News from Now On — Edition 4

Lead Forecast · Middle East / Lebanon · Medium

Israel–Lebanon talks open in Washington as the truce holds only tenuously

The week's concrete diplomatic event moves to Washington, where Israeli and Lebanese delegations are due to meet from 23–25 June. The Forecast Engine expects halting progress toward a framework rather than a finished deal — and on the ground, a ceasefire that holds only tenuously: sporadic Israeli strikes and Hezbollah responses across the southern front, the IDF signalling it stays in place, and Hezbollah, excluded from the room, rejecting terms set without it. The talks and the violations are likely to run in parallel all week.

Thread LIN-ISRAEL-LEB — first called D+1 on 22 Jun 2026 (Medium). Prior call graded Hit. Forecast revised: rolled from D+2 to D+1 and elevated to lead.

Based on: PBS, Times of Israel, Al Jazeera

Nuclear / Diplomacy · Medium

US–Iran talks grind on; a date for the next round is floated

Negotiations continue inside the 60-day window after the Switzerland restart. Expect a venue or date for the next round to surface, with verification, IAEA access and the down-blending of Iran's ~440kg highly-enriched-uranium stockpile the central sticking points.

Market: Polymarket puts a 80% chance on a US–Iran permanent peace deal by 31 Dec 2026 (up from 74%). Polymarket · captured 23 Jun

Thread LIN-IRAN-NUKE — first called D+2 on 13 Jun 2026 (Medium). Prior calls graded Hit, Hit, Hit. Rolled from D+2 to D+1.

Based on: NPR, Al Jazeera, TIME

Markets / Energy · High

Crude range-bound in the mid-$70s to low-$80s as Hormuz traffic recovers

Iran shipped 30M+ barrels in a week and Washington granted a 60-day oil-sales license, capping the upside even as IRGC-managed routing keeps transits below normal. Expect two-way volatility on talks and Lebanon headlines rather than a decisive break.

Thread LIN-OIL — first called D+1 on 13 Jun 2026 (Medium). Prior calls graded Hit, Hit, Hit, Partial. Rolled from D+2 to D+1; range nudged toward the mid-$70s.

Based on: Al Jazeera, CNBC, OilPrice

Ukraine / Russia · High

Strikes continue; a limited gesture may resurface but the impasse holds

No ceasefire breakthrough is expected: the competing-ceasefire confusion persists and a prisoner swap or partial-truce proposal may resurface without shifting the front. Kyiv keeps leaning on France, Germany and the UK.

Market: Kalshi prices a Russia–Ukraine ceasefire by end-2026 at 45% (down from 48%). Kalshi · captured 23 Jun

Thread LIN-UKR-WAR — first called D+1 on 13 Jun 2026 (High). Prior calls graded Hit, Hit, Hit. Rolled from D+2 to D+1.

Based on: Al Jazeera, CFR, Chatham House

Middle East / Energy Logistics · Medium

Hormuz normalization accelerates under Iran's 60-day oil license

More commercial transits and gradually easing war-risk insurance premiums as Washington's licence takes effect — though IRGC-managed routing keeps traffic below pre-crisis norms and a permanent settlement stays out of reach.

Market: Polymarket's permanent-peace-by-year-end line sits at 80%. Polymarket · captured 23 Jun

Thread LIN-IRAN-DEAL — first called D+1 on 13 Jun 2026 (Medium). Prior calls graded Hit, Hit. Rolled from D+2 to D+1.

Based on: NBC News, Al Jazeera, OilPrice

Lead Outlook · Middle East / US Foreign Policy · Medium

Can the interim MOU convert into a durable peace before the August deadline?

With the 60-day clock running, the Forecast Engine expects the debate to sharpen over whether the interim memorandum can be turned into a lasting settlement. Divergent US and Iranian readings — on sanctions sequencing, asset releases and the long-term administration of the Strait of Hormuz — are likely to surface publicly, testing how much of the de-escalation is durable and how much is a pause.

Market: Polymarket's permanent-peace-by-year-end line sits at 80%, a benchmark against this thread. Polymarket · captured 23 Jun

Thread LIN-IRAN-DEAL — first called D+1 on 13 Jun 2026 (Medium). Rolled from D+3 to D+2.

Based on: CSIS, TIME, PBS

Middle East / Lebanon · Low

Washington round wraps with at most a partial framework

As the talks conclude (25 Jun), expect a partial framework at best; questions grow over any IDF-withdrawal timeline while Hezbollah, excluded, rejects the terms and low-level friction continues in the south.

Thread LIN-ISRAEL-LEB — first called D+1 on 22 Jun 2026 (Medium). Rolled from D+3 to D+2.

Based on: Times of Israel, Al Jazeera

Ukraine / Russia · Low

Strikes persist as European capitals press for a leaders' track

Expect movement in rhetoric and summit pressure rather than agreement on a ceasefire date, with European aid commitments continuing in parallel.

Market: Metaculus community puts a 2026 ceasefire at 40% (down from 47%). Metaculus · captured 23 Jun

Thread LIN-UKR-WAR — first called D+1 on 13 Jun 2026 (High). Rolled from D+3 to D+2.

Based on: Washington Post, Chatham House

UK Politics · Low

Immigration politics hardens as the post-riot reckoning continues

Expect polling shifts and tougher enforcement proposals at Westminster and Stormont, with charges proceeding (including minors) and policing under scrutiny.

Thread LIN-UK-RIOTS — first called D+1 on 13 Jun 2026 (High). Prior calls graded Hit, Partial, Partial, Hit. Rolled from D+3 to D+2.

Based on: Al Jazeera, Background

Nuclear / Verification · Low

First technical-level contacts on verification reported or contested

Preparatory talks on nuclear verification and HEU down-blending may be reported as the parties try to stabilize the negotiating track.

Thread LIN-IRAN-NUKE — first called D+2 on 13 Jun 2026 (Medium). Rolled from D+3 to D+2.

Based on: Al Jazeera, NPR

Lead Extended Outlook · Middle East / Nuclear · Low · speculative

Down-blending Iran's HEU stockpile becomes the crux of the talks

By the extended horizon the Forecast Engine expects the technical heart of the deal — how to down-blend Iran's roughly 440kg stockpile of highly enriched uranium and restore IAEA monitoring — to dominate. Expect contested progress and a real risk the round slips again as August-deadline pressure builds. This is a speculative outlook; confidence is Low by design at three days' lead.

Market: Polymarket's permanent-peace line at 80% frames the optimism the technical track has to live up to. Polymarket · captured 23 Jun

Thread LIN-IRAN-NUKE — first called D+2 on 13 Jun 2026 (Medium). Prior calls graded Hit, Hit, Hit.

Based on: TIME, Al Jazeera

Middle East / US Foreign Policy · Low

Sanctions-relief sequencing and ceasefire durability dominate

Markets keep pricing a year-end peace deal as likely even as implementation doubts and Lebanon spillover persist. The gap between market optimism and on-the-ground friction widens.

Thread LIN-IRAN-DEAL — first called D+1 on 13 Jun 2026 (Medium).

Based on: CSIS, NPR

Middle East / Lebanon · Low

Low-level friction persists; settlement stays out of reach

Southern Lebanon sees continued skirmishing; the IDF's open-ended presence and Hezbollah's exclusion from the Washington track keep a comprehensive settlement elusive.

Thread LIN-ISRAEL-LEB — first called D+1 on 22 Jun 2026 (Medium).

Based on: Times of Israel, Al Jazeera

Ukraine / Russia · Low

Stalemate holds; a fresh limited measure may be floated

A prisoner exchange or localized truce may be floated without altering the front or the ceasefire impasse.

Market: Kalshi end-2026 ceasefire line at 45%. Kalshi · captured 23 Jun

Thread LIN-UKR-WAR — first called D+1 on 13 Jun 2026 (High).

Based on: Chatham House, CFR

Markets / Energy · Low

Crude range-bound as Iran's exports ramp under the licence

Iran's rising exports weigh modestly on price while a residual risk premium from Lebanon and implementation doubts keeps a floor under the range.

Thread LIN-OIL — first called D+1 on 13 Jun 2026 (Medium).

Based on: OilPrice, EIA STEO

Forecast Scorecard — Edition 3 (D+1) Resolved

Edition 3 (22 Jun) forecast Tue 23 Jun at D+1; those five have now resolved. Edition 3's D+2/D+3 calls have target dates still ahead and were rolled forward into this edition. Each forecast is graded at the horizon it was made at; accuracy = (Hit + 0.5·Partial) / graded. Figures below are cumulative across all resolved editions.

Skill Curve — accuracy by horizon (cumulative)

D+1 90%

D+2 80%

D+3 70%

Skill curve holds its order across 20 graded: D+1 (10) 90% > D+2 (5) 80% > D+3 (5) 70%. Overall 82.5%.

Confidence Calibration

Tier	Graded	Score
High	4	87.5%
Medium	10	90%
Low	6	66.7%

High and Medium now run neck-and-neck (a High-tagged oil call landed Partial), both comfortably above Low — confidence still tracks accuracy.

Model vs Market

Comparable forecasts	2
Model correct	2
Market correct	2
Model beats market	0

First benchmark scoring. The Iran-nuclear call (vs Polymarket 74%) and the Ukraine call (vs Kalshi 48%) both resolved as the model predicted — but the markets pointed the same way, so no model-beats-market edge yet.

Per-Forecast Grades — Edition 3 D+1

Forecast	Made	Conf.	Grade
US–Iran nuclear talks grind on in the 60-day window	D+1	Med	Hit
Israel–Hezbollah truce holds only tenuously	D+1	Med	Hit
Crude high-$70s/low-$80s; Hormuz recovering	D+1	High	Partial
Ukraine strikes; competing-ceasefire impasse	D+1	High	Hit
NI calmed; focus shifts to charges & politics	D+1	Med	Hit

Edition 3 D+1: 4 Hit · 1 Partial · 90% — the oil call slipped to Partial as crude stabilized nearer $74, just below the forecast band. First-call and final grades are identical (each resolved within one horizon step); cross-edition trajectories appear in Story Threads below.

Running record (cumulative): 20 graded · 14 Hit · 5 Partial · 1 Miss · 82.5% accuracy.

Story Threads — Lineage

LIN-ISRAEL-LEB — Israel–Lebanon / Hezbollah ceasefire · first called D+1 on 22 Jun (Medium) · active · history: (Hit → today's D+1 lead)

LIN-IRAN-NUKE — Iran nuclear question & verification track · first called D+2 on 13 Jun (Medium) · active · history: (three straight Hits → carried across D+1/D+2/D+3)

LIN-IRAN-DEAL — US–Iran war settlement & Strait of Hormuz · first called D+1 on 13 Jun (Medium) · active · history: (Hit → Hit → carried to D+1/D+2)

LIN-OIL — Oil / energy de-escalation · first called D+1 on 13 Jun (Medium) · active · history: (three Hits, then a Partial on a High call)

LIN-UKR-WAR — Ukraine–Russia war & ceasefire diplomacy · first called D+1 on 13 Jun (High) · active · history: (three straight Hits)

LIN-UK-RIOTS — UK / Northern Ireland immigration unrest · first called D+1 on 13 Jun (High) · active · history: (Hit → Partial → Partial → Hit)

LIN-ISRAEL-IRAN — Israel posture toward US–Iran deal · first called D+1 on 13 Jun (Medium) · dormant · history: (Partial; no fresh call since edition 1)

LIN-IRAN-INTERNAL — Iran leadership & succession dynamics · first called D+3 on 13 Jun (Low) · dormant · history: (Miss — did not materialize)

The Markets — Live Odds Feeding Today's Forecasts

Question	Venue	Implied	Captured
US × Iran permanent peace deal by 31 Dec 2026?	Polymarket	80%	23 Jun
Russia × Ukraine ceasefire agreement by end 2026?	Kalshi	45%	23 Jun
Russia–Ukraine ceasefire in 2026?	Metaculus	40%	23 Jun

Model–vs–market read: Polymarket's peace line firmed to 80% on the signed MOU, a touch above the Engine's Medium lead, which still weights the Lebanon spoiler and the unresolved HEU stockpile. The Ukraine ceasefire lines drifted lower (Kalshi 48→45, Metaculus 47→40), converging toward the model's High confidence in continued conflict.

The Record — What Actually Happened (to 23 Jun)

How the Forecast Engine Works

News from Now On is an experiment in forecasting future news the way a weather service forecasts weather. Each model run reads the day's verified reporting and extrapolates three horizons forward: D+1 (tomorrow, detailed), D+2 (outlook), and D+3 (extended, speculative). Everything below the masthead is a prediction, not a report of fact.

Comparability. Accuracy is scored by lead time — each forecast is graded at the horizon it was made at — so the skill curve (D+1 vs D+2 vs D+3) is comparable edition to edition. A healthy model is more accurate at shorter range.

Lineage. Forecasts persist as threads with stable IDs. When yesterday's D+2 becomes today's D+1 it keeps its thread ID and gains a revision note, so a story's confidence trajectory can be traced from first call to resolution.

Markets. Each run reads prediction markets (Polymarket, Kalshi, Metaculus) both to calibrate confidence and as a benchmark: when an event resolves, the model is scored against the market that priced it. This edition logged the first two comparable resolutions.

Grading. Hit / Partial / Miss against subsequent reporting; accuracy = (Hit + 0.5·Partial) / graded. Confidence tiers map to implied probability: High ≈ ≥75%, Medium ≈ 45–75%, Low ≈ <45% or no liquid market.

This edition note: edition 3's five D+1 forecasts resolved (4 Hit, 1 Partial); its D+2/D+3 calls were rolled forward with revision notes. The market benchmark recorded its first scored resolutions. Published via the local git pipeline to the active news-from-now repo.