The benchmark everyone trusts is structurally skewed.
The $50 trillion passive industry runs on one idea — market-cap weighting, traceable to Laspeyres' price index of 1871. Adopted for convenience, it carries an intrinsic positive bias.
As a stock rises it earns a bigger weight, so the index buys more of it — mistaking momentum for merit and quietly concentrating risk in a handful of names.
A 150-year-old convention, not a law of markets. 3N treats cap-weighting as one regime — and measures the other at the same time.
Most stocks sit still. A few carry the index.
The market is 500 balls — one per stock. Only a handful actually move, swelling and shrinking, while the rest sit still.
Estimate each stock's chance of growing, then tilt a broad basket toward the balls about to swell. Each stock as odds, not a verdict — persist, revert or transition.
Neither purely Normal, nor purely Non-Normal.
One idea — that markets are both at once — turned into a probabilistic engine. Three layers, one control.
Markets as a Markov chain of states — every asset carries measurable odds of persistence, reversion or transition.
Value bets on reversion; cap-weighting bets on persistence. 3N measures the prevailing regime and normalises across both.
Machine Beta rebuilds the index from non-linear structural factors, not market value — learning the structure of returns at lower tracking error.
Every asset carries measurable odds — persistence, reversion, transition.
Five return states form a live transition network. For every state, the odds of staying plus the odds of leaving always sum to 100% — and each link is two-way.
Over a single day each state holds itself — five separate circles. Lengthen the horizon and they knit into a chain; over long horizons every state reaches every other and self-persistence falls near 50% — far less predictable.
Two schools each bet on one regime. 3N measures both.
Value assumes prices return to a mean — cheap becomes dear, dear becomes cheap.
Cap-weighting assumes winners keep winning — size compounds into more size.
3N measures which regime prevails and normalises across both — correcting for recency, anchoring and confirmation bias instead of hard-coding one belief.
Machine Beta rebuilds the index from non-linear structural factors — not market value.
It learns the structure of returns directly, through non-linear mechanisms, at lower tracking error. The output is E&R — Exceptional & Rich: a rules-based basket ranked on relative cap growth rather than size — broad enough to track the market, tilted enough to beat it.
3N wasn't designed. It was derived — over fifteen years.
Each layer traces to peer-visible work, published openly on SSRN since 2010. The methodology is the accumulation of that record.
Structure, not luck — measured by scorekeepers we don't control.
E&R launched a portfolio every month-end since 2014 — 324 in all, run on one-, two- and three-year clocks. When a design wins from any start date and any cadence, it isn't luck — it's structure.