When the CLT Quietly Fails

"The CLT is reliable when its preconditions hold. The preconditions fail silently in most real-world data, and the bell still shows up because the audience is expecting it."

The Two Preconditions, Reviewed

The standard CLT requires:

Independence between the X's being summed.
Finite variance of each X.

When both hold, the standardized sum converges to a normal as N grows. When either fails, the conclusion no longer follows. The bell may still appear, but for the wrong reasons, or it may not appear at all — and either way, downstream tools that assumed normality are now mistaken.

Failure Mode 1: Hidden Dependence

The cleanest failure is when the X's are correlated. The CLT proof crucially depends on the variance of the sum being the sum of variances — an identity that only holds under independence. With positive correlation, the sum's variance is larger than the independence-assumed variance; the standardization step understates the true spread; the resulting tail probabilities are wildly underestimated.

This is exactly what happens to financial returns during a crisis. On normal trading days, asset returns are weakly correlated, and the CLT-derived risk model works reasonably well. On crisis days, correlations spike toward 1 — everything sells at once — and the 'rare' big move predicted by the bell becomes a Tuesday afternoon. The CLT didn't lie; the user forgot the independence precondition.

Failure Mode 2: Infinite Variance

The standard CLT needs each X to have a finite variance. Power-law-tailed variables with tail exponent α ≤ 2 have infinite variance, and the standard CLT does NOT apply. Instead, sums of such variables converge to stable distributions, which can be heavy-tailed and asymmetric. The bell never shows up; treating the average as if it were bell-shaped is mathematically meaningless.

This matters for natural phenomena with genuine power-law structure: earthquake magnitudes, file sizes on the web, social-media virality, certain asset returns. The fact that someone computed an 'average' for these quantities doesn't make the average meaningful — the underlying mathematics has refused to cooperate.

Failure Mode 3: Slow Convergence

Even when independence and finite variance both technically hold, the CLT can be slow to converge when the individual X's are very skewed or heavy-tailed (but with technically finite variance, like the lognormal). 'N is large enough' becomes a function of how non-normal the underlying X's are. For a highly skewed X, N=30 might be wildly insufficient; N=10,000 might still be wrong in the tails. The CLT is asymptotic — finite samples are always approximations.

The Citizen's Test

Before trusting a bell-shaped assumption, run the precondition check: are the underlying X's plausibly independent? Is the variance plausibly finite? Is N large enough relative to how non-normal the X's are? If any answer is 'no' or 'I don't know,' the bell is a guess, not a derivation. Track 07 will dismantle the worst real-world consequences of skipping this test.

Code

CLT under hidden correlation·python

import numpy as np
rng = np.random.default_rng(60)

# Show CLT failing under correlation.
# Sum of N=100 highly-correlated variables vs N=100 independent ones.
N = 100
M = 10_000

# Independent draws.
independent = rng.normal(size=(M, N))
sum_indep = independent.sum(axis=1)

# Correlated draws — all share a common factor.
common = rng.normal(size=(M, 1))
idiosync = rng.normal(size=(M, N))
correlated = 0.7 * common + 0.3 * idiosync
sum_corr = correlated.sum(axis=1)

# Standardize both.
def stats(x, label):
    z = (x - x.mean()) / x.std()
    p4 = (np.abs(z) > 4).mean()
    print(f"{label:>25s}: std={x.std():>7.2f}  P(|z|>4)={p4:.4f}")

stats(sum_indep, "independent (CLT holds)")
stats(sum_corr, "correlated (CLT broken)")

# The independent sum has the bell-like P(|z|>4) ~ 0.0001.
# The correlated sum's standard deviation is much larger because variance
# accumulates beyond the independence-assumed bound — but the bell-of-
# z still looks fine when you standardize. The trouble is the RAW values:
# extreme outcomes happen far more often than independence would predict.

Exercise

Pick a quantity in your work or life where you suspect hidden correlation between supposedly-independent observations (responses to a single email, traffic on different parts of your site, replies during a viral moment). Estimate: would the CLT's prediction of 'rare events' underestimate or overestimate the real-world frequency? Almost always: underestimate, because correlation amplifies tail behavior.

Hint

If observations share a common cause (a news event, a celebrity tweet, a system outage), their independence is illusory and the bell-derived 'this is a 1-in-a-million event' will fire much more often than once in a million.

The Two Preconditions, Reviewed

Failure Mode 1: Hidden Dependence

Failure Mode 2: Infinite Variance

Failure Mode 3: Slow Convergence

The Citizen's Test

Code

External links

Exercise

Progress

Comments 0