Methodology
# About Outlier Weekly
A weekly research newsletter applying a three-formula synthesis — Poisson distributions, Shannon entropy / KL divergence, and Taleb fat-tail asymmetry — to rare-event markets on Polymarket and Kalshi. Published with the math, the inputs, and the forward-tracked outcome on every call.
Most retail prediction-market traders use one indicator at a time: volume, simple Bayesian updates, raw historical frequency. That works on coin-flip-shaped markets where the price is already close to fair. It misses systematically on rare events — the long-tail outcomes where mispricing is largest.
The mispricing has a structural reason. *”Mediocristan is where we must endure the tyranny of the collective, the routine, the obvious, and the predicted; Extremistan is where we are subjected to the tyranny of the singular, the accidental, the unseen, and the unpredicted.”* [Source: Taleb, 2007, *The Black Swan*, ch. 3]. Prediction markets price rare events using Mediocristan tools. The result is persistent edge if (and only if) you bring Extremistan-aware machinery.
Outlier Weekly is that machinery — a synthesized three-formula system, the components fully open, applied weekly to a live market with the work shown.
The three-formula synthesis is documented in full as a public research concept page:
→ https://cemini23.github.io/outlier-weekly-methodology/
What is in that page: the Python function that combines the three components, the 40/35/25 weighting rationale, a worked numerical example with a $10,000 bankroll, the four operational override gates (RECHECK conditions), and a Dead-Ends section listing the failure modes already learned.
The page is not marketing. It is research-grade documentation. If you click through and decide the methodology does not match your priors, that is fine — the goal of the About page is to surface that decision before you subscribe, not after.
- **One weekly issue** on a live rare-event market. Inputs (market price, historical base rate, time to resolution, kurtosis bucket), three-formula model output, final weighted fraction, position recommendation (or skip), and a Kelly-sized exposure suggestion.
- **One monthly methodology deep-dive.** One formula component per month, drawing from the primary academic sources (Shannon 1948, Taleb 2007, recent peer-reviewed work on entropy-collapse insider signals).
- **A live forward-track sheet.** Every call published, every resolution disclosed. No survivorship bias, no quiet retractions. If a call resolves wrong, it stays on the sheet with a post-mortem.
- **”Sure things.”** No edge is certain. Rare-event trading is positive-EV variance, not omniscience. Position sizes reflect that — quarter-Kelly cap on every call.
- **Backtest screenshots without forward-tracking.** Backtests are notorious for over-fit; the only honest track record is forward-tracked from a public start date. The track sheet starts at Issue 1.
- **Financial advice.** Every issue is research on an open methodology applied to live markets. Readers are responsible for their own due diligence, position sizing, and risk management. Disclosed at the bottom of every page and every issue.
I run an independent OSINT and quant-research workspace — a working library of curated sources, concept syntheses, and methodology pages on prediction-market modeling, rare-event statistics, and quantitative research operations. The workspace is what produces the three-formula synthesis; Outlier Weekly is what publishes the application.
The workspace itself is private (research-in-progress is messy and not always shareable). The methodology pages relevant to Outlier Weekly are public: the three-formula synthesis, the Poisson-edge derivation, the Shannon / KL component, the Taleb fat-tail multiplier, and the quarter-Kelly sizing primitive are all linked from the methodology wiki page above.
I am building this newsletter in public. The track record starts at Issue 1 and accumulates from there — you watch it form rather than being shown a curated version of it.
Each issue follows the same seven-section structure: (1) the mispricing pattern this week, (2) the three formulas applied, (3) the 40/35/25 weighted output, (4) the worked example on the chosen market, (5) the forward-track entry for the call, (6) what is coming in the next two issues, (7) subscribe / share CTA.
If you only have five minutes: sections (4) and (5) are the trade. Section (4) is the math; section (5) is the position size and the resolution date you will be tracking.
If you have twenty minutes: sections (1)–(3) explain *why* the three-formula synthesis catches what single-indicator analysis misses on this specific market, in language calibrated for a reader with intermediate probability background. No proofs, no graduate-level statistics — but no over-simplification either.
Outlier Weekly’s forward-track sheet is published on the same page, updated within 24 hours of each resolution. URL: **[your forward-track page link]**. Every call resolves to one of four outcomes: HIT (call resolved in the predicted direction), MISS (call resolved against), SKIP (recommended no position; outcome logged for calibration), or VOID (market itself voided or canceled before resolution).
The sheet is the sheet. No survivorship bias, no quiet retractions, no “we are still bullish” rewrites. If a methodology produces three MISS calls in a row, that goes on the sheet and the methodology gets a post-mortem — not a marketing repackaging.
Questions on a specific issue: reply to the email or comment on the post — I read and answer every one. Methodology questions go in the comment thread; private trade-specific questions go to email and are answered without a paywall.
*Outlier Weekly* publishes research, not financial advice. Position recommendations are illustrative outputs of an open methodology; readers are responsible for their own due diligence and risk management. Forward-tracked outcomes are published whether positive or negative. No fiduciary relationship is created by subscription.