How to read this

What an option is

An option is a contract giving the right — not the obligation — to buy (a call) or sell (a put) the index at a fixed price, called the strike, up to a fixed date, called expiry. You pay a premium for that right. If the index never reaches your strike, the premium is lost.

Why time matters so much

An option is a wasting asset. Every day that passes, some of the premium disappears simply because there is less time left for the index to move. This is why so many patterns here are about days-to-expiry — and why several of them name a longer window than they actually measured: no price exists for an at-the-money option on expiry day itself in this dataset, so the final step into expiry cannot be measured directly. Each affected pattern's card says so explicitly, in the box headed "What was actually tested."

What each family asks

Expiry — how premium and implied volatility behave as an expiry date approaches — the most mechanically grounded family, since time decay is a structural feature of an option contract, not a claimed behavioural edge.
Volatility — whether implied volatility, once unusually high or low relative to its own recent history, tends to return toward normal.
Positioning — whether where the crowd has placed its money — open interest, the put-call ratio — says anything about what the index does next.
Calendar — whether particular days of the week, months, or scheduled events (like the Union Budget) behave differently from an ordinary session.

Explain and Analyst

Every pattern on this site is described twice: a plain-English paragraph that is always visible, and a full statistics table one click away, under Analyst in the header. Switching modes never removes or replaces anything — Explain mode simply keeps the statistics table folded away so the page is easier to skim. Your choice is remembered on this device. What was actually pre-registered and tested is shown a third way, separate from both: in a box under every pattern's name, visible in either mode, because that claim should never depend on which button you happened to click.

How to read a verdict

Confirmed means it held on years it was not found on, and survived a correction for the number of ideas tested. Candidate means it looked real once but did not repeat — treat it as not proven. Rejected usually means there is no evidence for it at all — but not always: a handful are rejected because the evidence is real and decisive in the opposite direction from what was predicted before testing, which fails this project's own test just as surely, but is a different fact than "nothing is there." Each rejected card says which one it is.

Not tested is different from all three, not a quieter way of saying any of them. It means this project's own checks cannot yet tell whether the idea is true or false, so it was never run at all — most often because the idea's own event selection and its outcome measurement rest on the same number, which can manufacture an apparent pattern out of pure noise if not corrected for, and the correction has not been built yet. A "Not tested" pattern has not failed. It is unanswered.

Always read the yellow and red caveat boxes attached to a pattern. A pattern can be genuinely real in the data and still be useless — most importantly if, like weekly_final_two_days, it only ever worked under market rules that no longer exist.