We found a market law. Then we spent six months trying to destroy it.

Every price move has two properties: a size and a direction. How big, and which way.

For a century, the dream of quantitative finance has been to predict both. A system that nails size and direction is a money machine. A system that nails neither is noise. Almost everyone selling a "signal" claims to be closer to the money machine than they are.

We are going to tell you something stranger, and we think more useful.

We built a single number — one transparent measurement, computed the same way across stocks, bonds, currencies, Treasuries, and crypto — and asked it those two questions separately. The answer came back with an asymmetry so clean, and so stubborn, that it survived every attempt we made to break it:

The number reliably tells you how large the coming move will be. It tells you nothing about which way. And — this is the part that matters — the strength of the size-signal isn't random. It's graded by the depth and maturity of the market. Strongest in deep, structured markets like gold and Treasuries. Faintest in thin, reflexive venues like prediction markets.

That gradient is the whole result. Let's unpack why it should interest you whether you manage a personal portfolio or a risk desk.

The intuition: a crowd, not a crystal ball

Imagine a room full of forecasters, each watching the same asset through a different lens — one tracks momentum, one tracks trend, one tracks the trading range, and so on. At any moment you can ask: how much do they agree?

When they're scattered, the market is incoherent — pulling in every direction at once. When they snap into alignment, the market is coherent — everyone leaning the same way.

We compress that agreement into one number we call the coherence hub. And here is the finding: high coherence reliably precedes larger moves. The crowd tightening up is a tell that volatility is coming.

But — and this is the beautiful, frustrating part — coherence says nothing about where the crowd is about to run. A tightly-aligned market is a market about to move hard. Which way it breaks is, as far as our number can see, free.

That's the two-sentence version a lay investor can carry around: coherence predicts the storm's strength, not its heading.

Why we tried to destroy our own result

Here is what separates this from the usual "coherence predicts X" paper, and why an institutional reader should keep reading.

A coherence statistic is dangerously easy to fool yourself with. Pool enough heterogeneous assets and eras together and you can manufacture predictability that dissolves the moment you look inside any single unit or test it on unseen data. The history of quantitative finance is a graveyard of signals that were real in-sample and gone the moment real money touched them.

So before we believed anything, we built a falsification-first protocol designed to break our own findings:

• Within-unit confirmation. A pooled correlation is provisional until it shows up inside individual assets, not just across the pile.
• Spline-controlled partials. We strip out the possibility that our number is just a disguised nonlinear function of recent volatility.
• Block-permutation nulls at the autocorrelation length — because naive shuffling over-rejects on autocorrelated series and flatters you.
• Chronological out-of-sample splits. Fit on the first half of history, test on the second. "Significant but time-unstable" is an allowed, reported verdict.
• A Texas-sharpshooter guard. Once a finding survives in-sample, we stop re-slicing the same era. Only genuinely new forward data is allowed to de-risk it further.

The magnitude law passed. Across substrates that share almost no microstructure, the coherence-hub → forward-volatility correlation is positive and survives the full gauntlet, with pooled significance clearing block-permutation at p < 0.003.

The direction claim failed every time we tried to make it work — which is exactly what Phase Sovereignty predicts. We threw eight independent forms of attack at it. Every apparent directional whisper died under hardening, with the same signature: anti-persistent across halves, confidence intervals straddling zero, agreement at chance. In Bitcoin — the most reflexive, most momentum-driven asset we tested — coherence captures essentially none of the direction while a momentum control grabs a strong slice of it. That's the cleanest demonstration of the wall.

The honest part: where the edge is, and where it isn't

This is the section most pitches would never write. We're writing it first, on purpose.

A lawful statistical structure does not automatically mean a tradeable edge. So we separated the science from the economics and reported the nulls with equal weight:

• Where it works. As a volatility-targeting and drawdown-timing overlay, the signal beats naive volatility targeting — clearest in crypto and metals. In crypto, the magnitude channel appears mispriced relative to options: it predicts the volatility-risk-premium (BTC Spearman 0.32, ETH 0.23, pooled 0.27, p = 0.0014), monotone across sorts, strategy t > 2, robust to costs. A cross-market synchronization measure over 27 coins flags states that precede ~1.6× volatility elevation at one day, out-of-sample stable across both time-splits and disjoint coin baskets.
• Where it doesn't. Against traded 30-day implied volatility, the signal does not beat the market in any asset class — vol-buyer win rates land at 26–36%. We tried two theory-driven routes to hidden alpha; both were clean nulls. The honest conclusion: the options market already prices the coherence level at standard tenors. The volatility edge is concentrated at the 1-day horizon and decays to nothing by 30 days — which makes the 30-day option structurally the wrong instrument.

The one remaining venue for standalone option-relative alpha is short-dated (0DTE) options, where continuous market-maker hedging breaks down. That's a microstructure question, not a signal question — and we have a live forward-collector capturing 0DTE straddle data on BTC and ETH every six hours to settle it. A first read is expected within weeks, and we will report it regardless of sign.

Why this matters to you

If you're a lay investor: the takeaway isn't a stock tip. It's a lens. Most volatility models are quietly confident right up until the moment they're wrong. Coherence is an independent early read on how big the next move could be — a way to size positions and hedges with the storm's strength in mind, while staying honest that nobody, including us, can reliably call its direction.

If you're an institutional candidate or partner: what you're looking at is a portable, interpretable measurement apparatus — one transparent complex number, no per-asset tuning, no black box — plus a falsification-first protocol that has already killed half a dozen tempting-but-false signals before they could embarrass anyone. The durable contribution isn't a single big R². It's the machinery that separates a genuine cross-asset law from a pooled mirage, a confirmed HAR blind-spot detector for risk overlays, and an honestly-open economic question with a forward test now collecting.

We're not selling certainty. We're offering a tool that knows the difference between what it can and cannot see — and tells you which is which.

The geometry is computed; the aim is free.