Ayan Jain · Backtest Audit
PaRRVA preparation, for algo providers, research analysts and advisers

Your track record, audited before the verifier sees it.

Under SEBI's framework for the verification of past risk and returns, what comes out verified is what the exchange data supports, not what the marketing said. This is the audit before the verifier looks, so the first verification cycle is the last.

Write for a quote → How the pack works ↓

One person runs every check. One person signs the audit note. Named, and liable for it.

Audit note, per algoFORMAT
Look-ahead / point-in-timePASS
Survivorship
The universe is missing 41 names that stopped trading.
FLAG
Cost realism
The edge survives to 45bps; 20 was assumed.
FAIL
Selection / deflated SharpeFLAG
Ledger gaps and restatementsPASS
Reconciliation to exchange dataPASS

Illustrative, not a client note. Every check gets a verdict, a measured magnitude where possible, and a fix, before anything is submitted.

Where it stands

The framework is live, and the enrolment window has closed

An investment adviser, a research analyst or an algo provider who wants to show performance enrols with the verification agency, submits the recommendations or trades through its data centre by API or file, and the data centre validates them against exchange and clearing data before computing the metrics. The agency does not reconcile a record for you, and every algo whose performance is claimed now has to carry a verified record of its own. A record that comes out short costs the cycle and the claim you wanted to make.

4 May 2026CARE Ratings operational as PaRRVA, NSE as its data centre, under SEBI's circular of 29 April 2026 (HO/38/14/(4)2026-MIRSD-POD/I/10557/2026).
3 September 2026Enrolment deadline for advisers and research analysts, after the extension of 3 August. Not enrolled means no certified past performance can be communicated to clients.
3 May 2028From this date only verified metrics may be shown at all. Exchange-empanelled algo providers are covered through the stock brokers that run their algos.
Scope

The pack, per algo or track record

01The audit

Look-ahead, survivorship, cost realism against measured costs, the selection effect across every version you tried with the deflated Sharpe beside the raw one, gaps and restatements in the ledger. The same protocol as a Backtest Audit Check, applied to the history you intend to have verified.

02The metrics the framework asks for

Computed from your ledger rather than from a spreadsheet: return, risk-adjusted return, maximum decline, consistency over the period, per algo, with the working shown.

03The submission, reconciled

Your recommendations or trades in the shape the data centre's API or file upload expects, reconciled beforehand to what the exchange and clearing data will show, with a written audit note stating what was checked, what was found and what was corrected.

04Problems flagged before submission, not after

A record that would not survive the verifier's comparison is told to you first, in writing, with the fix. Delivery within ten working days of a complete ledger; missing data pauses the clock, and I ask once.

What this is not. It is not the verification: only the agency verifies, and no preparation changes what your trades or recommendations were. It is not a guarantee of the outcome, and it is not investment advice or a rating of your strategy. It is an honest account of what the history is, before someone else reads it.

Price

One price per record. The first three clients pay less.

Verified-performance pack

Rs 1,00,000 per algo or track record

The first three clients take it at Rs 75,000. Several algos from one provider are quoted together. Payable in advance, invoice on request, GST as applicable.

The audit inside this pack carries the same guarantee as a Check: if the audit note does not go beyond what you already wrote in the intake about your own record, the fee comes back.

Write for a quote →

Who this is for

Algo providers empanelled with the exchanges, white box or black box, whose verification runs through the stock brokers that carry their algos, and who want to show performance per algo rather than make an unverified claim.

Research analysts and investment advisers who want to publish verified performance rather than an unverified claim.

Anyone enrolling for the first time who would rather know a record's weak points from me than from the verifier.

The case

Why me

159Of my own strategies tested in public, 146 killed with the cause named, every row git-timestamped in the graveyard.
5Retractions of my own results admitted in print. The method that catches mine is the one that will catch yours.
2,136Configurations counted against my own book in a public self-audit, the deflated number printed beside the raw one.

The Backtest Audit desk and its published method run the same checks this pack runs.

To start

Write with the count, the period and the ledger's form

Email ayan@financebroski.com with the number of algos or records, the period each covers, and the form your ledger is in. You get the intake and a quote back within one business day.

Ayan Jain, Jaipur, India. Independent; not affiliated with SEBI, any exchange, or the verification agency. Nothing here is investment advice. Terms · Privacy.