Resources / Historical research
Research archive
The studies behind the tools, including rules, historical results, and limitations. A historical finding is not a live alert or a forecast.
GoldSignal research register
Every hypothesis gets a record.
A running archive of the signals, ownership patterns, and market behaviors tested across a decade of public filings.
Most ideas do not survive. The ones that do become GoldSignal research signals.
10 YRS
Filings window
243
Historical convergences
SIG-030 SURVIVES
Flagship test
The register is the research trail.
GoldSignal begins with public disclosures: insider transactions, major ownership filings, institutional positions, and strategic investments. Each potential pattern becomes a defined hypothesis. The rule is written down. Historical occurrences are identified. Forward returns are measured. Results are compared against a benchmark. Most hypotheses are weak, inconclusive, or fail entirely. That is expected. The purpose of the register is to preserve the entire research trail — not just the winners.
Public disclosure → testable rule → historical evidence. Not a marketing funnel.
Research register
SIG-030 · SURVIVES
Strategic Convergence
Two independent 10%+ holders converging on the same company within 12 months.
Rule: Two independent parties each cross the 10% ownership threshold in the same company within a 12-month window. Definition
243
Historical convergences
25
Investable events
10 YRS
Research window
One large shareholder may be wrong, affiliated, promotional, or acting for reasons unrelated to valuation. Two independent sophisticated investors crossing the same legally reportable ownership threshold is a different event. The hypothesis asks whether that convergence contains information.
As of 14 Aug 2026.
In-sample held after audit. Out-of-sample not run. Not a recommendation.
What “survives” means
A hypothesis survives when the pre-registered in-sample test held after audit. Survival is a recorded outcome on this register, not a promotion and not formal statistical validation. Out-of-sample confirmation has not been run. A survived test is not a recommendation.
Each signal begins as a thesis.
“When two independent investors each become 10%+ owners of the same small mining company within 12 months, subsequent returns may differ from the broader junior-mining market.”
Define it. Find every occurrence. Measure what happened. Try to disprove it. Keep the result.
Research framework
- Hypothesis
- The claim written down before the test is run.
- Rule
- The event definition that decides what counts.
- Universe
- The names that can enter the test.
- Outcome
- The forward result that is measured.
- Benchmark
- The comparison the result is judged against.
- Status
- The recorded verdict on this register.
- Core principle
- Every number should be traceable back to its source.