Due formati pubblicati · two published formats
PULSE INSIDER guarda dentro un soggetto · FP3 OUTLOOK guarda avanti dentro un mercato. Entrambi gratuiti, datati, con le previsioni a registro pubblico e criterio di riscontro pre-impegnato. Nessun documento pubblicato è a pagamento: ciò che si paga è un servizio, mai un documento →
EN · A PULSE INSIDER looks inside a subject; an FP3 OUTLOOK looks ahead inside a market. Both free, dated, on the public register with pre-committed resolution criteria. Nothing published is paywalled — what is sold is a service, never a document.

FP3NEWS · Editorial Track Record

Predictions & Outcomes

A dated public register · prediction precedes outcome · resolved on published evidence · updated 20/07/2026

Versione completa bilingue · full bilingual register →

Every forecast on this record is certified by the date of the document that contains it — a dossier, a working session, a dated communication — so that publication precedes outcome. Nothing is scored after the fact; each entry states its original wording, its dated source and its outcome. The edge FP3 claims is not a ranking but an architecture: registered press + a public outcomes register + certification by date + §17 falsifiability + Italy-macro depth + bilingual delivery. The resolved sample is small and growing — and we say so.

57resolved correctly ✓

317open, awaiting outcome

380on register

6missed

90%resolved hit-rate57 of 63 resolved · small-N, stated openly

317open, untested83% of the register

m50latency discipline≥⅓ of entries at <12-month horizon

Full register — machine-readable export
⇩ CSV
⇩ JSON

The hit-rate is computed over resolved items only (57 confirmed + 6 missed); the 317 open entries are untested and can move it. No Brier score: outcomes are binary, on published evidence, at resolution date. Anteriority proof and specifics available on request.

Confirmed — selected

Football · FIFA World Cup 2026issued: 08/06/2026 (pre-tournament) · scenario tree 18/07/2026 (eve of final)Confirmed

Spain champions as the top pick of a distributional call (Spain 16% · France 13%) logged before the 11 June kickoff; on the eve of the final, a pre-committed scenario tree assigned Spain 0.56 against Argentina, with the disconfirming branch published in advance.

Outcome — 19/07/2026: Spain beat Argentina 1-0 in extra time (Ferran Torres, 106'), MetLife Stadium — second title after 2010. Both the pre-tournament modal pick and branch A of the final tree are realised. Symmetric §17 note on record: the «first-time winner ~36%» branch did not materialise.

Dated sources: analytical session 08/06/2026 (pre-kickoff) · PULSE FLASH «The Asymmetric Final» 18/07/2026 · outcome: FIFA · ANSA · Sky, 19–20/07/2026 (Tier-1). Register codes: MONDIALE-2026-DISTR · FINALE-SCEN-180726.

Geopolitics · USA · Presidential electionissued: October 2024 (reserved study) · communicated 01/11/2024Confirmed

Donald Trump to win the 5 November 2024 US presidential election — thesis developed in a reserved study of September–October 2024 and communicated to ~20 contacts on 1 November 2024, four days before the vote, while the race was rated a coin-flip by mainstream forecasters (Silver · Economist ≈ 50/50) amid a late surge of Democratic optimism.

Outcome — 05/11/2024: Trump wins with 312 electoral votes, sweeps all 7 swing states and the popular-vote plurality. Direction confirmed.

Dated source: reserved FP3 · FreePress3News study, Sep–Oct 2024 · communication to ~20 contacts 01/11/2024 · Director Iacopo Mutascio. Anteriority self-attested (dated reserved communication); the outcome is public and certain, the origin date is attested by the Director, not by a third party — explicit §17 flag; proof on request.

Geopolitics · USA–Iran · Hormuzissued: 16/05/2026 (genesis 30/03/2026)Confirmed

The ceasefire as managed friction, not peace: resumption of significant US kinetic action against Iran within 30–90 days [P = 0.52], with no structural normalisation of the Strait of Hormuz — issued while the deal was being hailed as a diplomatic victory.

Outcome — 07–08/07/2026: CENTCOM strikes 80+ targets in Iran; Iran retaliates on US bases in Kuwait and Bahrain; the ceasefire is declared "over". Above §17 threshold.

Dated source: analytical sessions 30/03/2026 & 16/05/2026 (per-window table, §17 threshold) · PI-005.

Politics · Referendumissued: March 2026, pre-voteConfirmed

NO victory in the constitutional referendum on justice reform, 52–54% band.

Outcome — NO at 53.74% (22–23/03/2026). Direction and magnitude within band.

Dated source: PULSE INSIDER 005bis + pre-blackout analytical sessions.

Geopolitics · USA–Italyissued: March–April 2026Confirmed

USA–Italy trajectory rupture after the Sigonella denial: the Trump–Meloni freeze as a structural, not episodic, consequence.

Outcome — Trump–Meloni freeze materialised and documented; confirmation registered 29/06/2026, thesis genesis certified to March/April 2026.

Dated source: March–April 2026 papers and analytical sessions.

Open the full bilingual register →

Expanding register. Predictions documented in dated working sessions and communications are added as original sources are retrieved and quoted. Each entry can be verified by opening the source document indicated. The full historical documentation remains available in the archive; positioning and methodology are stated in the disclosure and the method.

◆ Calibrazione · calibration

Il tasso di riscontro non è una misura di bravuraA hit-rate is not a measure of skill

Un previsore calibrato che lavora su domande difficili sbaglia circa il 30% delle chiamate dichiarate al 70%. Un tasso aggregato molto alto può quindi segnalare il contrario di ciò che sembra: che la difficoltà media delle domande registrate sia bassa. La misura che conta non è il totale, è l’esito dentro ciascuna banda di probabilità. Questa tabella è calcolata in diretta sul registro derivato, non scritta a mano. A calibrated forecaster working on hard questions gets about 30% of their 70% calls wrong. A very high aggregate rate can therefore signal the opposite of what it appears to: that the average difficulty of the registered questions is low. What matters is the outcome within each probability band, not the total. This table is computed live from the derived register, not written by hand.

Voci registrate e mai risolteRegistered and never resolved

Il denominatore onesto di un registro include le voci che sono state iscritte e non hanno mai ricevuto un esito. Qui sotto il conteggio, con la quota di quelle la cui data codificata nell’identificativo è già passata. Alcune restano aperte per costruzione, perché il criterio pre-impegnato rimanda a una pubblicazione successiva; altre no. La distinzione è contestabile e va contestata. The honest denominator of a register includes entries that were filed and never received an outcome. Below is the count, and the share whose date encoded in the identifier has already passed. Some stay open by design, because the pre-committed criterion defers to a later publication; others do not. The distinction is contestable and should be contested.

Metodo: la banda è letta dalla probabilità puntuale dichiarata nel testo della voce; le voci senza probabilità puntuale sono contate a parte e non entrano nelle bande. La data di scadenza è ricavata dal suffisso dell’identificativo. Entrambe le derivazioni sono meccaniche e verificabili su fp3-registro.json. Storico delle calibrazioni: calibration record →. Method: the band is read from the point probability stated in the entry text; entries without a point probability are counted separately and excluded from the bands. The due date is derived from the identifier suffix. Both derivations are mechanical and checkable against the JSON.