The India Credit Card Devaluation Index — September 2026: The 15 Most-Devalued Cards, Ranked

Our devaluation study answered the market question: Indian credit cards were cut 4.5 times for every improvement. This index answers the question that follows immediately — which cards, and which banks?

It is the ranked edition of the same dated, sourced change record, published monthly. Nothing below is opinion: every count is arithmetic over events you can browse, each with its date, its direction and its source. This is the September 2026 edition, computed over the 24 months to 16 August 2026. Editions are never edited after publication — when the record moves, the next dated edition carries the delta.

The 15 most-devalued cards in India (24 months to 16 Aug 2026)

A card is counted once for every distinct bank act that touched it. "Major" uses the record's own severity flag for that card — the kind of change that materially alters what a card returns, not a rounding of terms. A bank-wide act can be major for one card in it and minor for another, and this column takes each card at its own word rather than crediting it with a sibling's severity.

Distribution behind the table (cards with ≥1 cut, n=238): 1 cut ×50 · 2 ×51 · 3 ×49 · 4 ×33 · 5 ×19 · 6 ×17 · 7 ×4 · 8 ×6 · 9 ×6 · 10 ×2 · 12 ×1. Median cuts per moved card: 3.

Three things stand out. The top of the table is not where the famous cards are. The flagship premium cards that dominate headlines sit outside the top 15; the heaviest cutting landed on mid-market portfolios. India's most-devalued card took a cut every two months. Twelve distinct devaluations in 730 days is not a repricing — it is a product being dismantled in instalments. And nine of the fifteen most-recent cuts carry the same date, 15 June 2026 — bank-wide acts land on whole portfolios at once, which is exactly why the issuer view below counts decisions, not card-hits.

The issuer league table

Counting rule changes here: one bank decision counts once, however many cards it touched. "Cuts / card" divides distinct cuts by the issuer's active cards — the intensity number, and the ranking column. Issuers with fewer than five active cards are footnoted, not ranked; intensity is meaningless on a one-card portfolio.

Small portfolios, footnoted for completeness: Federal made 5 distinct cuts (4 major) against 1 improvement across 4 active cards; CSB 3 cuts (all major) against 2 improvements on a 1-card book; Utkarsh made 1 cut against 1 improvement on 1 card; Slice logged 1 improvement and no cuts.

The league reads cleanly. IndusInd is India's most aggressive devaluer: 30 distinct cuts across its 23 cards — 1.30 per card, 16 of them major — ahead of ICICI's 1.12 in second place, and it managed to touch every single card in its portfolio. HSBC is the only major issuer whose cards got better more often than worse — 7 improvements against 3 cuts — while Kotak, SC and RBL cut less often but improved nothing at all. And the biggest books are the calmest per card: SBI (0.50) and Axis (0.47) run large portfolios at a third of IndusInd's pace, a reminder that portfolio size and cutting intensity are different things — which is precisely why this table ranks intensity, not raw counts.

How the index is computed

The inputs are the same public record behind the changes tracker and the market-level study: dated events per card, each with a direction (devaluation, improvement, neutral), a severity flag, and a source. The index adds no model on top — no weights, no rupee estimates, no editorial adjustment. Where the two tables differ is only the counting scope, stated above each table. Severity is the record's own major flag, read per card in the first table and per act in the second: an act counts as major in the issuer league if it was major for any card it touched, which is the right reading of a bank decision but would overstate any single card, so the per-card table never inherits it. No rupee amounts appear in this edition: putting a ₹ figure on each cut requires replaying card states through a valuation engine, which is a different, harder claim than counting documented acts — we would rather publish the arithmetic that is airtight.

Dates deserve one caveat: we date an event from when it became publicly observable, which can be later than the day a bank first notified its own cardholders. "Dated 15 June" means our record dates it there — nothing more.

A recurring, citable index

This edition is frozen. The next one — October 2026 — will open with the exact delta: which acts landed since 16 August, which cards moved in or out of the top 15, and which issuers changed pace. That is the point of an index over a report: the record keeps moving, and the editions keep score.

Everything here is free to cite with attribution. The machine-readable edition lives at /data/devaluation-index.json — same numbers, same counting rules, stable schema — alongside the full event-level open dataset (JSON and CSV, CC BY 4.0). The event record behind every count is browsable, card by card, on the changes tracker. If you are covering a specific issuer and want the cut of this data for it, ask — we are happy to prepare it.