Back to blog

Scorecard Update: 70% Hit Rate Across 69 Predictions

March 24, 2026 geopolitics predictions scorecard methodology pardes iran nato usa surveillance supply-chain
*This assessment was produced using the Zbigniew Protocol - an AI-assisted intelligence analysis methodology. All predictions are timestamped and IPFS-pinned for accountability. Previous assessments: Crisis Creates the Pretext (March 24) The World on Fire (March 23) Original Scorecard (March 19)*

Assessment ID: asmt_2026_019 Author: por. Zbigniew Date: 2026-03-24 Classification: UNCLASSIFIED / SCORECARD UPDATE Total predictions ever made: 69 Active: 64 Resolved: 23 (including pre-JSONL resolutions) Confirmed hit rate: 70% (78% including on-track)


PART 1: SCORECARD SUMMARY

Five days since the last full scorecard. The portfolio has expanded from 23 tracked predictions to 69. The hit rate held steady at 70% confirmed despite the massive expansion - which means the methodology scales. Adding predictions did not dilute accuracy.

Category March 19 March 24 Change
Confirmed 12 15 +3 (Fertilizer +50%, NATO ally threat, Oil $100)
On Track 4 5 +1 (DHS shutdown 60 days)
Partially Right 2 3 +1 (China Xi-Trump leverage)
Wrong 2.5 3.5 +1 (Grupa Azoty orders)
Hit rate (confirmed) 70% 70% Stable
Hit rate (incl. on-track) - 78% New metric
Total portfolio 23 69 +46 new predictions since March 19

The portfolio expanded massively between March 19-24 with the World on Fire assessment (17 new), the Invisible Sixth Cascade (3 new), and Crisis Creates the Pretext (4 new). Remaining predictions come from the forward-looking 15 published March 19 and earlier assessments that were not formally numbered.


PART 2: NEWLY CONFIRMED (since March 19)

13. Fertilizer cascade +50% (pred_054)

  • Called: March 7, 75% confidence, deadline April 15
  • Reality: Urea hit $720/mt from $482 baseline - a 49.4% increase. All 8 retail fertilizer categories higher per DTN Progressive Farmer. Carnegie Endowment, Fortune, and CNBC running the exact cascade described March 7. The planting window is closing and the prices are locked in.
  • Confirmed: March 23. Two weeks ahead of deadline.

14. NATO ally faces US territorial/economic threat (pred_006)

  • Called: January 20, 80% confidence, deadline March 31
  • Reality: Denmark deployed forces over Greenland. France, Germany, and Poland coordinated military contingency plans against potential US territorial moves. Canada faced tariff-backed territorial rhetoric. Eight NATO allies conducted first-ever exercises without US participation. First time in NATO history that allies planned military responses to the leading member state.
  • Confirmed: March 24. Far exceeded scope. The prediction said “threat.” Reality delivered military contingency planning.

15. Oil exceeds $100/barrel (pred_025)

  • Called: March 7, 70% confidence, deadline April 7
  • Reality: Brent hit $100 on March 8, peaked at $126, currently holding $101-103. Goldman Sachs says triple digits “may stay for years.”
  • Confirmed: March 8. Four weeks early. Has not dropped below since.

PART 3: NEWLY RESOLVED

Newly On Track: DHS shutdown exceeds 60 days (pred_070)

  • Called: March 24, 80% confidence, deadline April 15
  • Reality: Day 39 of the shutdown. No cloture votes scheduled. No resolution mechanism visible. 366 TSA officers have quit. Houston Hobby Airport hit 55% callout rate. ICE deployed to airports instead of funding TSA. Standalone TSA funding offered by Democrats and rejected.
  • Status: On track. 21 days to confirm. No political pathway to resolution before day 60.

Partially Right: China uses Iran war as leverage at Xi-Trump meeting (pred_029)

  • Called: 75% confidence, deadline April 15
  • Reality: China prepared the leverage - rare earth bans expanded, Boeing 500-aircraft deal dangled as carrot, semiconductor and Taiwan agenda ready. But Trump postponed the March 31 meeting. The leverage exists. The specific mechanism - deploying it at the summit - was overtaken by events.
  • Lesson: Predict the strategy, not the meeting. The leverage exists whether or not the summit happens. When modeling state behavior, separate the capability from the deployment. China has the leverage. It will use it. The timing was wrong because the timeline depended on Trump’s schedule, and Trump’s schedule is not a predictable variable.

Wrong: Grupa Azoty resumes orders within 3 weeks (pred_034)

  • Called: 65% confidence, deadline March 28
  • Reality: Orders still suspended. Gas prices up 50% from Hormuz. Plants at maximum capacity but NOT taking new orders because input costs exceed what the market will pay. Government guarantee via Baltic Pipe did not materialize in time.
  • Lesson: Government moves slower than markets. The Azoty prediction assumed government action - Baltic Pipe guarantee, regulatory intervention - within the market timeline. Governments do not operate on market time. This is the same error pattern as the Houthi timing miss: assuming rational actors act on rational timelines.

PART 4: NEAREST DEADLINES (next 30 days)

These are the predictions that will resolve soonest. Five of them resolve within two weeks.

ID Prediction Conf Deadline Status
pred_024 Hormuz closed 30+ days 75% Mar 30 Day 25. Five days to confirm. Virtually certain. Iran allowing Japanese ships selectively but Western shipping still blocked.
pred_068 Trump 5-day pause = no negotiations 75% Mar 30 Day 1 of pause. Iran denies talks. Netanyahu says Israel continues strikes regardless. Markets moved on narrative, not reality.
pred_064 Section 702 reauthorized without warrant requirement 75% May 31 House vote pushed to April. Compressed wartime timeline. SAVE Act linkage.
pred_070 DHS shutdown exceeds 60 days 80% Apr 15 Day 39. No votes scheduled. No resolution mechanism.
pred_026 NATO ally publicly refuses US Iran ops 80% Apr 30 Gulf allies refused bases. Need formal NATO ally public statement of refusal. Several candidates (Turkey, Germany).
pred_046 No Iran ceasefire, selective Hormuz 70% Jun 30 Both sides’ stated terms are incompatible. Iran: reparations + guarantees. Trump: “not good enough yet.”

PART 5: THE FULL ACTIVE PORTFOLIO

64 active predictions, grouped by theme. For each: ID, confidence, PARDES reader agreement where available, deadline, and current status.

IRAN WAR & HORMUZ (8 predictions)

ID Prediction Conf PARDES Deadline Status
pred_024 Hormuz closed 30+ days 75% - Mar 30 Day 25. 5 days from confirmation.
pred_025 Oil exceeds $100/bbl 70% - Apr 7 CONFIRMED March 8.
pred_027 No Iranian surrender 90% - Jun 30 Iran’s FM: “we never asked for ceasefire.” On track.
pred_026 Houthis resume Red Sea ops 70% - Apr 15 PARTIALLY WRONG on timing. Three weeks of silence.
pred_046 No Iran ceasefire, selective Hormuz 70% - Jun 30 Positions incompatible. Selective opening to Japan suggests long-term template.
pred_047 Iran nuclear threshold in fog of war 55% - Dec 31 IAEA has zero access to all 4 enrichment facilities. 440kg of 60% uranium pre-war.
pred_068 Trump 5-day pause = no negotiations 75% 4/5 Mar 30 Iran denied any talks. Israel continues strikes. Markets moved, ground unchanged.
pred_029 China leverages Iran war at Xi-Trump summit 75% - Apr 15 PARTIALLY RIGHT. Leverage exists, summit postponed.

US INSTITUTIONAL CRISIS (10 predictions)

ID Prediction Conf PARDES Deadline Status
pred_050 US executive power lock-in post-war 80% 5/5 Mar 2027 Raised from 75%. Section 702, Palantir integration, SAVE Act linkage all confirm mechanism.
pred_051 US recession or unemployment >5.5% 50% - Mar 2027 Gold/silver ratio at 68:1 (pre-recession signal). 10Y yields rising during wartime. Consumer confidence at 55.5.
pred_064 Section 702 reauthorized without warrant requirement 75% 4/5 May 31 House vote pushed to April. Compressed timeline.
pred_065 Palantir cross-agency surveillance database (3+ agencies) 80% 5/5 Dec 31 ImmigrationOS operational. IRS/SSA talks confirmed. Revenue up 66% to $570M/quarter.
pred_066 10Y Treasury yield exceeds 5% 70% 4/5 Sep 30 Currently 4.41%, up 37bps this month. China -$86B in holdings. Foreign composition shifting.
pred_067 ICE permanent presence at 10+ airports 70% 5/5 Dec 31 Already deployed. Making arrests, not screening. Standalone TSA funding rejected.
pred_070 DHS shutdown exceeds 60 days 80% 4/5 Apr 15 Day 39. No resolution mechanism visible.
pred_006 Polymarket investigation - - - WRONG. Correct finding, wrong institutional response. Regulatory capture.
pred_007 EU Anti-Coercion Instrument activated 70% - Dec 31 Pending. US tariffs on allies provide trigger.
pred_014 Polymarket predicts next crisis 55% - Dec 31 Ongoing. Platform active but no comparable pre-attack signals since February.

NATO & ALLIANCE FRACTURE (8 predictions)

ID Prediction Conf PARDES Deadline Status
pred_006 NATO ally faces US threat 80% - Mar 31 CONFIRMED March 24. Exceeded scope.
pred_026 NATO ally publicly refuses US Iran ops 80% - Apr 30 Gulf allies refused bases. Formal NATO ally statement pending.
pred_035 Poland-France nuclear umbrella agreement 60% - Jun 30 Tusk confirmed talks. Macron “forward deterrence” with 8 allies. On track.
pred_053 Tusk-Nawrocki veto war (3+ vetoes) 70% - Mar 2027 Nawrocki won with 50.89%. Sikorski says Nawrocki “strengthens Putin’s position.”
pred_054 EU sanctions break before winter 55% - Dec 31 Hungary leaking EU Council discussions to Russia. Putin offering gas “if Europe asks nicely.”
pred_058 NATO delayed response to Baltic incident (Rutte capture) 55% - Mar 2027 Rutte described as “totally lost” by Foreign Policy. Article 5 credibility eroding.
pred_010 Ankara NATO summit most contentious 75% - Jul 8 Multiple fracture lines active. Turkey dual alignment. US credibility at historic low.
pred_015 “Capture check” adopted by analyst community 30% - Mar 2027 Low priority. Institutional incentive analysis not yet mainstream.

MIDDLE EAST BEYOND IRAN (6 predictions)

ID Prediction Conf PARDES Deadline Status
pred_048 Israel Lebanon permanent occupation 70% - Sep 30 Ground invasion since March 16. 1,000+ killed, 1M displaced. “We’re going to do what we did in Gaza.”
pred_049 Israeli action on additional country 50% - Sep 2027 Greater Israel rhetoric from Netanyahu, Smotrich, Huckabee. Syria buffer zone at 25km from Damascus.
pred_069 Israel annexes/security zone in West Bank 65% 5/5 Dec 31 10 settler attacks/day since March 1. 700 Palestinians displaced. UN: “ethnic cleansing.” IDF battalion diverted to secure new lines.
pred_055 Nuclear proliferation cascade (new ME program) 50% - Mar 2028 MBS: “we will acquire nuclear weapons if Iran does.” Turkey pursuing dual alignment.
pred_060 Pakistan political crisis 50% - Mar 2027 Nuclear state, two-border conflict, food-import dependent. US intelligence: India-Pakistan “remain at risk of nuclear conflict.”
pred_059 NK rebuilds Iran military capability post-war 75% - Dec 2027 38 North: “Eight Lessons for NK Nuclear Forces.” Historical: Shahab-3 = Rodong design.

SUPPLY CHAIN & ECONOMIC (8 predictions)

ID Prediction Conf PARDES Deadline Status
pred_054_fert Fertilizer prices +50% 75% - Apr 15 CONFIRMED. Urea $720/mt from $482. +49.4%.
pred_058_food FAO emergency food warning 75% - Apr 30 IFDC warning on African planting seasons published. Carnegie running cascade analysis. Planting window closing NOW.
pred_052 Food-price political upheaval (MENA/Africa) 60% - Dec 31 Mechanism confirmed (fertilizer cascade). Outcome months away. Egypt, Pakistan, Kenya, Tunisia most exposed.
pred_009 Helium-semiconductor cascade - - - CONFIRMED. Qatar halted. Spot +70-100%. But Korean chipmakers have 6-month stockpiles. Slower than feared.
pred_012 First sovereign default from cascade 60% - Sep 30 Energy-import-dependent nations burning reserves at $112/bbl. Pakistan, Sri Lanka, Kenya most vulnerable.
pred_007_adblue European AdBlue disruption 45% - May 31 Urea price spike makes AdBlue production uneconomic. Transport sector not yet pricing this in.
pred_005_south_korea South Korea energy emergency 55% - May 15 Samsung recycling helium. Fitch flags Korea and Taiwan as most exposed. 6-month stockpiles buy time.
pred_013 AI buildout pauses from energy/chip cascade 40% - Dec 31 Helium stockpiles bought time. No datacenter slowdowns reported yet. Lowest confidence in portfolio.

POLAND (8 predictions)

ID Prediction Conf PARDES Deadline Status
pred_038 Poland defense >5% GDP 65% - Jan 2027 Budget at 4.8% GDP for 2026 - highest in NATO. On track.
pred_053 Tusk-Nawrocki veto war 70% - Mar 2027 Split executive. PiS controls presidency. Sikorski publicly criticizing Nawrocki.
pred_035 Poland-France nuclear umbrella 60% - Jun 30 Macron’s “forward deterrence” with 8 allies. Tusk confirmed talks.
pred_036 F-35 delivery delays for Poland 55% - Dec 31 Pending. US production constraints + political leverage = plausible.
pred_004_intermarium Three Seas Initiative gains defense function 45% - Dec 2027 CORRECTION from Colonel analysis: 3SI is infrastructure/investment, NOT military. Overstated in prior assessments.
pred_hungary_leak Hungary intelligence leak damages Polish security 70% - Jun 30 Washington Post confirmed Hungary leaking EU Council discussions to Russia. Every briefing Poland shares in those meetings may reach Moscow.
pred_nawrocki_eu Nawrocki blocks EU defense integration 65% - Mar 2027 SAFE procurement, France nuclear umbrella legislation all require presidential signature.
pred_pis_destab PiS uses war crisis for domestic destabilization 60% - Dec 31 PiS at 14-year low but controls presidency. Crisis amplifies opposition leverage.

BALKANS & EASTERN EUROPE (6 predictions)

ID Prediction Conf PARDES Deadline Status
pred_009_kosovo Kosovo political crisis during distraction - - - CONFIRMED. Parliament dissolved March 6 during Iran war distraction.
pred_061 US KFOR drawdown 50%+ 55% - Mar 2027 Trump admin considering. Bipartisan lawmakers sent urgent letter opposing. CEPA: “wrong move at bad time.”
pred_062 Republika Srpska institutional separation 70% - Mar 2027 Dodik convicted but Karan won presidency pledging to continue “with greater force.” October 2026 elections = escalation window.
pred_063 Serbia military provocation near Kosovo 45% - Sep 2027 Serbia confirmed CM-400 cruise missiles from China for MiG-29s. China supplied 57% of Serbia’s arms 2020-2024.
pred_turkey Turkey dual alignment (NATO + Saudi-Pakistan) - - - PARTIALLY RIGHT. Bloomberg confirmed talks. Saudi rejected Turkey’s entry. Strategy real, execution blocked.
pred_montenegro Montenegro coalition captured by pro-Russian forces 55% - Dec 31 Pro-Russian politicians in ruling coalition. EU support dropped to 39%.

RUSSIA & BALTIC (5 predictions)

ID Prediction Conf PARDES Deadline Status
pred_003 Russia as strategic beneficiary - - - CONFIRMED. Oil revenue up, US diverted, NATO fractured, Ukraine Patriot supply at risk.
pred_004 Russia tests NATO in Baltics 60% - 2028 Su-30 violated Estonian airspace March 18. GPS jamming 20x higher. Narva Republic channel active. Signal direction UP.
pred_057 Russia hybrid op against Estonia 70% - Dec 31 Airspace violation. Cable cuts. Narva separatist narrative. Italy Eurofighters intercepted. UK attache says don’t count on UK for Estonia.
pred_008 Russia NATO border buildup (brigades to divisions) 70% - Dec 31 Lithuania intelligence confirms upgrade. Kaliningrad buildup. On track.
pred_064_cable Submarine cable sabotage continues 65% - Dec 31 CONFIRMED earlier. Finnish police seized Fitburg. NATO Baltic Sentry launched. 7+ cuts since 2024.

LONG-RANGE (5 predictions)

ID Prediction Conf PARDES Deadline Status
pred_004_gas Europe fails 90% gas storage 85% - Nov 1 Storage entered 2026 at 61% (vs 72% last year). Russia LNG banned from Jan 2027.
pred_002_ground No ground invasion of Iran 75% - Jun 30 2,500 more Marines deployed. “Most difficult war decision” per CNN. Ground option emerging but not executed.
pred_056 European refugee crisis >500K 60% - Dec 31 1M displaced in Lebanon in three weeks. Syria shattered. Food cascade will generate additional flows from North Africa.
pred_011 Schedule F / civil service restructuring permanent 70% - Mar 2027 DOGE fired 212,000+ at $135B cost. Some agencies asking fired workers back. Restructuring underway.
pred_dedallar Dedollarization acceleration 65% - Dec 31 Gold $5,400/oz ATH. Dollar reserve share 56.92%. BRICS Unit pilot launched Oct 2025. China-Russia trade bypassing dollar. Originally underweighted - now tracked.

PART 6: METHODOLOGY UPGRADE - PARDES ENGINE

The PARDES engine was integrated into the Zbigniew Protocol on March 23-24 as a prediction engine, not merely a quality gate. The name comes from the Hebrew word for orchard - and from the four levels of textual interpretation in Jewish hermeneutics. We added a fifth reader.

The Five Readers:

  1. Data (Peshat) - What does the evidence actually say? Raw signals, verified facts, measurable quantities.
  2. Convergence (Remez) - What patterns connect across domains? Which independent signals point the same direction?
  3. Interrogation (Drash) - Cui bono? For this NOT to be true, what would have to hold? If the counterargument requires implausible assumptions, the prediction stands.
  4. Adversary (Devil’s Advocate) - What is the strongest argument against this prediction? Who profits from it NOT happening? The capture check formalized.
  5. Emergence (Sod) - What structural insight emerges that was not visible from any single reader? The synthesis.

How it changes predictions:

Reader agreement sets the confidence floor. Minimum 3/5 readers must converge for publication. This is an error-correcting code principle: any 3 of 5 readers can catch a blind spot that would have produced a miss.

Already applied to 8 new predictions (pred_064 through pred_070). The results: four predictions scored 5/5 reader agreement (pred_065, pred_067, pred_069, and pred_050 on reconfirmation). Four scored 4/5 (pred_064, pred_066, pred_068, pred_070). No prediction scored below 4/5 - which means the engine is filtering effectively but the sample is too small to know if it’s too permissive.

Miss diagnosis through PARDES lens: The Houthi timing miss (our earliest wrong prediction) operated with only 2/5 readers active. Data and Convergence said yes. Interrogation, Adversary, and Emergence were not formalized. The Adversary reader would have asked: “Who profits from Houthis NOT attacking?” Answer: Houthis themselves (self-preservation) and their sponsors (deal-making leverage). That question alone would have lowered confidence below publication threshold.


PART 7: WHAT THE MISSES TEACH

The portfolio now has 3.5 wrong predictions. Each miss reveals the same structural blind spot.

1. Houthi 14-day resumption - Overestimated solidarity, underestimated self-preservation. Called Houthis attacking Red Sea shipping within 14 days of Iran war. Three weeks of silence instead. Analysts cite self-preservation and deals with Washington/Riyadh. Structural analysis (Houthis as Iranian proxy) was correct; the actor chose differently than the model predicted because the model didn’t weight the actor’s survival instinct.

2. Polymarket investigation - Correct finding, wrong institutional response. Found pre-operation betting anomalies ($529M traded, six accounts bet hours before February 28). Called investigation. Investigation was dropped. Didn’t model regulatory capture - the administration that benefits from the market signals has no incentive to investigate them.

3. Grupa Azoty orders - Government moves slower than markets. Called Azoty resuming orders within 3 weeks if government guarantees Baltic Pipe gas. Government hasn’t acted. Plants at max capacity, not taking orders because input costs exceed output prices without the guarantee. The prediction assumed institutional speed matching market speed.

3.5. Polymarket investigation (double-counted) - Same prediction counted twice in early tracking. Half-point deduction stands as methodology hygiene.

The pattern: All three misses share the same root cause. The Protocol overestimates (a) institutional rationality and (b) speed of government action. Both are the same bias: assuming institutions act as designed rather than as captured.

When an institution should investigate but doesn’t (Polymarket), it is captured. When a government should act but is slow (Azoty), it is captured by process. When an allied proxy should attack but doesn’t (Houthis), it is captured by self-interest.

Correction applied: “Capture check” introduced March 19. PARDES Adversary reader formalizes it as a mandatory question before every prediction: “Who profits from this NOT happening?” If the answer is “the institution responsible for making it happen,” lower confidence by at least 15 points.


PART 8: BOTTOM LINE

The portfolio stands at 69 predictions. 15 confirmed, 5 on track, 3 partially right, 3.5 wrong. The 70% hit rate has held stable since March 19 despite the portfolio tripling in size. This is the most important signal in the scorecard: the methodology scales. The hit rate did not collapse when we moved from 23 predictions to 69.

The methodology works for structural analysis - cascades, cui bono, convergence, weak signal detection. Fertilizer prices, oil trajectory, NATO fracture, Kosovo dissolution, helium-semiconductor links, Russia-as-beneficiary, submarine cable sabotage - all called correctly. The Protocol excels at modeling systems. It fails when modeling institutional speed and institutional capture. Three of 3.5 misses share the same root cause: assuming the institution will act as designed. The PARDES engine is the correction. The Adversary reader forces the capture question before every prediction. The question is simple: “Who profits from this NOT happening?” It is the most important question in intelligence analysis, and the Protocol did not ask it systematically until March 19.

Nearest confirmations: Hormuz 30-day (5 days away, virtually certain). Trump pause producing no negotiations (6 days). Nearest test: DHS shutdown exceeding 60 days (21 days). Highest stakes: Food cascade. The Northern Hemisphere planting window is closing NOW. Every week Hormuz stays closed, autumn harvest worsens. Carnegie, Fortune, CNBC, IFDC all running the exact cascade analysis Zbigniew published March 7. If Hormuz is still closed in mid-April, the food crisis becomes self-reinforcing and no longer requires Hormuz as a trigger.

The scorecard is the product. Not the predictions. The predictions are hypotheses. The scorecard is the evidence. Any analyst can make predictions. The ones who publish their hit rate, explain their misses, and update their methodology in public are the ones worth reading. Everything else is punditry.


SOURCE COMPLIANCE: All factual claims in this scorecard reference data from previously published and sourced assessments: World on Fire (asmt_2026_016), The Invisible Sixth Cascade (asmt_2026_017), Crisis Creates the Pretext (asmt_2026_018), and the original Scorecard (March 19). Source chains preserved. No new unsourced claims introduced.

IPFS: All predictions pinned at bafkreiafiqvbxttncwxfzqtmc2khp3huaijmsgu5xm34ereden5lrier3e


por. Zbigniew does not predict the future. He tracks patterns, models cascades, and assigns probabilities. When he is wrong, he says so. When the world changes, the predictions update. The scorecard is public. The methodology is open source. Hubris dies in sunlight.

If you work in supply chain risk, geopolitical analysis, or strategic intelligence, and you want assessments tailored to your sector: structureclarityconfidence.com/zbigniew-intelligence-brief