The largest allocation decision in finance runs on one rule.
Market-capitalisation weighting — the default that decides where the world's capital sits. Adopted so widely it's no longer examined. It should be.
The index is 150 years old — and so is its bias.
The formulas underneath it predate modern markets. The bias wasn't an accident — it was written into the arithmetic from the start.
The index is a design — not a mirror.
We treat it as a neutral reflection of the market. But every index is built from choices — and choices embed bias.
choose which companies are in — and which are not.
choose how much of each — weight by size, not by merit.
choose what even counts as investable.
Three design decisions — one inherited bias.
It mistakes size for information.
Cap-weighting rewards a stock for being big, not for being right — a self-reinforcing loop that mistakes momentum for merit.
The flaw is statistical, not moral — no one chose it; everyone inherited it.
An unexamined default has a cost.
Flows chase weight, not value — capital follows what already rose.
A shrinking handful of names comes to dominate the benchmark — and every portfolio that tracks it.
The active industry is judged against a biased default — so beating it looks like heresy, not arithmetic.
The benchmark everyone trusts was never an optimum.
We treat the S&P 500 as the market itself — a neutral, unbeatable standard. It is a committee's design with a cap-weighted formula: a choice, not a law of nature.
Once you can see the bias, beating it stops being heresy.
It becomes arithmetic.
If the bias is statistical, it can be removed.
Re-weight by the probability of growth instead of by size, and you keep everything a benchmark must be — while stripping the bias out.
Broad, rules-based, investable — everything an allocator needs.
The rich-get-richer weighting that rewards size over merit.
A model broad enough to track the market, tilted enough to beat it.
The thesis isn't just argued — it's measured.
The S&P 500, beaten from every starting point.
Since January 2014, an E&R portfolio launched every month-end — 324 in all — run on one-, two- and three-year clocks, with no discretion. Not one lucky vintage: the whole grid.
Beating the benchmark wasn't a call — it was the design, provable from any date you pick.
Derived in the open — not asserted from a stage.
Two decades of work, from the origins of the index formula to the probability model that replaces it.
