What happened later?
A later move can trigger review. It does not decide whether the original call was wrong.
DECISION PROOF
Proof, not hindsight.Freeze the decision before the outcome is known.
A Decision Receipt preserves what FlipForge knew and why it acted. Later evidence is compared against that frozen record.
CURRENT STATUS
Proof100 results are withdrawn while we re-audit them.
The Proof100 study froze 100 decisions on September 9–10, 2026. Its earlier results were withdrawn because an evidence-classification issue affected the original study. The issue has been fixed, and the original study records are preserved.
Our study rules say that a fix which changes how decisions are made requires a new cohort. A new, independently reviewed cohort, frozen on the corrected build, must establish any future performance claims.
Frozen-decision tracking does not prove that FlipForge will always make the right decision, guarantee profit, or establish a universal performance percentage. It holds every saved decision to account, including when we got something wrong.
HOW TO READ RESULTS
Three different things must stay separate
A later price move is not the same thing as a good original decision. Three layers stay separate:
Was BUY actually justified at T0?
Outcome evidence is reconnected to the original price, profitability gate, margin of safety, and Maximum Buy Price.
Did economics allow BUY but the recommendation still hold back?
Only a BUY-eligible non-BUY decision becomes a policy-review candidate. No rule changes automatically.
METHOD
The original decision stays frozen
The same frozen decision is followed through time. Later evidence adds context; it never rewrites T0.
T0
Exact identity, qualified evidence, supported value, risk, recommendation, and original economics are preserved.
T7
When sufficient governed outcome evidence exists, FlipForge records an observation and compares it with the frozen T0 decision.
T14
The same study denominator and original decision remain intact. New evidence adds another checkpoint rather than replacing history.
T30
Longer-horizon evidence is added under the same rules, enabling more stable calibration and cross-cohort comparison.
Review
Repeated findings may become a human-reviewed policy hypothesis, but no automatic self-training or production threshold change is authorized by this public proof layer.
WHAT COMES NEXT
A new cohort, held to the same rules
Unknown outcomes are never counted as wins or losses, and FlipForge does not lower evidence standards just to report more results.
Freeze a new cohort
A new set of decisions, predeclared and frozen on the corrected build before any outcome is known, and independently reviewed.
T7, T14 and T30 checkpoints
Later evidence is compared with each frozen decision. Unresolved outcomes stay unresolved.
A larger cohort
Longer horizons and a larger predeclared cohort come before any stronger performance claim.
Bring FlipForge a real decision.
Evaluate a real card, inspect the reason trail, and track how the decision holds up.