Tools chosen: kalshi_direct kalshi_related claude_news gdelt_news wikipedia polymarket_related code_execution

Sub-questions (Fermi decomposition)

Planner reasoning

Oil has been the largest single source of global primary energy for decades (~30-32% share vs coal ~26% and gas ~23% in the Energy Institute Statistical Review), so the base case strongly favors Yes; the key uncertainties are the resolution source/accounting method (substitution vs direct-equivalent for renewables) and whether coal or renewables could plausibly overtake oil by 2030. Research should anchor on the Kalshi price for all sibling outcomes in the event, then verify current shares and projected 2030 shares from IEA/EI/EIA outlooks.

# Current state Kalshi's Oil-largest-source market prices YES at 67% (up from ~59% a month ago, down slightly over 7 days). All major primary-energy datasets and forecasts (Energy Institute Statistical Review, IEA WEO 2025, OPEC WOO 2025) show oil is currently the largest primary energy source by a substantial margin and is expected to remain so through 2030, though the market price implies more uncertainty than trend-extrapolation models suggest. # Timeline of key events - 2023 (confirmed): Energy Institute baseline — Oil 196 EJ, Coal 164 EJ, Gas 145 EJ, oil-coal gap ~32 EJ. - 2025-06 (confirmed): Statistical Review of World Energy 2025 (covering 2024 data) — global energy 592 EJ; Oil 199 EJ (33.6%), Coal 165 EJ (27.9%), Gas 149 EJ (25.2%); methodology shifted to "direct equivalent" for renewables/nuclear (deflates their EJ contribution vs. old substitution method). - 2025 (confirmed): IEA WEO 2025 — STEPS scenario projects oil demand peaking ~102 mb/d around 2030, then flattening/declining; coal peaks slightly earlier (before end of decade); gas continues growing to 2035. - 2025 (reported): Carbon Brief analysis of IEA WEO 2025 — renewables projected to overtake oil as largest energy source only in the "early 2040s," not by 2030. - 2025 (reported): OPEC WOO 2025 — most bullish on oil, projecting it retains largest share even through 2050 (just below 30% share), with oil+gas combined staying above 50%. - 2026-07 (confirmed): Statistical Review 2026 (2025 data) — fossil fuels still 86% of total energy supply; renewables 5.9%, hydro 2.7%, nuclear 5.2% (direct-equivalent basis), reinforcing oil's continued dominance. - 2026 (reported, GDELT): scattered news of "clean power as largest source of new energy additions in 2025" — refers to new capacity/generation growth, NOT overtaking oil in total primary energy stock. # Event Will Oil be the single largest source of global primary energy consumption in 2030? (Kalshi: KXPRIMEENGCONSUMPTION-30-OIL) # Outcomes to forecast Yes / No (Oil is/is not the largest primary energy source in 2030) # Kalshi market anchor **YES (Oil) = 67%** (as of latest data). 7-day change: -1pp; 30-day change: +8pp (trending up). Range over 89 days: 47%–68%. Avg daily volume: 98 contracts (thin/low liquidity). This is the primary consensus to beat. # Sub-question answers 1. **Kalshi prices for each outcome** — Only Oil's own YES price (67%) is directly observed; sibling Coal/Gas/Renewables/Nuclear markets in the KXPRIMEENGCONSUMPTION-30 series were not found in Kalshi-direct data (kalshi_related found 0 series matches). Code-execution tool used illustrative (not live) sibling prices implying de-vigged Oil ≈73%, Gas ≈13%, Coal ≈6% — not verified live data. 2. **Current actual shares** — Energy Institute Statistical Review 2025 (2024 data): Oil 33.6% (199 EJ), Coal 27.9% (165 EJ), Gas 25.2% (149 EJ); oil-coal gap ≈34 EJ. [Statistical Review 2025] 3. **Accounting convention** — Methodology shifted from substitution method to "direct equivalent" for renewables/nuclear starting with the 2024 Statistical Review, which *reduces* renewables' counted EJ share relative to the old method — this favors oil retaining top rank under Energy Institute data. [dieselnet.com] 4. **2030 projections** — IEA WEO2025 STEPS: oil peaks ~102 mb/d around 2030 then flattens/declines; coal peaks slightly before 2030; gas grows to 2035; renewables overtake oil only in the "early 2040s." OPEC WOO2025: oil retains largest share through 2050. [IEA, Carbon Brief, OPEC] 5. **Oil demand plateau/decline** — IEA STEPS projects oil demand plateauing around 2030 (~102 mb/d peak), a much gentler decline trajectory than coal, which is already near/at peak and declining faster in relative terms; gas and renewables continue growing through 2030 and beyond. [IEA WEO2025] 6. **Any forecast of overtake by 2030** — No credible source found projecting coal, gas, or renewables overtaking oil by 2030; the earliest credible overtake date cited (renewables) is the early 2040s per IEA/Carbon Brief. OPEC sees no overtake through 2050. [Carbon Brief, OPEC WOO] # Key facts (high-confidence, factual) 1. [Statistical Review 2025] 2024: Oil 199 EJ (33.6%) > Coal 165 EJ (27.9%) > Gas 149 EJ (25.2%); gap oil-coal ≈34 EJ. 2. [Statistical Review 2026] 2025: fossil fuels 86% of total energy supply; oil remains dominant. 3. [IEA WEO2025] Oil demand plateaus ~2030 (~102 mb/d peak) under STEPS. 4. [Carbon Brief/IEA] Renewables overtake oil only in "early 2040s," not 2030. 5. [OPEC WOO2025] Oil retains largest share of energy mix through 2050 (~30% share). 6. [code_execution Monte Carlo] Using historical growth-rate ranges, oil remains largest in 100% of 200,000 simulated trajectories to 2030; even in an extreme "oil bear/clean bull" stress scenario, oil still finishes ahead (by ~13 EJ margin vs. coal). # Cross-market signals - Kalshi related: No live sibling markets (Coal/Gas/Renewables) found for direct triangulation; series appears sparsely populated. - Polymarket: No matching markets found (0/100 scanned). - Sportsbook implied: N/A (not applicable to this event type). # Analyst opinions and speculation - code_execution's illustrative de-vig exercise (not live data) suggested market-implied P(oil not largest) ≈27%, notably higher than trend-extrapolation's ~0%, implying markets price meaningful tail risk (policy shocks, faster EV/renewables adoption, data/methodology revisions) — but this was demonstrative, not observed live pricing. - Real Kalshi YES=67% is far below both the naive trend-model near-certainty and the demonstrative de-vig figure, suggesting actual market participants harbor substantial uncertainty, possibly due to resolution-criteria ambiguity, thin liquidity, or awareness of methodology/accounting risk (e.g., which primary energy convention resolution source uses). # Directional lean per outcome - **YES (Oil largest in 2030):** Strongly supported — current oil lead is large (~34 EJ over coal, ~50 EJ over gas), all IEA/OPEC scenarios keep oil on top through 2030 and beyond, and quantitative extrapolation makes an overtake essentially implausible within 7 years given renewables' small base. - **NO (Oil not largest):** Weak support — relies on the 67% Kalshi price being systematically too low, or a black-swan scenario (rapid EV/renewables surge + oil demand shock) plus possible favorable-for-non-oil methodology in final resolution source, none of which is currently evidenced. # Gaps / unknowns - Exact resolution methodology/data source Kalshi will use (Energy Institute vs IEA vs EIA; substitution vs direct-equivalent) is unspecified in rules — could matter at the margins if oil's lead narrows further, though current gap is large enough to be robust either way. - No live sibling market data for Coal/Gas/Renewables/Nuclear to cross-check implied probabilities. - Why Kalshi YES sits at 67% rather than closer to ~90%+ implied by trend data is unexplained — possibly reflects genuine model uncertainty, thin/illiquid market pricing, or unaccounted resolution ambiguity. # Calibration anchors - Kalshi current YES price: 67% (anchor). - Precedent: Oil has been the largest primary energy source continuously per Statistical Review data for decades; margin over #2 (coal/gas) remains ~30+ EJ (~15-20% relative gap) as of 2024-2025 data, with no credible mainstream forecast (IEA, OPEC, EIA) projecting an overtake before the early 2040s at earliest.

1. Both forecasts converge on identical 0.90/0.10 despite claiming independent reasoning, and neither seriously grapples with why Kalshi sits at 67% (up 8pp in 30 days) rather than near-certainty — Forecast 1's "time-value discount" explanation (0.92/1.05^6≈0.69) is a post-hoc rationalization not evidenced in the brief and doesn't actually explain a 67% price on a binary event (time discounting on a YES/NO contract doesn't work this way; it would depress both YES and NO prices symmetrically, not create a specific 67/33 split). 2. Neither forecaster addresses the resolution-source ambiguity with concrete probability weight: the brief explicitly flags "unspecified" methodology (Energy Institute vs IEA vs EIA, substitution vs direct-equivalent) as a real gap, yet both treat this as a minor residual rather than modeling scenarios where a substitution-method source could make renewables/hydro/nuclear collectively rank oil lower — this is glossed over rather than quantified. 3. Both anchor exclusively on trend-extrapolation/Monte Carlo (100% oil-largest in 200,000 sims) but underweight the illustrative de-vig finding that market-implied P(no)≈27%, which — even if "not observed live" — suggests sophisticated traders may see resolution/tail risks the brief's fundamental data doesn't capture; dismissing this entirely as "demonstrative, not observed" may be too quick given it's the only cross-market signal available. 4. Neither forecast considers a scenario where "largest source" could plausibly be contested not by renewables but by a coal resurgence in a bifurcated world (e.g., Asian coal growth outpacing IEA STEPS assumptions) — the brief notes coal peaking "slightly before 2030" but a delayed coal peak combined with oil demand plateau could theoretically narrow the gap faster than modeled, and this alternative pathway to NO is unexamined. 5. Both hold 10% for "black-swan" and "methodology risk" but never decompose this residual — is it 5% methodology + 5% demand shock, or all methodology? Given the brief's Monte Carlo shows ~0% overtake probability under demand-growth uncertainty alone, nearly all of the 10% NO should logically be attributed to resolution/methodology ambiguity, and this asymmetry deserves explicit acknowledgment since it changes how one should weight the unresolved Kalshi rules-source question. 6. Neither forecaster explores whether the 67% price and its 8pp monthly rise might reflect informed trading anticipating a specific rules clarification (e.g., Kalshi confirming its data source imminently), which would be a more parsimonious explanation than generic thin-liquidity noise, and neither flags the thin volume (98 contracts/day) as a reason to discount the market signal more heavily than they do (both still net out to trusting fundamentals ~23pp above market, an identical adjustment that looks more like anchoring on each other's plausible answer than independent convergence).

90% (75% confidence)

Yes: 90% No: 10%

Fundamentals are unusually one-sided: oil led coal by ~34 EJ and gas by ~50 EJ in 2024-25 Energy Institute data, and no mainstream projection (IEA STEPS, OPEC WOO, EIA) has any source overtaking oil before the early 2040s; the direct-equivalent convention further suppresses renewables' counted EJ. Even the critique's alternative NO pathways are weak — a coal resurgence closing a 34 EJ gap in five years while coal is at/near peak is implausible, and under a substitution-method source renewables aggregate still falls well short of ~190+ EJ oil by 2030. The Kalshi 67% anchor is thin (98 contracts/day) and resolves ~6 years out, so capital-lockup/opportunity-cost and inattention plausibly explain much of the discount, though the specific 67/33 split is not fully explained by that mechanism, which justifies not going to 95%+. Consistent with the critique's point 5, I attribute nearly all of the residual 10% NO to resolution-source/methodology ambiguity and definitional surprises rather than to genuine demand-trajectory risk. Final: Yes 0.90, materially above market but short of the near-certainty implied by pure trend extrapolation.

Total pipeline time: 300.8s