Our Methodology
Five analysts. Five frameworks. One obsession: turning uncertainty into something you can read, challenge, and verify.
Who We Are
FUTARCHY.media was founded in 2024 by a small group of forecasting obsessives — five people who spent years watching prediction markets, reading intelligence reports, and arguing about base rates over late-night spreadsheets.
We come from different worlds. Former traders. Political science researchers. A sportswriter who got hooked on Bayesian updating. An engineer who spent three years tracking regulatory timelines across 40 jurisdictions. A journalist who covered three elections and realized that narrative journalism without probabilities was intellectually dishonest.
What we share is a conviction: the public deserves the same calibrated, probabilistic analysis that hedge funds and intelligence agencies take for granted. Not dumbed down. Not behind a paywall. Not dressed up with false certainty.
Every article on this site is researched, written, reviewed, and fact-checked by our team. We argue about weights, challenge each other's assumptions, and reject drafts that don't meet our editorial standards. The methodology below is the system we built for ourselves — and the contract we offer to our readers.
What We Do
FUTARCHY.media produces data-driven forecasts on politics, finance, technology, sports, and world affairs. Every article assigns a probability to a specific, falsifiable claim — and explains exactly how we got there.
We don't predict the future. We estimate its likelihood, show our reasoning, and tell you when we'll know if we were right.
Our editorial posture is simple: here's what the data shows, here's our reading, decide for yourself. We are a transparent aggregator, not an oracle. Every forecast includes the assumptions that built it, the data that supports it, and the conditions that would break it.
The Dossier Format
Every FUTARCHY article is structured as a prospective dossier — a layered document designed for three reading speeds. The format draws from the intelligence community's "Bottom Line Up Front" doctrine (DIA Style Manual, 4th Ed.) and Tetlock's work on structured analytical techniques.
The probability, the claim, and the confidence interval. Visible at the top of every article before you scroll. You know what we think before you read a single paragraph.
Key findings, the critical metrics, scenario cards showing possible outcomes, and a stress test that tells you what would change our mind. Enough for a decision-maker who needs the essentials.
The complete analysis in four chapters: Context, Analysis, Reading, and Resolution. For the reader who wants to verify every step of our reasoning.
We borrowed this structure from intelligence briefings, data journalism, and prediction market interfaces. The goal is never to trap you into reading 2,000 words. It's to give you the right amount of information at whatever depth you need.
Five Analytical Frameworks
Each analyst works within one of five frameworks, developed internally and refined across hundreds of forecasts. The framework is determined by the subject category. Each embodies a distinct analytical tradition — from game theory to sports sabermetrics.
Structures geopolitical probability assessments using a multi-actor analysis grid inspired by Allison's Essence of Decision (1971) and modern structured analytic techniques. Typically weights opposition resilience, resource alignment, coalition dynamics, and leadership calculus. The exact components and weights vary by subject.
Will the EU enforce its carbon border tax on schedule? Will Iran's Strait of Hormuz disruption persist through summer?
Analyzes financial predictions through market data and pricing signals, historical precedent, fundamental analysis, and sentiment. Rooted in the efficient market hypothesis debate — Fama vs. Shiller — and the conviction that capital at risk reveals more honest probabilities than commentary without skin in the game.
Will SpaceX close above $1.5T on IPO day? Will Bitcoin's Q2 recovery break its April base rate?
Evaluates competitive outcomes by weighting form, momentum, home advantage, squad depth, and historical precedent. Influenced by sabermetrics and the Elo rating tradition — designed for one-off events where small margins and variance play an outsized role.
Will Arizona win the NCAA championship? Can PARIVISION dominate PGL Bucharest without Tier-1 experience?
Blends regulatory readiness, political momentum, industry pressure, and technical standards progress. Draws from the Collingridge dilemma and regulatory impact assessment methodology — built for questions where policy, technology, and institutional dynamics intersect.
Can 24 states save America's climate rulebook? Will Europe's carbon border tax survive retaliation?
Tracks cultural momentum, public sentiment, media cycles, and historical base rates. Applies information cascade theory and social proof dynamics — for questions where virality and timing matter more than structural analysis.
How We Build a Probability
Every forecast follows the same six-step construction method, regardless of framework. Our approach is grounded in the structured analytic techniques formalized by Richards Heuer (Psychology of Intelligence Analysis, CIA, 1999) and the calibration research of Philip Tetlock (Superforecasting, 2015).
What Our Probabilities Mean
A 65% probability means: if we made 100 forecasts at 65% confidence, we'd expect roughly 65 of them to resolve YES and 35 to resolve NO. A well-calibrated forecaster at 65% is wrong 35% of the time — and that's not a failure. That's the point.
We express all probabilities in multiples of 5. We don't write 62.4% because that level of precision is dishonest for the kind of analysis we do. The difference between 60% and 65% is meaningful. The difference between 62.4% and 63.1% is noise. This is consistent with Tetlock's finding that even top-performing superforecasters gain marginal returns from granularity beyond 5-point increments.
Every probability comes with a confidence interval — a numerical range (e.g., 55%-75%) that captures our uncertainty about the estimate itself. A narrow interval means we're relatively sure about our probability. A wide one means the underlying situation is volatile.
Scenarios and Resolution
Every dossier presents at least three scenarios that sum to 100%. This isn't a rhetorical trick — it's a forcing function borrowed from scenario planning methodology (Schwartz, The Art of the Long View, 1991). If you can't make your scenarios sum to 100%, you haven't thought through all the possibilities.
Resolution criteria are non-negotiable. Every forecast has a specific date and measurable conditions. When that date arrives, the forecast resolves YES or NO based on observable facts — not our interpretation of events, not a revised reading, not "well, it sort of happened." Binary resolution keeps us honest.
The Team
FUTARCHY's forecasts are researched and written by five analysts, each with distinct expertise, background, and a recognizable analytical voice. We write under pen names — not to hide, but because the analysis should be judged on its reasoning, not its author's credentials.
Former derivatives trader turned forecasting obsessive. His analysis tends toward market microstructure — liquidation cascades, order flow, futures curves. He reads positioning data the way most people read headlines. Believes that price is the most honest opinion poll ever invented.
International relations background with a focus on game theory and multi-actor strategic dynamics. She's most comfortable when the question involves multiple actors making simultaneous calculations under uncertainty. Spent two years tracking EU regulatory timelines before joining the team.
The most conversational of the five — poker analogies, momentum reads, the texture of a specific match or moment. Started in sports journalism, pivoted to quantitative analysis after realizing that narratives without base rates were just stories. Skeptical of models that ignore variance in single-game outcomes.
Systems thinker who tracks feedback loops, regulatory timelines, and the gap between what institutions promise and what they deliver. Studied environmental policy and got frustrated by how rarely forecasts accounted for implementation lag. The most likely to tell you what she got wrong last time and why.
The team's resident contrarian. Covers through the lens of underdogs and base rates, hunting for spots where consensus is wrong and historical precedent suggests a different outcome. Joined after building a personal tracker of 500+ prediction market positions over two years.
Editorial Standards
The DATA - READING - CAVEAT Pattern
Every section of every article follows the same three-beat structure. First, present the data — the numbers, the facts, the observable evidence. Then, offer our reading — what we think it means and why. Finally, the caveat — what could make us wrong, what we're not sure about, what the data doesn't capture.
This pattern forces intellectual honesty into every paragraph. It's easy to write a confident analysis. It's harder — and more useful — to tell the reader where your confidence breaks down.
Honest Precision
We calibrate the precision of our claims to the precision of our knowledge. Probabilities come in multiples of 5. Confidence intervals are always numerical, never vague labels alone. We write "Medium-High (60%-80%)" — not just "Medium-High."
We never invent acronym expansions for our frameworks. ORACLE is ORACLE. Our frameworks are structured editorial methods with names — not proprietary algorithms. There's no black box. You see every weight, every score, every assumption.
Sources and Verification
Every forecast cites a minimum of 8 sources with names and dates. Claims are fact-checked against 2+ independent sources before publication. Every claim is classified as confirmed, rumored, or speculative — and the article's tone adjusts accordingly. We prioritize primary sources (official filings, live data feeds, court documents) over secondary reporting.
Peer Review
No article is published by the analyst who wrote it. Every dossier goes through internal peer review — a second analyst reads the full draft, challenges the weights, and flags reasoning gaps. We disagree on record. When the reviewing analyst's probability estimate diverges from the author's, we note it in the article's stress test section.
The Narrative Arc
Our articles are structured as analytical narratives, not research reports. This is a deliberate editorial choice, informed by the conviction that good analysis requires more than data — it requires storytelling discipline. Every dossier includes three narrative elements that serve both readability and intellectual honesty.
One passage per article that departs from the main thread to explore a tangent — a historical parallel, an unexpected data point, a connection to another domain. Good analysis makes unexpected links. We leave them in.
One moment per article where the analyst argues against their own thesis. This forces the analyst to take the opposing view seriously and shows the reader where the argument is weakest. Inspired by Heuer's "Analysis of Competing Hypotheses."
Every article closes without a neat bow. The data says one thing; the analyst's instinct says another. We don't wrap up. We stay with the uncertainty. That's what makes a forecast useful.
The Editorial Process
Every article passes through a three-phase editorial process before publication. No shortcuts, no exceptions.
The assigned analyst identifies the question, gathers sources, builds the component model, and writes the full dossier. Text quality comes before everything else — no data visualizations, no widgets, no formatting at this stage. Just the argument.
A second analyst reviews the draft independently — challenging assumptions, verifying sources, and checking the math on weighted composites. Separately, all factual claims are verified against primary sources. If the reviewer's own probability estimate differs by more than 10 points, the dossier goes back for revision or the disagreement is documented.
Final formatting, data visualization, and a last editorial pass. The article goes live only after both quality gates are cleared. No partial publications, no draft states visible to readers. Once published, the forecast is locked — we don't retroactively adjust probabilities.
Calibration and Accountability
We will get forecasts wrong. A 70% probability means 30% of the time, the thing doesn't happen. That's not a bug in our methodology — it's the entire point. A forecaster who's right 100% of the time is either lying about their confidence or only predicting obvious things.
What we commit to is calibration — that over time, our 60% forecasts resolve YES about 60% of the time, our 80% forecasts about 80% of the time. We measure this using Brier scores, the standard metric in forecasting research (Brier, 1950), and we'll publish our calibration curves as our track record grows.
Every resolved forecast stays on the site with its original probability, so readers can evaluate our record themselves. We don't delete wrong predictions. We don't rewrite history. The whole point of probabilistic forecasting is that you can score it — and we intend to be scored.
If you disagree with a weight, a score, or an assumption, you have everything you need to recalculate the probability yourself. That's the point. We don't ask you to trust us. We ask you to check our work.
Intellectual Foundations
- Philip Tetlock, Superforecasting (2015): Calibration, base rates, and the evidence that ordinary people can outperform intelligence analysts when they follow structured methods.
- Richards Heuer, Psychology of Intelligence Analysis (1999): Cognitive biases, Analysis of Competing Hypotheses, and the architecture of structured analytic techniques we adapted for our frameworks.
- Nate Silver and the Silver Bulletin: Data first, opinion is calibrated, humility is a feature — not a weakness.
- Prediction markets (Polymarket, Kalshi, Metaculus): People with skin in the game produce more honest probability estimates than people with opinions.
- Intelligence community tradecraft (DIA Style Manual, 4th Ed.): BLUF structure — the conclusion leads, the evidence follows. Probability language standards.
- Data journalism (FiveThirtyEight, Bloomberg Graphics, The Economist): Show your work, use multiple scenarios, make probability legible.
We combined these influences into something specific to FUTARCHY: a forecast that reads like journalism, structures itself like an intelligence brief, resolves like a prediction market contract, and holds itself accountable like a scientific hypothesis.