OSINT-037-strategic-forecast-and-estimate-collection

OSINT-037 Strategic Forecast & Estimate - Collection Map

OSINT-037 Strategic Forecast & Estimate - collection workspace. Central topic = the forecast question set / strategic situation being assessed. Branches = the data-point categories for this strategic forecast. Drop each collected datum as a child node; expand it with the source, date accessed, and what it implies for the relevant forecast question. Paste raw tool output into the node’s Notes. → feeds deliverable OSINT-037.

00 · Collection Plan - PIRs & EEIs

The questions this collection must answer + the essential elements of information (EEI) for each. Tick as satisfied. → drives §3 Key Judgments, §4 Forecasting Questions & PIRs, §8 The Forecast (Calibrated Outcome Set), §10 Signposts & Warning Indicators, §15 Collection Gaps & RFIs.

FQ-1 - Primary forecast question (will [event/outcome] occur by [date]?)

• EEI: current state of the primary driver most influencing this question
• EEI: historical base rate from the reference class (how often did comparable cases resolve this way?)
• EEI: observable confirming signposts present / absent at collection date
• EEI: observable disconfirming signposts present / absent at collection date
• EEI: actor behavior / policy signals bearing on this question

FQ-2 - Second forecast question

• EEI: driver state and trajectory relevant to FQ-2
• EEI: reference-class base rate for FQ-2
• EEI: leading indicators specific to FQ-2

FQ-3 - Third forecast question

• EEI: driver state relevant to FQ-3
• EEI: base rate and comparison cases for FQ-3

PIR-A - Driver intelligence (§6 Strategic Drivers & Trends)

• EEI: current state and measurable trajectory of each key structural driver
• EEI: current state and measurable trajectory of each proximate driver
• EEI: driver interaction (reinforcing, offsetting, conditional) - documented evidence
• EEI: rate of change signals (economic data, polling, indicators, official statistics, events)

PIR-B - Base-rate / reference-class intelligence (§7 Reference Class & Base Rates)

• EEI: identified reference class(es) with n≥5 comparable past cases
• EEI: historical base rate (outcome frequency) within each reference class
• EEI: quality / limits of comparison (analogical validity, sample size, era)
• EEI: source for base-rate data (conflict datasets, election databases, economic series, etc.)

PIR-C - Signpost / warning indicator intelligence (§10 Signposts & Warning Indicators)

• EEI: each signpost from §10 - present / emerging / absent at collection date
• EEI: lead-time estimate for each high-priority signpost
• EEI: new potential signposts surfaced during collection not yet in §10

PIR-D - Alternative outcomes & wildcard intelligence (§9)

• EEI: evidence bearing on the principal alternative outcome
• EEI: signals for any low-probability / high-impact wildcard or discontinuity
• EEI: whether conditions for a full Scenario Analysis (OSINT-038) are warranted

Collection gaps / RFIs (running)

• [gap: base-rate source unavailable] → escalate to OSINT-029 Intelligence Estimate
• [gap: signpost not currently observable] → route to OSINT-040 I&W Program
• [gap: driver state contested] → cross-check against OSINT-026 Country / Regional Study
• [open item]

01 · Drivers & Key Variables

The forces shaping how the situation will develop. Collect evidence for each driver’s current state and trajectory - these anchor the §6 driver table and the §8 probabilities. Distinguish structural (slow-moving) from proximate (fast-moving) drivers. → feeds §6 Strategic Drivers & Trends, §8 The Forecast (Calibrated Outcome Set).

Structural drivers

Long-horizon, slow-moving forces. Collect trend data, official statistics, academic / institutional reporting. Grade each source (A–F/1–6). → §6 driver table, rows with Type = Structural.

• [structural driver - e.g. demographic / economic / institutional / geopolitical trend]
  • Current state (measured):
  • Trajectory (direction + rate): [↑ / → / ↓ + pace]
  • Evidence / data source:
  • Time horizon over which it shapes outcomes:
  • Interactions with other drivers:
  • Forecast relevance: [Dominant / Significant / Marginal]
  • Uncertainty: [High / Med / Low]
  • Confidence: [H/M/L]
  • Source grade (A–F/1–6):
  • Notes / tool output:
• [structural driver 2]

Proximate drivers

Fast-moving, near-term forces. Collect current-event data, policy announcements, market signals, leadership decisions, conflict incident data. → §6 driver table, rows with Type = Proximate.

• [proximate driver - e.g. policy decision, leadership change, conflict escalation, market event]
  • Current state:
  • Trajectory: [↑ / → / ↓]
  • Evidence / event:
  • Interaction with structural drivers:
  • Forecast relevance: [Dominant / Significant / Marginal]
  • Uncertainty: [High / Med / Low]
  • Confidence: [H/M/L]
  • Source grade (A–F/1–6):
  • Notes / tool output:
• [proximate driver 2]

Driver interaction synthesis

• [Reinforcing pair - driver A × driver B]:
• [Offsetting pair - driver A dampens driver B]:
• [Conditional interaction - driver A matters only if driver B exceeds threshold]:

02 · Indicators & Warnings

Observable, monitorable signposts that confirm or disconfirm each forecast path. Each indicator is a separate clonable block. Collect current status, lead-time estimate, and direction of probability change. These seed §10 Signposts & Warning Indicators and hand off to OSINT-040 I&W Program. → feeds §10, Appendix D Signpost & Warning Matrix.

Confirming indicators (would raise central-forecast probability)

→ §10, disposition: Watch / Notify client / Re-forecast. Clone per signpost.

• [signpost - e.g. observable event, metric threshold, policy action, statement]
  • Forecast question it bears on: [FQ-1 / FQ-2 / FQ-3]
  • Probability change if observed: [e.g. FQ-1: 55%→75%]
  • Lead time:
  • Current status: [Not present / Emerging / Present]
  • Source / where to monitor:
  • Disposition: [Watch / Notify client / Re-forecast / Route to OSINT-040 I&W]
  • Notes / tool output:
• [signpost 2]

Disconfirming indicators (would lower central-forecast probability)

→ §10. A disconfirming signpost lowers the probability of the central forecast.

• [signpost - e.g. non-event, reversal, policy contradiction]
  • Forecast question it bears on: [FQ-1 / FQ-2 / FQ-3]
  • Probability change if observed: [e.g. FQ-1: 55%→35%]
  • Lead time:
  • Current status: [Not present / Emerging / Present]
  • Source / where to monitor:
  • Disposition: [Watch / Route to OSINT-040 I&W]
  • Notes / tool output:
• [signpost 2]

Wildcard / discontinuity tripwires

Low-probability, high-impact signals that would invalidate the forecast. → §9 wildcards.

• [tripwire - e.g. exogenous shock, black-swan indicator]
  • Type: [Discontinuity / Black-swan-adjacent / Wildcard]
  • Estimated probability: [<5% / <10%]
  • Impact if triggered: [Crit / High / Mod]
  • Current status: [Absent / Trace signal]
  • Notes:

03 · Scenarios / Hypotheses

The central forecast and its plausible alternatives, for ACH and premortem. Collect evidence that confirms, disconfirms, or is neutral to each hypothesis. → feeds §8 The Forecast, §9 Alternative Outcomes, §12 ACH, §14 Premortem.

Central forecast - most likely outcome

• Forecast question: [FQ-1 / FQ-2 / FQ-3]
• Hypothesis label: [H1 - e.g. Outcome A obtains by [date]]
• Evidence FOR (consistent - C):
  • [evidence item]:
  • [evidence item 2]:
• Evidence AGAINST (inconsistent - I):
  • [evidence item]:
• Neutral evidence (N):
  • [evidence item]:
• Current point probability:
• Range:
• ICD 203 band:
• Analytic confidence: [H/M/L]
• Notes:

Principal alternative outcome

• Hypothesis label: [H2 - e.g. Outcome B obtains / timeline slips / magnitude reduced]
• Evidence FOR (C):
  • [evidence item]:
• Evidence AGAINST (I):
  • [evidence item]:
• Estimated probability:
• Impact if it occurs:
• Leading indicators specific to H2:
• Notes:

Additional alternative / wildcard

• Hypothesis label: [H3 - e.g. discontinuity / exogenous shock]
• Evidence FOR (C):
  • [evidence item]:
• Estimated probability: [e.g. <5%]
• Impact: [Crit / High]
• Notes:

Key Assumptions Check (KAC) items

Assumptions underpinning the forecast - collected evidence for/against each. → §13 KAC.

• [assumption - e.g. reference class is genuinely comparable]
  • Basis / evidence:
  • Confidence: [H/M/L]
  • Impact on forecast if wrong:
  • Linchpin?: [Y/N]
  • Notes:
• [assumption - dominant driver continues current trajectory]
  • Basis / evidence:
  • Confidence: [H/M/L]
  • Impact:
  • Linchpin?: [Y/N]
• [assumption - no major exogenous shock intervenes within horizon]
  • Basis / evidence:
  • Confidence: [H/M/L]
  • Impact:
  • Linchpin?: [Y/N]

Premortem failure modes

Imagine the central forecast was badly wrong at resolution date. → §14 Premortem.

• [failure mode - what most plausibly caused the miss]
  • Assumption / driver that broke:
  • Likely bias: [Anchoring / Confirmation / Base-rate neglect / Recency / Optimism]
  • Corrective / early catch (feeds back to §10):

04 · Data Sources & Feeds

Every material source type - institutional, statistical, media, dataset, human - with cadence, reliability grade, and access route. Clone per source. → feeds §17 Annex A - Sources & Methodology, Appendix H Full Source Register.

Institutional / official sources

National statistics agencies, government portals, international organization databases. Tools: direct agency portal scrape, UN/World Bank/IMF data APIs, Freedom of Information requests.

• [institution / dataset - e.g. UN Comtrade / national statistics office / central bank]
  • Type: [Primary official / Secondary official]
  • Coverage scope:
  • As-of date / cadence:
  • Self-interest / reliability caveat:
  • Reliability (A–F):
  • Credibility (1–6):
  • Access URL / route:
  • Notes / tool output:
• [institution 2]

Conflict / political event datasets

For base-rate and reference-class collection. Tools: ACLED, Uppsala UCDP, GDELT, ICEWS, V-Dem, Polity, PRS Group, Economist Intelligence Unit, Crisis Group, IISS, Jane’s, SIPRI.

• [dataset - e.g. ACLED / UCDP / V-Dem / Polity5 / GDELT]
  • Cadence / currency:
  • Reference class supported:
  • Base rate derivation from this dataset:
  • Reliability (A–F) / Credibility (1–6):
  • Access route / licence:
  • Notes:
• [dataset 2]

Economic / financial data feeds

Tools: World Bank Open Data, IMF WEO/IFS, BIS, central bank portals, Bloomberg/Refinitiv (where licenced), FRED, Trading Economics, Eurostat.

• [feed - e.g. IMF WEO / World Bank Databank / central bank portal]
  • Metric / series:
  • Cadence:
  • Reliability (A–F) / Credibility (1–6):
  • Notes:
• [feed 2]

Open-source news & media feeds

Tools: Feedly, GDELT GKG, Google News API, RSS aggregation, Nexis/Factiva (licenced), OSINT search operators (site:, filetype:, date:).

• [outlet / feed - e.g. Reuters / AP / regional outlet / aggregator]
  • Coverage:
  • Cadence:
  • Self-interest / state-control caveat:
  • Reliability (A–F) / Credibility (1–6):
  • Notes:
• [outlet 2]

Academic / think-tank / institutional reporting

Tools: Google Scholar, JSTOR, SSRN, Brookings, RAND, Chatham House, IISS, ICG, CSIS, SWP, ECFR, CFR, regional area-studies journals.

• [publication / institution]
  • Relevance to forecast:
  • Cadence:
  • Reliability (A–F) / Credibility (1–6):
  • Notes:

Forecasting / prediction aggregators

Tools: Good Judgment Open, Metaculus, Manifold Markets, Polymarket (prices as market signal, not ground truth); Superforecaster consensus as calibration cross-check.

• [platform - e.g. Metaculus / Good Judgment Open / Manifold]
  • Question URL:
  • Current community probability:
  • Resolution criteria:
  • Cadence / last updated:
  • Notes / deviation from internal forecast:
• [platform 2]

Human / in-country sourcing (where commissioned)

Tools: analyst network, in-country correspondents, expert interview, Track-II contacts.

• [source type / designator - never name if sensitive]
  • Reliability (A–F):
  • Credibility (1–6):
  • Basis of reliability assessment:
  • Notes:

05 · Key Actors & Entities

The actors whose decisions, behavior, or capacity most shape the forecast outcomes. Clone per actor. → feeds §6 Strategic Drivers (actor as proximate driver), §9 Alternative Outcomes, §10 Signposts (actor behavior as signpost). ★ recursive block.

State / government actors

Tools: government portals, Hansard / congressional records, official statements, diplomatic cables (public), Wikileaks (lawfully released), FOIA releases; OpenCorporates for state-owned enterprises.

• [actor - e.g. government, ministry, ruling party, central bank, military command]
  • Role in forecast:
  • Current posture / stated position:
  • Trajectory / trend in behavior:
  • Decision levers they hold relevant to FQ-1/2/3:
  • Constraints on their action:
  • Signpost: what actor behavior would confirm forecast path:
  • Signpost: what actor behavior would disconfirm:
  • Source / basis:
  • Confidence: [H/M/L]
  • Notes / tool output:
• [actor 2]

Non-state actors

Opposition groups, armed groups, civil society, major corporations, multilateral bodies, IFIs. Tools: NGO databases, UN Panel of Experts reports, OFAC/EU sanctions lists, company registries.

• [actor - e.g. opposition party, armed faction, IGO, major firm, lobby]
  • Role in forecast:
  • Current posture:
  • Trajectory:
  • Constraints:
  • Signpost (confirming):
  • Signpost (disconfirming):
  • Source / basis:
  • Confidence: [H/M/L]
  • Notes / tool output:
• [actor 2]

External / third-party actors

Foreign states, international institutions, transnational actors whose intervention could move forecast probabilities (wildcards or major factors).

• [actor - e.g. foreign power, multilateral institution, diaspora network]
  • Stake in outcome:
  • Likely posture / intervention vector:
  • Evidence of current activity:
  • Probability impact if they act:
  • Source:
  • Confidence: [H/M/L]
  • Notes:

06 · Events Timeline

Chronological record of events, policy decisions, and milestones that bear on the drivers, signposts, or forecast questions. Clone per event. → feeds §6 Strategic Drivers, §10 Signposts, §11 Key Findings Summary, Appendix G Time-Phasing Chart. ★ recursive block.

Past events (pre-collection date)

Events that establish baseline, reference-class cases, or driver trajectory.

• [event - date + brief description]
  • Date:
  • Description:
  • Bearing on: [driver / signpost / reference class / alternative outcome]
  • Forecast question affected: [FQ-1 / FQ-2 / FQ-3]
  • Significance:
  • Source:
  • Source grade (A–F/1–6):
  • Notes:
• [event 2]

Current-period events (within collection window)

• [event - date + brief description]
  • Date:
  • Description:
  • Bearing on: [driver / signpost / forecast probability]
  • Probability impact: [FQ-1 moves from X% to Y% because…]
  • Source:
  • Source grade (A–F/1–6):
  • Notes:
• [event 2]

Scheduled / anticipated future events (within forecast horizon)

Elections, summits, contract renewals, fiscal years, leadership transitions, legal deadlines. These are potential signposts or resolution-date anchors.

• [anticipated event - expected date + type]
  • Expected date:
  • Type: [Election / Summit / Deadline / Leadership transition / Economic release]
  • Forecast question affected: [FQ-1 / FQ-2 / FQ-3]
  • Scenario implication:
  • Source / basis for timing:
  • Notes:
• [anticipated event 2]

07 · Assumptions & Uncertainties

The analytic and evidentiary assumptions underpinning the forecast, and the irreducible uncertainties that widen the probability range. → feeds §5 Forecast Framing & Methodology, §13 Key Assumptions Check, §14 Premortem / Red-Team Check, §17 Annex A methodology statement.

Reference-class assumptions

• [assumption - e.g. reference class genuinely comparable to this case]
  • Basis / evidence:
  • Confidence: [H/M/L]
  • Impact on forecast if wrong:
  • Linchpin?: [Y/N]
  • Notes:

Driver-continuity assumptions

• [assumption - e.g. dominant driver continues current trajectory through horizon]
  • Basis:
  • Confidence: [H/M/L]
  • Impact if wrong:
  • Linchpin?: [Y/N]
  • Notes:

Exogenous-shock assumptions

• [assumption - e.g. no major exogenous shock within horizon]
  • Basis:
  • Confidence: [H/M/L]
  • Impact if wrong:
  • Linchpin?: [Y/N]
  • Notes:

Data / base-rate quality limitations

• [limitation - e.g. reference class small, base rate fragile]
  • Nature of limitation:
  • Effect on range (does it widen the range?):
  • Mitigation:
  • Notes:

Reflexivity / model uncertainty

• [reflexivity concern - e.g. forecast publication or client action could alter outcome]
  • Nature:
  • Assessed probability that reflexivity materially distorts:
  • Mitigation:

Bias audit

Structured check before finalising §8 probabilities.

• Anchoring: [Were initial probabilities adjusted sufficiently away from the anchor?]:
• Confirmation bias: [Was disconfirming evidence genuinely weighed?]:
• Base-rate neglect: [Was the outside view applied before the inside view?]:
• Recency bias: [Were current vivid events over-weighted vs. base rate?]:
• Optimism / pessimism: [Was the range widened appropriately for uncertainty?]:

99 · Collection Admin

Working register - audit trail behind the deliverable. Not a deliverable section.

Source register

Every material datum traceable to a graded source. Admiralty two-axis code (A–F / 1–6).

• [S-1 - source name / designator]
  • Type: [Primary official / Secondary / Tertiary / Dataset / HUMINT]
  • Reliability (A–F):
  • Credibility (1–6):
  • Date accessed:
  • Coverage scope:
  • Self-interest / caveat:
• [S-2]
• [S-3]

Evidence archive

• [capture ref + hash + URL + timestamp + description]:

Forecast question register (working)

Mirrors §4 / Appendix A in the deliverable. Track current probabilities here during collection.

• [FQ-1 - resolvable question text]
  • Resolution criteria:
  • Resolution date:
  • Current probability (working):
  • Range:
  • ICD 203 band:
  • Analytic confidence: [H/M/L]
  • Last updated:
• [FQ-2]
• [FQ-3]

Open gaps / RFIs (running)

• [base-rate source gap - pending] → OSINT-029 Intelligence Estimate
• [signpost not observable - standing watch needed] → OSINT-040 I&W Program
• [driver state contested - corroboration needed] → OSINT-026 Country / Regional Study
• [full alternative-futures set warranted] → OSINT-038 Scenario / Alternative Futures Analysis
• [weak-signal sweep needed] → OSINT-039 Horizon Scan / Emerging Threats Report
• [live tracking / scoring surface needed] → OSINT-041 Probabilistic Forecasting Dashboard
• [item submitted, not yet returned]