Astrology makes claims about the world. "Markets bottom near the new moon." "A grand cross brings crisis." Whatever you believe about why, those are statements that can be checked — and checking them is a craft with its own traps.
This is a lesson about that craft. It is also, in the second half, a report on what happened when we pointed it at our own work.
Someone says the market fell after a difficult aspect. You look, and it did. Case closed?
No — because markets fall roughly half the time. The claim only means something if it beats what you would have expected anyway. That baseline is called a null model, and almost every argument about astrology founders on not having one.
It is not a hostile idea. It is the only way a hit can count as evidence at all: if you cannot say what chance looks like, you cannot say you have beaten it.
The sharpest version of the test is disarmingly simple. Take the astrological signal — a planetary curve, say — and scramble it. Keep the shape, keep the rhythm, destroy only its alignment with the dates. Then run exactly the same analysis.
If the real signal does no better than its own shuffled copy, it carried no information. You have not tested it against a hostile assumption; you have tested it against itself.
Here is the trap that catches careful people.
Test thirteen configurations. One comes back striking. Is that a discovery?
Usually not — because at the conventional threshold, roughly one test in twenty looks interesting by chance, and you ran thirteen. The p-value beside that one result does not know how many others you tried. It is answering a question you did not ask.
The fix is to put the search itself inside the test: how often does a made-up configuration, given the same thirteen chances, look as good? That number is the honest one, and it is usually much larger than the first.
Last, and most neglected: could this ever be settled?
An effect has a size, and the thing you are measuring has a natural wobble. If the effect is one-twentieth of the daily noise, you need thousands of observations to see it reliably. If the events happen a few times a year, "thousands" can mean centuries.
A claim like that is not false. It is unresolvable — it cannot be confirmed or refuted by waiting, and no amount of new data will settle it. Knowing that in advance saves a great deal of time.
We ran these tests on our own material, because it seemed dishonest to describe the method without applying it.
We tested planetary curves as forecasting inputs. We tested eight thousand dated sky events — new moons, full moons, eclipses, stations, ingresses, outer-planet aspects — across seventy-five years of market history. We tested the classic chart configurations, four and a half thousand times over, in five separate national markets.
The results were null. Consistently, and with controls.
One figure did come back striking, and it survived a run of further checks: it replicated in several markets, it was not a seasonal artefact, it was not a calendar quirk. Then we corrected for the fact that it was the best of thirteen, and it dropped to exactly where chance predicts. And when we asked how long a forward test would take to settle the matter, the answer came back in centuries.
So: no result. We think publishing that is more useful than not, and it is the sort of thing this field very rarely reports.
It does not mean the sky is uninteresting, and it does not touch most of what astrology is for. A birth chart read well is a conversation about a life — about pattern, timing, and the meaning someone makes of them. That was never a claim about market prices, and testing prices says nothing about it.
What the exercise does mean is narrower and worth keeping: the market-timing corner does not survive its own controls, on our data, by our tests. When you meet a confident claim in that corner — ours or anyone's — the useful questions are the four above. What is the baseline? Does it beat its own scrambled copy? How many other things were tried? And could this ever be settled at all?
Those questions are good company. They are also, we would argue, the most respectful thing you can bring to a tradition worth taking seriously.