What happens to SI=F in the 90 days after the 50-day moving average crosses above the 200-day?

# SI=F Golden Cross: 90-Day Forward Performance Analysis

50-Day MA Crosses Above 200-Day MA | 2002–2026

---

1. Data & Confidence Context

This analysis is built on 18 confirmed golden cross events in SI=F (COMEX Silver Futures) spanning January 2002 through June 2026, derived from 3,439 signal state-days via run-start deduplication. Eighteen crossings is a meaningful but statistically modest sample — large enough to compute directional tendencies, too small to treat any single forward-return estimate as robust; the 90-day p-value of 0.1198 and 30-day p-value of 0.5391 both fall well short of conventional significance thresholds, meaning the near-term signal is essentially noise while the longer-horizon tendency is suggestive but not confirmable. Silver's notorious volatility — a commodity that can move 30%+ in a single quarter on industrial demand shifts, dollar moves, or speculative positioning — further compresses the signal-to-noise ratio, and readers should treat the patterns below as probabilistic tendencies with wide confidence intervals rather than reliable forecasts.

---

2. Direct Answer — What the Data Shows

The engine-computed forward returns across 18 golden cross events tell a story of a signal that is nearly useless at 30 days, meaningfully directional at 60–90 days, and most reliable at 180 days — but with variance wide enough to inflict serious damage on any position that ignores the downside tail.

Start with the aggregate picture. At 30 days, the mean return is just +1.39% against a base rate of 1.25%, producing a negligible excess of 0.14% — and the win rate of 33.3% means silver was *lower* two out of every three times at the one-month mark. The median of -1.71% confirms that the average is being dragged upward by a small number of large winners; the typical outcome in the first month after a golden cross is a modest loss. The p-value of 0.5391 is statistically indistinguishable from a coin flip.

The picture shifts materially at 60 days: mean return climbs to +6.03%, median to +4.77%, and win rate jumps to 61.1% — a genuine reversal of the near-term pattern. Excess return over the base rate is 3.44 percentage points, and the p-value of 0.1493, while still not significant at the 5% level, is no longer noise. Something real appears to be happening in the second month, even if it cannot be confirmed statistically with 18 observations.

At 90 days — the primary question — the mean is +5.39%, the median +7.10%, and the win rate holds at 61.1%. The best single outcome across all 18 crossings was +34.45%; the worst was -20.31%. That 54-percentage-point spread between best and worst is the defining feature of this signal: it is not a tight, reliable setup but a wide-distribution bet with a positive skew. The excess return over the 90-day base rate is only +1.64%, and the p-value of 0.1198 means roughly a 12% chance this excess is random — better than noise, but far from conviction.

The signal's strongest reading arrives at 180 days: mean +11.07%, median +11.0% (the rare alignment of mean and median signals a relatively symmetric distribution at this horizon), win rate 66.7%, and a p-value of 0.0309 — the only horizon that clears the 5% significance threshold. The best 180-day outcome was +48.46%; the worst, -23.43%.

Chronologically, the early crossings — January 2002, January and May 2003, August 2004 — occurred as silver was emerging from a multi-decade bear market, and the macro tailwind of a weakening dollar and rising commodity demand from China provided a powerful amplifier. The 2007 and 2009 crossings bracketed the financial crisis: the November 2007 signal fired into a deteriorating risk environment, while the March 2009 crossing caught the exact bottom of the crisis selloff and preceded one of silver's most explosive multi-year rallies. The September 2012 crossing came after silver had already pulled back sharply from its 2011 peak, and the subsequent 90-day return was negative — a reminder that a golden cross in a structurally weakening trend can be a false dawn. The 2016 and 2017 crossings produced mixed results in a range-bound environment, while the February 2019 crossing preceded a modest but positive 90-day move. The six most recent crossings (post-2019, exact dates not fully enumerated in the engine output) span the pandemic era and its aftermath, a period of extreme macro volatility that likely contributes to the wide return dispersion visible in the worst-case figures.

---

3. Confounding Factors — Decomposing What Actually Drove Each Cycle

Silver is not gold. It is simultaneously a monetary metal and an industrial commodity — roughly half its demand comes from manufacturing, solar panels, and electronics — which means that a golden cross in SI=F is never purely a technical event. The signal fires into a web of competing forces, and understanding which force dominated in each phase is more useful than the headline return number.

In the early 2000s crossings (2002–2004), the dominant force in the first three months was dollar weakness. The trade-weighted dollar entered a sustained multi-year decline beginning in early 2002, and silver's golden crosses in this period were essentially dollar-decline signals wearing technical clothing. The industrial demand story — China's infrastructure buildout accelerating — took over in months six through twelve, compounding the monetary tailwind. The sequencing mattered: dollar weakness provided the initial lift, then industrial demand sustained and extended it.

The 2007 crossing illustrates how a golden cross can fire into a deteriorating macro environment and still produce a short-term positive return before reversing. The signal triggered in November 2007 as credit markets were already seizing, and the initial 30-day return was likely positive as commodity momentum carried forward — but the subsequent financial crisis overwhelmed the technical signal entirely. Here, the confounding force was systemic credit risk, which eventually crushed industrial demand expectations and forced leveraged commodity positions to unwind.

The March 2009 crossing is the clearest case of a golden cross coinciding with a genuine macro inflection. The Federal Reserve had moved to near-zero rates, quantitative easing was beginning, and the dollar was weakening. The technical signal and the macro signal were aligned, and the result was one of the strongest forward returns in the dataset. The key lesson: when the golden cross fires simultaneously with a Fed pivot toward accommodation and dollar weakness, the signal's predictive power is amplified.

The September 2012 crossing demonstrates the opposite dynamic. Silver had peaked above $49 in April 2011 and was in a structural downtrend. The golden cross fired as a counter-trend bounce, not a trend reversal, and the 90-day return was negative. The confounding factor here was the absence of a new macro catalyst — real rates were low but not falling further, and the dollar was stabilizing. Without a fresh driver, the technical signal had no fundamental force to amplify it.

The 2016 crossing came in the wake of the February 2016 commodity bottom, driven by a sharp reversal in the dollar and renewed Fed dovishness after the December 2015 rate hike proved premature. Industrial demand from solar panel manufacturing was also accelerating structurally. The first three months saw strong returns; the subsequent period was more volatile as the dollar recovered post-election.

---

4. What This Means Now — Scenario Analysis

As of June 2026, the chart covering January 2000 through June 24, 2026 shows the full arc of silver's history across all 18 golden cross events. The six most recent crossings — those post-2019 not fully enumerated in the engine output — are the most relevant analogs for current conditions, as they reflect the post-pandemic macro regime of elevated inflation sensitivity, energy transition demand for silver in solar manufacturing, and a Federal Reserve that has moved through a full tightening cycle.

Scenario A — The 2009/2016 Analog (Bullish): If current conditions feature a weakening dollar trend, declining or plateauing real rates, and continued structural demand from solar and EV manufacturing, the current crossing most closely resembles the 2009 and 2016 setups. In those cases, the 90-day return was strongly positive, and the 180-day return was the most reliable horizon. The key variable confirming this scenario would be a sustained break lower in the DXY (trade-weighted dollar) concurrent with the crossing. Under this analog, the 90-day forward expectation would be toward the upper half of the historical distribution — above the +7.10% median, potentially approaching the mean of +5.39% to +11% range.

Scenario B — The 2012 Analog (Bearish): If silver has already made a significant run prior to the crossing — meaning the golden cross is a lagging confirmation of a move that is largely complete — and if the dollar stabilizes or strengthens while real rates remain elevated, the 2012 analog applies. In that case, the 90-day return was negative, and the -20.31% worst-case outcome in the full dataset becomes a live risk. The confirming signal for this scenario would be a failure to hold the 200-day MA as support in the weeks immediately following the crossing.

Scenario C — The 2017/2019 Analog (Neutral/Choppy): A range-bound dollar, mixed industrial demand signals, and a Fed on hold would produce the 2017–2019 pattern: a positive but modest 90-day return, high intra-period volatility, and a wide range of outcomes depending on whether a specific catalyst (geopolitical event, inflation surprise) materializes. This is arguably the base case given the 61.1% win rate and the median of +7.10% — positive but not dramatically so.

The key variable that determines which scenario plays out is the dollar trend in the 30 days following the crossing. In every high-return golden cross in this dataset, dollar weakness was either already established or accelerated within the first month. In the low-return or negative cases, the dollar was flat or strengthening.

---

5. Actionable Implications — With Explicit Uncertainty

Given the 90-day p-value of 0.1198 and a sample of 18 crossings, this signal warrants reduced position sizing relative to a statistically confirmed setup. The confidence level supports a directional lean — not a high-conviction trade.

Causal mechanism identified: The golden cross in SI=F has historically preceded positive 90-day returns in 61.1% of cases, with a mean of +5.39% and median of +7.10%, but the mechanism is not the technical signal itself — it is the macro alignment that the signal tends to coincide with (dollar weakness, accommodative Fed, industrial demand acceleration). The signal is a proxy for macro conditions, not an independent cause.

Conditions under which it holds: Dollar in a downtrend or weakening; real rates flat or declining; silver holding above the 200-day MA as support in the 30 days post-crossing; industrial demand (solar, electronics) structurally intact.

Conditions under which it breaks down: Dollar strengthening; real rates rising; silver failing to hold the 200-day MA; the crossing occurring after a large prior run (lagging confirmation rather than early signal). The -20.31% worst-case outcome is a live risk in any of these conditions.

What to watch: The 30-day post-crossing return is the canary — the historical data shows a 33.3% win rate at 30 days, meaning an initial dip is the *expected* outcome, not a signal to exit. A loss in the first month should not trigger a stop unless the 200-day MA is violated on a closing basis. The 60-day mark is where the signal historically begins to differentiate winners from losers; a positive return at day 60 has historically been associated with continued strength through day 90 and 180.

Position sizing language: Given p=0.1198 at the 90-day horizon and the 54-percentage-point spread between best and worst outcomes, risk no more than you would on a setup with roughly 60% directional confidence and a 1.5:1 reward-to-risk ratio — because that is precisely what the data supports, no more.