Law of Large Numbers
The principle that averages of large samples converge to the expected value—valid for thin-tailed distributions but not for fat-tailed ones.
As defined in et_747
Other definitions
A statistical principle that as a sample grows very large, no single observation can sway the total and the average converges to a stable value.
Nassim Nicholas Taleb on Black SwansA statistical principle that the average outcome of many independent trials converges to the expected value, so a small edge becomes reliable over many bets.
#387: Jim Simons Built The World’s Greatest Money-Making Machine
Where it comes up · 24
- The Strange Math That Predicts (Almost) Anything7 claims
- Nassim Taleb — Meditations on Extremistan3 claims
- Nassim Nicholas Taleb on Skin in the Game3 claims
- Nassim Nicholas Taleb on Black Swans3 claims
- Luca Dellanna on Risk, Ruin, and Ergodicity2 claims
- Rational and Religious (with Ross Douthat)2 claims
- Nassim Taleb - The #1 Rule to Not Be a Sucker (The Ergodicity Principle)2 claims
- Terence Tao2 claims
and 16 more
Most important claims · 10 of 42Search all claims for “Law of Large Numbers” →
- The central error in misusing averages is mistaking a vertical (ensemble) average for a time average: 100 gamblers each gambling one day produces a valid expected return via the law of large numbers, but a single person gambling for 100 days yields a completely different process because if they go bankrupt on any day the sequence ends — so the two are governed by different mathematics.Nassim Taleb - The #1 Rule to Not Be a Sucker (The Ergodicity Principle)
- The clean statistical results (law of large numbers and central limit theorem) only hold for perfectly executed random sampling; in practice sampling plans are rarely executed perfectly because you never get near 100% cooperation, and the non-responders differ systematically from responders in unobservable ways, producing skews rather than mere random noise.Doug Rivers on polling 7/21/2008
- Berlekamp argued that frequent trading reduces the impact of any single bad trade, so a casino-style approach that profits from being right on a bare majority of trades, around 51 percent, can yield large profits through the law of large numbers.#108 Jim Simons (Money Printer)
- Medallion was modeled on a casino: with many trades and a slight statistical edge, the law of large numbers means the firm only needs to be right about 51 percent of the time to profit.#387: Jim Simons Built The World’s Greatest Money-Making Machine
- In Mediocristan, variables like weight follow Gaussian distributions where no single observation can dominate the total (the heaviest person is an inconsequential fraction of total weight), whereas in Extremistan, variables like income are dominated by a small number of extreme observations (adding Bill Gates radically changes the total), so the exceptional is consequential.Nassim Nicholas Taleb on Black Swans
- Markov built a two-state predictive machine using transition probabilities (e.g., 13% chance vowel-to-vowel, computed by dividing 0.06 by 0.43) and showed that running it produced a ratio converging to 43% vowels and 57% consonants—the exact split he had counted by hand—proving dependent events still follow the law of large numbers.The Strange Math That Predicts (Almost) Anything
- A risk that is ruinous for an individual (Russian roulette dropping expected win from $5,000 to $0 over time) can retain its positive expected value for a company that employs many players and simply hires replacements when one is lost, because the population invokes the law of large numbers the individual cannot.Luca Dellanna on Risk, Ruin, and Ergodicity
- The central error in statistics and risk management is mistaking a vertical/ensemble average for a time average: sending 100 gamblers to a casino for one day gives an accurate expected return by law of large numbers, but sending one gambler for 100 days produces a completely different outcome even with identical positive expected value, because a single ruin event (bankruptcy on day 28) eliminates all future compounding.Nassim Taleb - The #1 Rule to Not Be a Sucker (The Ergodicity Principle)
- You only obtain the expected average outcome if you are allowed to continue to play the game; if ruin removes you, no amount of favorable odds delivers the average return.Luca Dellanna on Risk, Ruin, and Ergodicity
- Reinsurance exists because insurers face correlated shocks: an insurer concentrated in earthquake or flood coverage in Florida or California may have no claims some years and claims from 80% of customers in others, so it lays off some risk to a larger, more diversified pool with a bigger capital stock that can bear it — unlike the independent coin-flip case where a large pool nearly eliminates the chance everyone is wiped out.232 | Amy Finkelstein on Adverse Selection and Hidden Information