Methodology

Every CPI and IIP release has a consensus estimate — the market's best guess for the number before it is published. The surprise score measures how far the actual print landed from that consensus.

To make surprises comparable across months (when the usual miss size changes), we standardize: surprise σ = (actual − consensus) ÷ the historical standard deviation of past misses for that event type. A reading of +1.5σ means the print beat consensus by one and a half historical standard deviations; −1.5σ means it missed by the same margin. MPC decisions have no single numeric consensus, so they are analysed separately by policy action (hike, cut, hold).

For each event we capture the cross-asset price path at five checkpoints: T-60 (one hour before the announcement), T0 (the announcement), T+30, T+2H, and T+1D (one trading day later). The reaction at each window is the percentage move from the T-60 baseline.

For very recent events (roughly the last two months) we use intraday prices, so T-60, T0, T+30 and T+2H are genuinely distinct moments. For everything older, only daily data is available from the price source, so those four windows all fall on the same daily close — in that case the "reaction" is effectively the move from the prior trading day's close to the event day's close. This is flagged per asset on every event page.

The event study averages price paths across many events of the same kind. Each individual path is scaled to 100 at the event-day close (T0), then we take the arithmetic mean across all events. Scaling to 100 lets paths from different periods sit on a common footing.

For MPC, events are grouped by policy action — hike, cut, or hold. For CPI and IIP, they are grouped by surprise direction — above or below consensus. The shaded band is the ±1 standard-deviation spread of the individual paths around the mean, showing how dispersed outcomes are within each group (not a confidence interval for the mean).

Paths use trading days T-2 through T+2 around each event, skipping weekends and market holidays. All events are weighted equally.

The scatter plots each CPI/IIP print by its consensus surprise (σ) against the subsequent market reaction, with an ordinary-least-squares (OLS) regression line through the points. A steep, high-R² slope implies markets reprice sharply when prints miss consensus; a flat line implies surprises are largely already priced in.

Alongside the slope and R² we report the sample size (N) and the regression's p-value, so you can tell a real relationship from one that could be noise. We analyse CPI and IIP separately by default because their surprises are measured on different bases.

Suppose the consensus for a CPI print is 5.00% and the actual comes in at 5.40%. The raw miss is +0.40pp. If the historical standard deviation of CPI misses has been 0.25pp, the surprise is +0.40 ÷ 0.25 = +1.6σ — a clear upside surprise.

If Nifty 50 then falls 0.8% from T-60 to T+2H, that point lands at x = +1.6σ, y = −0.8% on the scatter. Over many such events the regression line shows whether big surprises tend to produce big moves in a consistent direction.

Prices come from a public market-data provider (Nifty 50, USD/INR, India VIX, and a gilt-fund proxy for the 10-year sovereign). Macro prints come from RBI press releases and MOSPI, with consensus from curated estimates and a market-data API. Live indicators are cached hourly.

The 10-year sovereign is represented by a gilt mutual fund (SBI Magnum Gilt). A gilt fund's value moves with bond prices, which move inversely to yields — so a positive "G-Sec" move means bond prices rose (yields fell), the opposite direction to a yield chart. It also carries small daily coupon accrual and fund-level frictions, so treat it as a directional proxy, not a precise yield.

These are raw indexed returns, not "abnormal returns" stripped of broad market moves, so part of any path reflects the market's general direction rather than the event alone. Indexed paths and regressions are descriptive averages; past reactions do not predict future moves. This is an analytical instrument, not investment advice.