Verified, Not Predicted: How the NiftyDesk Score Is Graded Every Session

The only feature that actually matters

Every directional service in India sells the same thing: screenshots of winning trades, a wall of testimonials, and a vague promise of accuracy. The losing trades are absent. The base rate of how often the system actually works is unknown to the buyer.

The NiftyDesk Score is built on the opposite assumption. Every directional call we publish is graded — publicly, unedited, against the same-day Nifty 50 close — and the rolling 30-day record is on /today for anyone to read before they sign up. No login wall. No "request access". You can audit how the score has actually performed first, then decide.

If you want to skip to the data, /today is open right now. If you want to act on it, the 30-day free trial is open with no card required.

How the grading actually works

The mechanism is deliberately boring, because it has to be.

1. The call is the call. Whatever directional call (BULLISH, BEARISH, or NEUTRAL) the page shows during the session is the call we are accountable for. The score evaluates live and refines through the day; the day's final direction is what is graded.
2. The verdict is the Nifty 50 close. No adjustments. No "well, it touched our level intraday so we count it as a win". The session close is the session close.
3. The rule is simple, and it is the rule on the homepage FAQ:
   • BULLISH is correct if Nifty closes above the open.
   • BEARISH is correct if Nifty closes below the open.
   • NEUTRAL is correct if the close is within a pre-published narrow band of the open.
4. The result goes onto the public record. The 30-day strip on /today shows per-day verdicts and the rolling accuracy. We do not get to revise calls after seeing the close.

That is the whole mechanism. No asterisks, no curation, no buried losses.

What you actually see on /today

When you open /today, the verification surface is two pieces:

• The rolling 30-day accuracy — a single percentage covering the last thirty sessions. Updated daily.
• The per-day strip — a row of the last thirty sessions, each one tagged correct, wrong, or neutral. Visual and auditable at a glance.

That is intentionally simple. We do not publish a separate "accuracy by conviction band" table, and we do not publish a curated highlights reel. The strip is the strip, the percentage is the percentage, and the calculation rule is the FAQ rule above. Anyone visiting the site for the first time can read all three in under thirty seconds.

Why a 30-day rolling window

Two reasons.

First, thirty sessions is long enough to be a meaningful sample. A trader who sees five sessions of results cannot distinguish signal from noise. A trader who sees a year of historical results is reading history, not present-tense performance — and present-tense is what matters for a live product.

Second, a rolling window is honest. A static "since inception" number lets a product hide a bad quarter behind a good one. A rolling 30-day number cannot hide a bad week — it shows up immediately. We chose the harder number on purpose.

What "verified" rules out

The grading methodology rules out a set of dishonesties common in the directional-signal industry. These are worth naming because they are everywhere:

• Selection bias. We cannot cherry-pick which sessions count. Every session lands in the public strip.
• Definitional drift. "Correct" cannot mean "the market went up" some days and "the market touched our level intraday" other days. The rule is the close.
• Retroactive adjustment. We cannot wait until 2:00 PM to refine a session's call into something more flattering.
• Survivorship bias. Old, bad calls do not disappear from the record. They roll out of the window on schedule.
• Testimonial substitution. We do not need a wall of trader quotes when the live record is on a public page.

If you have used directional services before, you have almost certainly encountered at least three of these. The mechanism is a deliberate response to all five.

Why this matters more than the score itself

A trader's job is to size capital against a system whose edge they trust. Trust is built from one thing: the ability to audit the system's history without the seller getting to choose what you see. Every other feature of the product — the live update, the regime-aware synthesis, the conviction tag — is downstream of this. If you cannot verify what the score has done, none of those features matter.

This is why we will spend more engineering effort on the integrity of /today than on any other surface of the product. A score that cannot be audited is just a confident opinion.

A note on what the accuracy is

We are deliberately not putting a specific accuracy number in this post, because the rolling window changes every session and any number printed here is wrong by tomorrow. Read the live number on /today — that is the only honest answer at any given moment.

What we will say:

• The accuracy is not 100%. No directional system is. Anyone selling 100% is selling a curated screenshot.
• Most NiftyDesk sessions resolve at MEDIUM or LOW conviction, by design. The score is calibrated to be honest about uncertainty rather than manufacture HIGH conviction every day.
• The honest base rate over a rolling 30-day window is what is on the page. Read it, decide whether it is tradable for your method, and then act — or do not.

What you should do before signing up

Three things, in order.

1. Open /today without logging in. Read today's call, the rolling 30-day accuracy, and the per-day strip.
2. Come back in a week. The window will have shifted by five sessions. See whether the calibration holds.
3. If the record reads as honest, start the 30-day free trial. No card required. Use the score on a handful of sessions before committing capital. Cancel any time if it does not earn its keep.

We would rather you audit the record for a week than sign up on a marketing impulse. Traders who audit before they pay stay subscribed; traders who pay on impulse churn — and we are not interested in churn revenue.

What this does not mean

The 30-day record does not mean the score will be right tomorrow. It does not mean the calibration will hold through every market regime. It does not mean that following the score blindly will produce returns. None of that is true of any directional product, and we would be lying to imply otherwise.

What it does mean is that the base rate of the score's accuracy is knowable, public, and current. That is the floor of honesty we hold ourselves to. Everything else — position sizing, strike selection, risk management — is the trader's job, and we are deliberate about not pretending to do that work for you.

What to read next

• What is the NiftyDesk Score? — the pillar explainer.
• How to use the NiftyDesk Score during a Nifty session — the practical reading guide.
• NiftyDesk Score for options traders — aligning the call with your options view.
• Why the NiftyDesk Score is regime-aware — regime as an input.

---

See the live 30-day verified record on /today. When you are ready, start the 30-day free trial — no card required.