The Equinox Test

The Equinox Test

A ninety-year-old claim, forty chances to check it, and one week to start.

W. D. Gann argued that markets turn with the seasons — that the four points where the sun changes direction, the two equinoxes and the two solstices, mark the places where trends are most likely to end. He was not alone in it. Traders have watched these dates for a century.

Most people hear that and do one of two things. They believe it, or they laugh at it. There is a third option, and it is the only one worth anything: go and look.

You have ten years of Nifty on the screen in front of you and four dated events a year. That is forty tests. Nobody has to tell you the answer.

The dates

March equinox, June solstice, September equinox, December solstice. Note they drift — the September equinox falls on the 22nd in some years and the 23rd in others, and March moves between the 20th and 21st. Working from "around the 21st" will put you on the wrong candle in half the years.

2016  20 Mar   ·   21 Jun   ·   22 Sep   ·   21 Dec

2017  20 Mar   ·   21 Jun   ·   23 Sep   ·   21 Dec

2018  20 Mar   ·   21 Jun   ·   23 Sep   ·   22 Dec

2019  21 Mar   ·   21 Jun   ·   23 Sep   ·   22 Dec

2020  20 Mar   ·   21 Jun   ·   22 Sep   ·   21 Dec

2021  20 Mar   ·   21 Jun   ·   23 Sep   ·   21 Dec

2022  20 Mar   ·   21 Jun   ·   23 Sep   ·   22 Dec

2023  21 Mar   ·   21 Jun   ·   23 Sep   ·   22 Dec

2024  20 Mar   ·   21 Jun   ·   22 Sep   ·   21 Dec

2025  20 Mar   ·   21 Jun   ·   22 Sep   ·   21 Dec

2026  20 Mar   ·   21 Jun   ·   23 Sep   ·   22 Dec

Dates in Indian time.

The orb, and what it costs you

Nobody serious claims the market turns on the exact day. The convention is a window either side — call it five trading sessions before and five after.

That is fair. It is also the single most important number in this exercise, and here is why.

Five sessions each side is an eleven-session window. Four events a year is forty-four sessions. An NSE year has about 246.

So your windows already cover roughly 18% of every trading year. Turns scattered completely at random would land inside them about one time in five. That is not an argument against the idea — it is the number the idea has to beat.

If you find turns in the window 20% of the time, you have found nothing at all. You have found the calendar. If you find them 60% of the time, you have found something worth a second look.

How to run the test

Fifteen minutes of setup saves you from an answer you cannot trust.

●      Write the rule down before you open the chart. What is a "turn"? Pick something you cannot argue with later — for example: a swing high or low that price moves at least 3% away from within the next ten sessions, and does not exceed again for twenty.

●      Mark all forty-four dates first. Every one, all ten years, before you look at what price did. Marking them as you go is how you end up seeing only the ones that worked.

●      Score each one hit or miss. Eleven sessions wide. It either contains a turn by your written rule, or it does not. No "close enough".

●      Count the misses out loud. Forty events. If thirty missed, that is your answer, and it is just as useful as the other one.

●      Then do the control. This is the step almost nobody does, and it is the only one that makes the rest mean anything.

The control

Pick forty dates at random across the same ten years — throw a dart at the calendar. Build the same eleven-session windows. Score them with the same rule.

Now compare. If your random dates score 8 out of 40 and the equinoxes score 9, the seasons are doing nothing; you were looking at how often markets turn at all. If the random dates score 8 and the equinoxes score 24, you have something that deserves more work.

Without the control you cannot tell a real effect from the ordinary restlessness of price. Every chart contains a turn somewhere near almost any date you choose. That is what makes this kind of claim so easy to believe and so hard to verify.

Where we are right now

The September 2026 equinox falls on Wednesday 23 September. With a five-session orb, the window runs from 16 September to 30 September — which means it opens today.

Watch it live. Note what you see before you know the outcome, not after. One observation proves nothing, but a note written in advance is honest in a way that a memory reconstructed afterwards can never be.

What we are not saying

We are not telling you the market turns on these dates. We have not published a study on it and we are not going to pretend otherwise. This is an idea with a long history, a plausible mechanism nobody has pinned down, and a body of evidence that is mostly anecdote.

What we are saying is that it is testable — and that you can test it this week with a chart and a notebook. That is the whole skill. Most of what circulates in trading is neither proved nor disproved, only repeated. The habit of checking is worth more than the answer to any single question.

If you run it, tell us what you found. Especially if it was nothing.

Educational content only. Nothing here is investment advice or a recommendation to buy or sell anything.