapping observations, and set the threshold at the most extreme 30% range of monthly readings; adjusting the threshold to 20% and 40%.
[5] Data range: December 27, 2005 to October 28, 2026.
Indicators are calculated based on daily data (CFTC positioning uses weekly data) and sampled at month-end between December 2005 and October 2026.
A drawdown phase is defined as the 50-day moving average remaining below the 200-day moving average for more than 20 consecutive trading days.
A signal is triggered when the indicator reading falls within the most extreme 30% range of the drawdown phase months.
Returns are forward three-month gold returns, with at least three months between signals.
Source: Bloomberg, World Gold Council.
[6] Data as of October 28, with positioning data as of October 27.
Source: Bloomberg, World Gold Council.
[7] Data as of October 28, 2026.
40 weeks is equivalent to 200 trading days.
Source: Bloomberg, World Gold Council.
Comments