This page is the companion to the backtest. The backtest is the evidence — six years of past UPSC papers, every claim verifiable. This page is the "how" behind the numbers, in plain English. There is no formula here. There never will be. There is also no engineering jargon. Just the working of an editorial filter that takes about 1,000 PIB releases a week and returns the ten we think you should read.
Aditya Tiwari Sir has been grading Prelims-style mocks since 2018. Two complaints kept surfacing in Law Optional batch reviews — students wasting six hours a week reading PIB notes that nobody would ever ask about, and missing the two or three releases that would, in fact, appear on a paper eighteen months later. A weekly email to his batches, titled "Current Developments in Law", started as a private fix.
The list grew. It moved out of Law Optional and into GS-2 prep. By 2023 it was less a topic round-up than a working theory of which kinds of PIB releases actually carry exam weight, and which are noise. That theory is the engine. We have spent the years since 2018 calibrating it against what the next set of mocks revealed students were getting wrong. pibtracker is that filter, opened to every aspirant — not just the Law Optional batches.
A typical PIB publishing day yields between 80 and 200 releases across all ministries and languages. We pull only the English ones (about one in six are English; PIB publishes in twenty-three languages). That leaves roughly 50–80 releases a day to read. Our system scores each one and ranks them. We hand the top of the ranking to a human reader. The human edits, rewrites the gloss, flags the trap-watch notes, and approves the final ten. The list goes live at 10 PM.
The scoring itself is a layered judgment, not a formula:
It does not predict which week of news will matter. We don't know which Cabinet approval becomes a Prelims question eighteen months later — nobody does. What we can do is make sure that if a release matters, it will be in the top ten the night it dropped, and inside our searchable index for the full year after. That is the promise the backtest tests, and that the backtest reports honestly on.
It also does not cover everything UPSC asks. Roughly a third of Prelims Current Affairs questions trace to sources outside PIB — Reserve Bank circulars, SEBI consultation papers, Supreme Court judgments, international press, scientific journals. We are clear about this. The backtest excludes these by design, and our subscriber product flags the blind-spot topics each week so an aspirant knows what extra reading is needed. The pibtracker filter is the PIB layer of UPSC CA prep, not a replacement for the rest.
Reading the right PIB release is not always enough. UPSC has, especially since 2023, been writing questions that punish candidates who memorised a release without reading it carefully. Wrong attribution, swapped numbers, conflated names, deliberate location-trap. Every daily Top 10 carries a trap-watch flag where Aditya Tiwari Sir or Umang has noticed the kind of error that surfaces in mocks.
Because the weighting is the product.
If we hand it over in a PDF, every coaching institute clones it in a weekend, every aspirant doubts our reasons for charging ₹599, and our subscribers lose the advantage they paid for. What we are willing to do — and what the backtest and the Verification Sheet PDF demonstrate — is publish the outputs. Every rank we assign, every week we score, every UPSC question we did or did not catch, sits in public view next to the original PIB release one click away on the official pib.gov.in site. If an aspirant disagrees with a specific call, they can email Aditya Tiwari Sir and we re-examine in public.
That is the contract we offer: full transparency on results, intentional opacity on method.
The work is not infallible. Our 2025 cohort backtest, for example, started out catastrophically — only four of the year's twenty-one PIB-anchored questions had survived the first research pass, because the original PRIDs we had recorded were Hindi, Tamil, Marathi or Gujarati editions of the right release. Once the labels were cleaned the cohort grew to twenty-one questions, and the engine's true score on 2025 came out at 29% top-10 — still our weakest year, but for an honest reason (we explain it on the backtest page).
Every such correction is documented. The Verification Sheet PDF lists every CA question from the last six papers with its source release and our verdict — a fresh aspirant can sit with it for an hour and find something we got wrong. When that happens, write to defactoias.com@gmail.com with the question number and the release you believe is correct. We update in public, with a date and a reason.
The point of publishing the backtest and the verification sheet is not to claim we are right — it is to make it cheap for anyone to find out where we are wrong. — from a draft of our pre-launch note, 28 May 2026
The backtest page has the numbers. The Backtest Summary PDF is the printable version of those numbers, written for a teacher or a parent who wants to read it offline. The Verification Sheet PDF is the full audit trail — every question, every source, every verdict, 37 pages. The sample week is a free taste of what subscribers actually see at 10 PM each night.
The thread underneath all four is the same. A patient, repeated reading of a single newswire (PIB) with a point of view about what carries exam weight, opened to every aspirant who cannot get to a De Facto IAS batch in person.
The backtest is six years of past UPSC papers honestly tested. The sample week is what subscribers see at 10 PM tonight. The free trial is seven days.