The Framework
Taking Calculated Risk
A regime signal does not tell you what will go up. It tells you whether the market is currently paying you to hold risk at all — and that is a narrower, more useful question. This page is the honest version of what the Alphameter can and cannot do for you.
What makes a risk calculated
Three things, none of them a prediction:
- You know the base rate. Before you act, you know how often this setup has worked, over how many days, and how badly it has failed. A signal without a denominator is a hunch.
- You know the blind spot. Every real edge has a market environment where it underperforms. If you cannot name yours, you have not looked hard enough — and you will abandon the strategy at the worst possible moment.
- The position is sized to survive it. The framework below spent seven years losing to a plain 60/40 portfolio. That is survivable at the right size and ruinous at the wrong one.
One input, not the answer
The Alphameter is a read on risk appetite. That is genuinely useful and it is also narrow. It does not know what a chart looks like, what a company earns, what you paid, how long you can hold, or how much of your account is already committed. Anyone who tells you a single indicator answers all of that is selling something.
Treat it as one leg of a stool. The other two are worth studying properly, in their own right:
Technical analysis
Structure, trend, levels, volume. This is what tells you where to enter, where your invalidation sits, and where to place a stop. The Alphameter has nothing to say about any of it — it gives you a climate, not a trade. Start with price action and the break-and-retest and anchored VWAP, both tested on 30 years of data — including the parts that did not hold up.
Fundamental analysis
Valuation, earnings, balance sheets, the macro data underneath. This is what tells you whether a thing is worth owning at all, and at what price — a question no regime score can answer.
Use it for confluence
The signal earns its place when it agrees with work you did independently. A setup you already like, in a market whose fundamentals you already understand, during a regime that favours it — that is confluence, and it is a better trade than any one of the three alone.
Disagreement is information too. When the chart says one thing and the regime says another, the honest response is a smaller position or none, not picking whichever input agrees with what you already wanted to do.
Worth your time in the school: position sizing, stop-loss strategies, drawdown, correlation, and the business cycle. Risk management is the part that decides whether you are still around to use a good signal when one arrives.
The one idea we have actually validated
With that caveat in place: the Alphameter's durable edge is not timing the stock market. It is rotation — moving between risk assets and defensive assets as the regime flips. The rule below is a worked example rather than a portfolio, tested as a pre-registered specification, meaning the success bar was written down before the backtest was run:
Risk-on episode
Equal-weight SPY / QQQ / IWM. Broad US equity beta, no stock picking.
Risk-off episode
Equal-weight TLT / GLD. Long Treasuries and gold, the two things that usually catch a falling market.
The signal is lagged one trading day (you trade on tomorrow's open, not on a close you could not have known), switches cost 7bps of turnover, and the rule fired about 7 times a year — 156 switches in 22 years. This is a slow framework, not a trading system.
| 2004–2026 | CAGR | Sharpe | Max DD |
|---|---|---|---|
| Regime rotation | 10.1% | 0.71 | 31.9% |
| SPY buy & hold | 9.0% | 0.55 | 56.5% |
| Static 60/40 | 6.0% | 0.58 | 32.8% |
Window 2004-11-18 → 2026-09-25 (5,496 trading days), starting at GLD's inception. Daily rebalance within each basket, 7bps per switch. Past performance of a backtested rule is not a forecast.
The blind spot, stated up front
It loses to 60/40 in a long, quiet bull market
From 2013 to 2019 the framework returned 3.3% a year against a 60/40 portfolio's double-digit run. Rotation earns its keep when volatility actually arrives; in a low-vol grind it pays you to sit in defensives while equities melt up without you. If you cannot hold through that, this is not your framework.
| Sharpe by period | Rotation | 60/40 | SPY |
|---|---|---|---|
| 2004–2012 | 0.81 | 0.40 | 0.21 |
| 2013–2019 | 0.34 | 1.11 | 0.97 |
| 2020–2022 | 0.64 | 0.13 | 0.36 |
| 2023–2026 | 1.23 | 0.89 | 1.34 |
Sub-windows were fixed before the test, not chosen afterwards to flatter the result. Three of four beat 60/40 on risk-adjusted return.
Using the dashboard for this
Everything on the site feeds one decision: how much risk to carry right now.
The composite score →
The regime itself, recomputed once a day on settled closing prices. Above +25 is risk-on, below −25 is risk-off.
Tactical overlay →
Looks up every prior occurrence of the exact six-indicator configuration showing right now, and tells you how much precedent there is for it — including when the answer is “none”. Read it as a precedent counter rather than a stock picker: its per-asset ranking has not cleared our own pre-registered bar, but its count of how unusual today is has nothing to prove.
Score momentum →
Direction matters more than level. A score climbing through zero is a different situation from one falling through it, even though both read “neutral”.
Caution score & flip probability →
Context for position sizing: how stretched the current regime is, and how often regimes this old have flipped. Use them to decide size, not direction.
Asset regime explorer →
How 33 assets behaved in each regime over 30 years. Read it as a map of asset-class behaviour — the risk-on ordering is broadly stable, but individual rankings are noisier than the table makes them look.
How to read the Alphameter →
The interpretation guide: what each of the six indicators is telling you and how the score is built.
When there is no precedent
Sometimes the six indicators line up in a configuration that has never occurred in thirty years of history. The overlay says so plainly instead of producing a confident-looking table from three stale matches, and it flags the related case too — when most of the matches it did find come from a single era, which makes them one macro episode rather than a sample.
This is where calculated risk stops being a backtest and becomes a discipline. With no precedent, the regime direction still tells you something — the six indicators are measuring today's risk appetite whether or not today has a historical twin — but the per-asset detail does not. The correct response is smaller size and a wider stop, not a more creative reading of a thin sample. A framework that admits when it is out of data is more useful than one that never does.
What not to do with this signal
Don't trade it intraday
Every result on this site comes from daily closing bars. A daily bar cannot validate an intraday strategy, so we make no claim about one.
Don't treat risk-off as a shopping list
Our own stability testing found the risk-off asset rankings are mostly noise — they do not hold up between halves of the sample. Rotate to defensives as a basket rather than chasing whichever asset topped a risk-off table.
Don't expect precision on the turn
The Alphameter has flagged every major crisis of the last 30 years, typically three to four weeks early — and it also raises risk-off warnings that never become crises. Early and occasionally wrong is the honest description.
Start with today's regime
See where the six indicators sit right now, then decide what size that justifies.
Not investment advice
Alphamancy publishes research, not recommendations. Nothing here is personalised financial advice, the baskets above are a test specification rather than a portfolio we are telling you to hold, and backtested results are not a forecast of future returns. Do your own work and consider speaking to a licensed adviser.