Gold Price Movement 30-90 Days After RSI Overbought Signals (79 characters)

1. Data & Confidence Context

The signal engine analyzed 83 confirmed RSI-above-70 crossings in GC=F (gold futures) spanning January 2001 through early August 2026 — a 25-year window that encompasses multiple full commodity supercycles, two financial crises, a pandemic, and a sustained post-2020 inflationary episode. With n=83, this is a statistically meaningful sample by event-study standards, though the macro regimes embedded within it are heterogeneous enough that aggregate statistics mask substantial cycle-to-cycle variance. The p-values across all three forward windows (0.48 at 30 days, 0.52 at 60 days, 0.29 at 90 days) are uniformly above conventional significance thresholds, which is itself the central finding: this report treats the signal as a framing device for understanding gold's momentum behavior, not as a reliable directional edge.

---

2. Direct Answer — What the Data Shows

The engine's verdict is unambiguous in its ambiguity. Across 83 RSI crossings above 70 in gold futures from 2001 to 2026, the mean 30-day forward return was +0.43% (median +0.18%, win rate 50.6%). At 60 days, the mean nudged to +0.50% (median +0.34%, win rate 50.6%). At 90 days, the mean improved to +0.92%, but the median *fell to -0.58%* and the win rate dropped to 48.2%. Against an unconditional base rate of roughly +0.96% per 30-day period, the RSI signal generated excess returns of -0.54%, -1.43%, and -2.05% at 30, 60, and 90 days respectively — meaning that on average, buying gold when RSI crosses 70 has slightly *underperformed* simply holding gold at random. None of these excess returns are statistically distinguishable from zero.

The earliest signals in the dataset, beginning August 17, 2001, arrived during gold's first stirrings out of a two-decade bear market. The September 14 and September 21, 2001 crossings came in the immediate aftermath of the World Trade Center attacks, when gold spiked on safe-haven demand — a classic case where the RSI signal was a symptom of a macro shock rather than a technical setup with predictive content. The January and February 2002 crossings followed as gold consolidated those gains, and the April and May 2002 signals captured the early phase of the commodity supercycle as the dollar began its multi-year decline. In each of these early cases, the 30-day outcome was highly path-dependent: signals that fired into genuine macro tailwinds (dollar weakness, geopolitical fear) tended to produce positive follow-through; signals that fired into exhaustion rallies tended to mean-revert.

The dispersion in outcomes is the most important number in this dataset. At 30 days, the best single outcome was +12.2% and the worst was -17.34% — a 29.5 percentage-point range around a mean of less than half a percent. At 60 days, that range widens to +16.17% to -19.9%. At 90 days, the best was +21.71% and the worst -14.2%. This is not a signal that clusters outcomes — it is a signal that fires across wildly different macro environments and produces wildly different results accordingly. The median diverging below zero at 90 days while the mean stays positive tells a specific story: a minority of very large positive outcomes (likely the 2007-2008 crisis rally, the 2011 peak approach, and the 2020 pandemic surge) are pulling the mean up, while the majority of 90-day outcomes are flat to slightly negative.

The chart covering 2001 through August 2026 shows gold's full arc from sub-$300 levels in the early 2000s through the secular bull market, the 2011 peak, the 2013-2018 consolidation, and the post-2019 breakout into what by August 2026 represents a substantially higher price regime. RSI crossings above 70 have occurred throughout this entire arc — in uptrends, in topping processes, and in sharp counter-trend rallies within downtrends — which is precisely why the aggregate statistics are so diffuse.

---

3. Confounding Factors — Decomposing What Actually Drove Each Cycle

The reason the RSI signal produces near-zero excess returns is not that it is random noise — it is that it fires in response to fundamentally different causal forces, and those forces have opposite implications for what happens next.

In the 2001-2002 cluster of signals, the dominant force in the first three months was dollar weakness: the trade-weighted dollar entered a multi-year decline that provided a structural tailwind for gold regardless of momentum readings. RSI crossing 70 in this environment was a confirmation of trend, not a warning of exhaustion, and follow-through was positive. By contrast, the December 2002 and January 2003 signals arrived as gold approached levels not seen in years, and the geopolitical premium from the buildup to the Iraq War was doing heavy lifting. When that premium partially unwound after the initial invasion in March 2003, signals that had fired in January saw negative 90-day outcomes even as the underlying dollar-driven bull market remained intact.

The 2007-2008 period almost certainly contains some of the dataset's best 90-day outcomes (contributing to the +21.71% best case). RSI crossings that fired in late 2007 and early 2008 were catching the beginning of the financial crisis safe-haven surge — a case where overbought momentum was the *entry point* into a sustained move, not a warning. The force that overwhelmed any mean-reversion tendency was systemic credit fear, which intensified over the subsequent 90 days rather than dissipating.

The 2011 signals near gold's all-time peak at the time represent the opposite dynamic. RSI crossings above 70 in mid-2011 fired into a market where the fundamental driver — fear of dollar debasement following QE2 — was already priced, and the subsequent 90-day outcomes were among the dataset's worst as gold entered a multi-year bear market. The sequencing mattered: in the first 30 days after those signals, gold continued higher on momentum; by days 60-90, the reversal was underway.

The post-2020 signals introduce a third regime: structurally elevated inflation and central bank buying from emerging market reserve managers, which created a demand floor that made mean-reversion after overbought readings shallower than historical norms. This likely explains why the 60-day mean (+0.50%) is slightly better than the 30-day mean (+0.43%) — in recent years, momentum has had more staying power than in earlier cycles.

---

4. What This Means Now — Scenario Analysis

As of August 2026, the relevant question is which historical analog current conditions most closely resemble. The chart shows gold in a substantially elevated price regime relative to its 2001-2019 history, with the post-2019 breakout having established a new structural range. An RSI crossing above 70 in this environment carries different implications depending on which of three scenarios is operative.

Scenario A — The 2007-2008 Analog: If the current RSI crossing is firing into a genuine macro stress event (credit deterioration, dollar weakness, or a geopolitical shock that is still in its early innings), the historical precedent suggests the overbought signal is a false warning. The best 90-day outcome in the dataset (+21.71%) likely came from exactly this setup. The key confirming condition would be a dollar that continues to weaken and real rates that remain suppressed or declining over the subsequent 30 days.

Scenario B — The 2011 Analog: If the current crossing is firing into a market where the macro catalyst is already fully priced — where the fear or inflation narrative that drove the rally is peaking — the 90-day outlook deteriorates sharply. The median 90-day return of -0.58% and the worst-case of -14.2% are most relevant here. The confirming condition would be dollar stabilization or a hawkish policy shift that pushes real rates higher within 30-60 days of the signal.

Scenario C — The Post-2020 Structural Bid Analog: If central bank demand and de-dollarization flows continue to provide a demand floor, the signal may produce outcomes clustered near the mean (+0.43% to +0.92%) with limited downside — not a strong positive, but not the sharp reversals seen in 2011-2013. The key variable here is whether institutional and sovereign buying continues at the pace established since 2022.

The single variable that most cleanly separates these scenarios is the direction of the trade-weighted dollar in the 30 days following the RSI crossing. Dollar weakness has historically been the force that converted overbought momentum into sustained follow-through; dollar strength has been the force that converted it into reversals.

---

5. Actionable Implications — With Explicit Uncertainty

The confidence framing from Section 1 must govern everything here: p-values of 0.48-0.52 at 30 and 60 days mean the signal has no statistically demonstrated directional edge. Position sizing language should reflect this directly.

Claim 1 (Low confidence, high conviction on uncertainty): An RSI-above-70 crossing in gold is not a reliable sell signal. The win rate of 50.6% at both 30 and 60 days is statistically indistinguishable from a coin flip. Mechanically fading gold on this signal — shorting or reducing long exposure solely because RSI crossed 70 — has no historical justification in this dataset. *Breaks down if:* the specific macro context matches the 2011 analog precisely (peak fear narrative, dollar bottoming).

Claim 2 (Moderate confidence): The 90-day window is the most dangerous. The divergence between the +0.92% mean and the -0.58% median at 90 days signals a skewed distribution where a minority of large winners distort the average. Traders holding gold positions through a 90-day window after an RSI crossing should be aware that the *typical* outcome (median) is slightly negative, even if the *average* outcome is slightly positive. *Watch for:* real rate direction at the 30-day mark as an early indicator of which tail is more likely.

Claim 3 (Tactical, conditional): The dollar trend in the 30 days following the signal is the highest-value observable for scenario discrimination. If the trade-weighted dollar weakens materially in the month after an RSI crossing, the historical analog shifts toward Scenario A and the 90-day outlook improves. If the dollar stabilizes or strengthens, Scenario B becomes more probable and reducing gold exposure at the 30-day mark has historical support. This is not a mechanical rule — it is a conditional update that should be weighted against position size and existing macro conviction.

What to watch: Real rate direction (TIPS yields), dollar index trend, and whether any geopolitical or credit stress event that may have triggered the RSI crossing is still in its early or late stages. These three variables, not the RSI reading itself, are what the historical record shows actually determined outcomes.