Forecasting Scaffold

A scaffold for quantitative forecasting and prediction tasks, designed to produce accurate numerical predictions with calibrated uncertainty.

heuristic-algebra forecasting-scaffold quantitative-prediction LLM
concept Updated 2026-04-12

Forecasting Scaffold (FS)

A scaffold for quantitative forecasting and prediction tasks, designed to produce accurate numerical predictions with calibrated uncertainty.

Origin: Forecasting campaigns F1-F4. Designed from practitioner domain knowledge for quantitative prediction tasks.

Type: Program scaffold (structure + content). Use alone — do not combine with additional heuristics, as they degrade performance.

Use: Numerical forecasting, market predictions, estimating future values of measurable variables. Use alone — do not combine with additional heuristics, as they degrade performance. Works best with actual data included in the prompt.

The Eight Principles

  1. Base Rate First — Start with historical base rate and distribution. Anchor to data, not narrative.
  2. Trend vs Mean Reversion — Choose Explicitly — State expectation and reasoning for each variable.
  3. Asymmetric Risk — Identify skewed risks and reflect in forecast, not just widen range symmetrically.
  4. Avoid Conservative Bias — Calibrate confidence correctly; “intellectual humility” doesn’t mean forecasting the mean.
  5. Quantify — Every forecast must be a specific number; express uncertainty through ranges after point estimate.
  6. Falsifiable — State what would prove the forecast wrong.
  7. Regime Awareness — Identify current regime (expansion, contraction, transition, shock).
  8. Update on Evidence — Weight recent data in regime changes, base rates in stable regimes.

The Full Prompt

You are a quantitative forecaster. Your goal is to produce the most ACCURATE numerical predictions possible, not the most hedged or cautious ones.

FORECASTING PRINCIPLES (apply to every forecast):

1. BASE RATE FIRST: Start with the historical base rate and distribution. What has this variable typically done? What is the recent trend? Anchor to data, not narrative.

2. TREND vs MEAN REVERSION — CHOOSE EXPLICITLY: For each variable, state whether you expect trend continuation or mean reversion, and WHY. Do not default to mean reversion out of caution — trending variables (rates in hiking cycles, assets in momentum) often continue longer than expected.

3. ASYMMETRIC RISK: Identify which direction the risks are skewed. If you're uncertain, the forecast should reflect the SKEW, not just widen the range symmetrically.

4. AVOID CONSERVATIVE BIAS: "Intellectual humility" does NOT mean forecasting the mean. It means calibrating your confidence correctly. A forecast of "roughly where it is now" is a specific prediction (no change) and is often WRONG for trending variables.

5. QUANTIFY: Every forecast must be a specific number. Express uncertainty through ranges AFTER committing to a point estimate. Do not hide behind qualitative language.

6. FALSIFIABLE: State what would prove your forecast wrong. If nothing could prove it wrong, it's not a forecast.

7. REGIME AWARENESS: Identify the current regime (expansion, contraction, transition, shock). Forecasts that ignore regime context systematically underperform.

8. UPDATE ON EVIDENCE: Weight recent data more heavily when there are signs of regime change. Weight base rates more heavily in stable regimes.

Example Application: 2026 Iran War Analysis

Evaluation of the 2026 Iran War under the Lucid Judgment (LJ) Framework
As of 12 April 2026

The 2026 Iran war commenced on 28 February 2026 with joint United States and Israeli airstrikes targeting Iranian leadership (including the assassination of Supreme Leader Ali Khamenei), nuclear facilities, ballistic missile infrastructure, and military sites. Iran responded with missile and drone barrages against Israel, U.S. regional allies, and Gulf states, while asserting control over the Strait of Hormuz through mining, toll imposition (in Chinese yuan), and threats to shipping. A two-week ceasefire, brokered by Pakistan and effective 8 April, required Iran to reopen the strait. However, as of 12 April (day five of the ceasefire), U.S.-Iran face-to-face talks in Islamabad have collapsed after 21 hours without agreement. President Trump has announced an immediate U.S. naval blockade of Iranian ports and the Strait of Hormuz, with enforcement beginning in the coming days; he has described U.S. forces as “locked and loaded” to “finish” any Iranian interference. No direct U.S.-Iran or U.S.-Israel exchanges of fire have occurred since approximately 8 April, though Israeli operations against Hezbollah in Lebanon continue and strain the truce. Iran’s conventional military capabilities have been severely degraded (air defenses, missile production, and naval assets largely destroyed).

LJ6 (Epistemic Collapse Detection): The assessment passes the preliminary system check. Beliefs are updating with fresh evidence from multiple independent sources (Institute for the Study of War, Al Jazeera, Reuters, Wikipedia summaries, and contemporaneous reporting). Evidence streams are not highly correlated; Western, Gulf, and Iranian-affiliated outlets converge on core facts (ceasefire fragility, failed talks, blockade announcement) despite differing emphases. Confidence tracks evidence quality rather than emotional or partisan conviction.

LJ1 (Evidence-Action Proportionality): The decision informed here is informational—providing a calibrated forecast for situational awareness rather than an irreversible high-stakes action (e.g., investment or policy commitment). Evidence standards are therefore set at a moderate threshold: recent, multi-source reporting suffices, with higher scrutiny reserved for claims of imminent full-scale resumption.

LJ3 (Asymmetric Disconfirmation): Confirming evidence (ceasefire holding for five days; U.S. statements of military objectives largely achieved) is deliberately down-weighted relative to disconfirming indicators (talks collapse, blockade implementation, ongoing Lebanon operations, Iranian warnings of “deadly whirlpools” in the strait, and proxy activity). The steel-manned opposing view—that Iran retains asymmetric leverage via proxies, residual mines, and global oil sensitivity, potentially forcing a U.S. climb-down—is acknowledged; however, Iran’s documented degradation and the U.S. naval superiority in the Gulf render this view weaker than it might appear at first glance.

LJ2 (Calibrated Confidence): All probabilistic forecasts below are explicitly scored and will be tracked against outcomes.

LJ4 (Thinking-Acting Threshold): Stakes are elevated (potential oil-price shocks, regional spillover), but the situation remains partially reversible (blockade can be lifted; diplomacy remains open). Delay costs are moderate (markets already pricing in risk), and marginal information value is declining rapidly given consistent reporting. The threshold for decisive forecasting has therefore been crossed.

LJ5 (Action-Inquiry Cycle): The present analysis enters the inquiry phase (“What is true?”) post-ceasefire initiation and talk failure, prior to any further action recommendation.

LJ7 (Negation Portfolio): No bounded negation of the axioms is required here; the context does not involve pure research, crisis command, or trusted institutional adjudication.

Current Assessment
The ceasefire is fragile but intact in direct U.S.-Iran terms. The primary flashpoint is the Strait of Hormuz, where U.S. blockade enforcement risks tactical incidents. Iran’s military posture is defensive and asymmetric; its leadership (now under Mojtaba Khamenei) appears focused on survival rather than escalation. Israel’s separate Lebanon campaign introduces linkage risk but does not yet breach the core truce. Global economic effects (rising oil prices) exert pressure toward de-escalation on all sides.

Forecasts with Calibrated Confidence (LJ2)

  • +7 days (19 April 2026): U.S. blockade will be actively enforced, with Iran likely responding via proxy harassment, mine-laying, or rhetorical threats rather than direct naval confrontation. Diplomatic channels (possibly via Oman, Pakistan, or China) may reopen under economic pressure. Probability of major direct kinetic escalation (U.S./Israel strikes on Iranian territory or Iranian missile response): 25 % (±10 %). Probability of sustained tense standoff with limited incidents: 65 %. Base case favors containment; disconfirming evidence (e.g., Iranian mine detonation on U.S. vessel) would update this upward sharply.

  • +14 days (26 April 2026): The original two-week ceasefire window expires around 22 April. Expect either a short extension or a controlled resumption of limited strikes if the blockade remains contested. Iran’s degraded capabilities reduce its capacity for sustained retaliation; U.S. objectives (nuclear/missile program degradation) are largely met. Probability of full-scale war resumption: 40 % (±15 %). Probability of negotiated pause or phased de-escalation: 50 %. Steel-manned alternative (prolonged Hormuz crisis driving oil to $150+/bbl and forcing U.S. concessions) is monitored but currently under-weighted by evidence of Iranian weakness.

  • +28 days (10 May 2026): Longer-term trajectory points toward a frozen or managed conflict rather than decisive victory or regime change. Iran will likely seek sanctions relief and proxy preservation; the U.S. will prioritize freedom of navigation and non-proliferation verification. Probability of sustained low-level hostilities (proxy/ naval incidents) without full war: 55 %. Probability of comprehensive diplomatic framework (possibly including IAEA inspections): 35 %. Probability of major escalation: 10 %. These estimates reflect diminishing marginal returns to further analysis absent new evidence and will be recalibrated upon any material change (LJ5 cycle).

In summary, the war has entered a high-tension pause characterized by U.S. coercive diplomacy via blockade rather than active combat. Evidence-action proportionality and asymmetric disconfirmation support cautious optimism for managed de-escalation, tempered by calibrated recognition of escalation risks around the strait and ceasefire expiration. Updates will follow the action-inquiry cycle as new data emerge.

See Also