How to read these pages
A plain-English guide to every figure on Unstructured Alpha: what a stock's typical weekly move with oil or interest rates means, what the 90% range tells you, and when a reading counts as Clear.
- The typical weekly move
- Beyond the market
- The 90% range
- Clear, Tentative, or not
- Beyond the core five
- What this cannot tell you
The typical weekly move
Each reading answers one question: in a week when this force moved by a set amount, how did the stock typically move? It is measured over the last 3 years of weekly returns. The set amounts are deliberately ordinary weeks, not crises:
- Interest rates: the 10-year Treasury yield rose 0.25 percentage points
- Inflation expectations: 10-year inflation expectations rose 0.25 percentage points
- U.S. dollar: the trade-weighted U.S. dollar rose 2%
- Oil and energy: the price of oil rose 10%
- Credit spreads: corporate credit spreads widened 0.25 percentage points
So "+2.4% with oil" means: in weeks when oil rose 10%, the stock typically rose about 2.4% more than it otherwise would have. A negative figure means it typically fell. Moves scale roughly in a straight line, so half the oil move means about half the stock move.
Beyond the market
Most stocks rise and fall with the market as a whole, and the market itself reacts to oil, rates and the rest. Each reading is measured after removing the stock market's own move, and the five core forces are measured together, so each shows what that one force adds on its own. A stock can rise in a week when oil rose because everything rose; that part is not counted as oil.
The 90% range
Three years of weekly data pin a reading down only so far. The 90% range is the band the true figure very likely sits in, given how noisy the weeks were. A narrow range is a precise reading; a wide one, a rough one.
The single most useful thing to check is whether the range crosses zero. If it does, the data cannot rule out that the stock has no link to that force at all.
Clear, Tentative, or not
Every reading carries one of three labels, from how far the figure sits from zero relative to its own noise:
- Clear The figure is far enough from zero that chance is an unlikely explanation, even allowing for the fact that five forces are tested at once (|t| ≥ 2.58).
- Tentative The 90% range excludes zero, but the bar for Clear is not met (|t| ≥ 1.64). Worth noting, not worth leaning on.
- Not distinguishable from zero The range includes zero: no link that can be told apart from noise. This is the most common answer, and it is an answer, not a gap.
A stock with fewer than 2 years of weekly prices gets no label at all: too little data to say anything. Pages rank and count only Clear and Tentative readings; the rest are shown grey, never dropped and never ranked.
Beyond the core five
The other forces (short-term rates, gold, bitcoin, the yen and more; 19 in all) are each measured with the stock market and the core five held fixed, so they show only what those do not already explain. Because so many are tested, their bar for Clear is stricter (|t| ≥ 3.01).
What this cannot tell you
- It is the past, not a forecast. A stock that moved with oil for three years may not over the next three. The track record measures how often readings held up a year later.
- It is not cause and effect. A reading says the two moved together, not why.
- It is not advice. Nothing here is a recommendation to buy, sell or hold any security.