Iptv Sports Channels
IPTV UK 2025: Watch Every Premier League Match Live and Slash Your Monthly TV Bill If you are one of those households that quietly…
Read full story → Est. 2016 • Fat-tail research & culture • Vol.09
THE WORLD IS HEAVY-TAILED DEAL WITH IT.
Welcome to HeavyTails.com — a field guide to extremes. Why one client beats 10,000, why markets crash, why memes explode, why climate breaks records. Gaussian thinking is comfortable. Reality lives in the tails.
80/20 → 95/5Pareto on steroids
α ≈ 1.16Typical wealth tail
∞ varianceWhen averages lie
● FAT TAIL — REALITY● THIN TAIL — TEXTBOOK
Live visualization: under a power law with α<2, a single observation can outweigh the sum of all others. That never happens with a Gaussian.
POWER LAWS • PARETO • BLACK SWANS • TAIL RISK • EXTREME VALUE THEORY • LINDY • ERGODICITY • FRAGILITY • ANTIFRAGILITY • ZIPF • POWER LAWS • PARETO • BLACK SWANS • TAIL RISK • EXTREME VALUE THEORY • LINDY • ERGODICITY • FRAGILITY • ANTIFRAGILITY • ZIPF •
/// 01 — ABOUT
We study what averages hide.
HeavyTails.com started as a quant blog in 2016 and grew into an independent lab, newsletter and culture zine about extremistan. We translate rigorous statistics — regular variation, stable laws, EVT — into plain language for founders, traders, engineers, climate nerds and curious minds.
Thin-tailed world (Mediocristan): height, weight, test scores. No single person doubles the total. Thick-tailed world (Extremistan): wealth, city sizes, wars, pandemics, book sales, crypto returns, internet traffic, wildfires. One event rules them all.
2016 — ORIGIN First post: Why your backtest lies. Variance is infinite, Sharpe is a joke.
2019 — TAIL LAB Open tools to simulate Pareto, Lévy, lognormal and compare ruin probabilities.
2023 — EXTREMES ATLAS 400+ datasets: earthquakes, floods, outages, hacks, viral hits mapped.
2026 — TODAY 84k readers, monthly print zine, workshops for risk teams and creators.
63×
More impact from top 0.1% vs median in our web traffic data
More impact from top 0.1% vs median in our web traffic data
Rule #1
Never cross a river that is on average 1m deep.
Never cross a river that is on average 1m deep.
/// 02 — CORE CONCEPTS
Six ideas to survive Extremistan
Hover cards. No PhD required. Only skin in the game.
α-stable • Power law
Fat Tails ≠ Fat Belly
P(X>x) ~ x-α. Smaller α = wilder world. Finance α~3, wealth α~1.5, wars α~1.1. Below α=2 variance is infinite. Below 1 even the mean explodes.
Ruin • Ergocity
Ensemble ≠ Time
100 gamblers one night vs you 100 nights. Expected value lies if ruin is possible. First survive, then optimize. No absorbing barrier, no game.
Antifragile
Love Convexity
Fragile hates volatility, antifragile feeds on it. Barbell: 90% ultra-safe + 10% wild bets with capped downside and open upside. Never middle-risk.
Zipf • Networks
Why Virality Is Pareto
City ranks, word frequencies, website hits, crypto caps — all Zipf. Preferential attachment + feedback loops = winner-takes-most. Design for distribution, not average user.
f(k) ~ 1/k — rank #1 is 2× #2, 10× #10
EVT • Climate
Records Are The New Normal
Floods, heat, fires follow Fréchet / Gumbel tails. A “1000-year flood” every 7 years means your model is thin-tailed. Extrapolate tails, not means.
Lindy • Practice
How To Act Tail-Wise
Redundancy over efficiency. Optionality over prediction. Via negativa: remove ruin risks first. Keep cash, keep slack, keep small experiments running.
/// 03 — INTERACTIVE LAB
Drag the tail. Feel the risk.
This is a live Pareto simulator. Lower α = heavier tail. Press “Run 2000 samples” and watch how the maximum dominates the sum. With Gaussian data the max is boring. With α=1.2 the max often beats the rest combined.
MAX / SUM RATIO—
TOP 1% SHARE—
VERDICTMove the slider →
/// 04 — EXTREMES ATLAS
Beauty lives in tails
What we track
Market crashes, startup outcomes, book sales, YouTube views, earthquake energies, cyber breaches, power outages, wildfire areas, city populations, hack ransoms. If it has outliers, it belongs here.
How to use it
Founders: size your bets. Risk managers: stress beyond VaR. Creators: play the hits game — many small shots, one monster tail. Citizens: demand buffers, not forecasts.
Explore journal →/// 05 — JOURNAL / NEWS
Fresh from the edge
Field notes, data dives and survival guides. Updated weekly. Click any story to open its full page.
/// 06 — FAQ
Thin questions, thick answers
What does heavy-tailed actually mean?
It means extreme events are not exponentially rare. The survival function decays like a power law. Practically: sample mean wanders, variance may be infinite, and history underestimates future maxima. Think wealth, not height.
Why should a founder / trader / creator care?
Because payoff lives in tails. One product, one trade, one video pays for 100 failures. Optimize for number of shots + size of upside, not average conversion. And cap ruin first.
Isn’t this just “black swans” hype?
No. Black swans are unpredictable large impacts. Heavy tails are measurable structures — we can estimate α, simulate ruin, build barbells. Less prophecy, more engineering.
How do you estimate tail risk without a PhD?
Plot log-rank vs log-size. Straight line = power law. Use Hill estimator for α, look at max-to-sum ratio, stress test 5–10× worse than history. Keep buffers. Details in our Lab section.
Can I republish HeavyTails charts?
Yes for non-commercial use with credit to HeavyTails.com. For commercial or book use, contact us via the form below.
/// 07 — NEWSLETTER
Join 84k tail-watchers.
One email / month. No spam, no predictions, only extremes: datasets, tools, book notes, failure post-mortems.
BY JOINING YOU AGREE TO LOVE OUTLIERS.
/// CONTACT