Methodology
How Gajamudra measures mutual funds
Every analytic on Gajamudra is computed from disclosed monthly portfolio holdings using the definitions below. We publish them in full so the numbers can be checked, compared and cited. This page is the authoritative reference for our metrics.
Conviction & concentration
How focused a portfolio is and how strongly the manager backs their best ideas.
- Gini coefficient
- The inequality of position weights across the portfolio (0 = every holding equal-weighted, 1 = all weight in one stock). Gajamudra's primary conviction metric. Conviction labels: Extreme (Gini > 0.50), High (> 0.38), Moderate (> 0.28), otherwise Low.
- HHI (Herfindahl-Hirschman Index)
- The sum of squared position weights, expressed in basis points. A higher HHI means weight is concentrated in fewer holdings.
- Top-5 / Top-10 / Top-20 weight
- The share of the portfolio held in the largest 5, 10 or 20 positions - a direct read on concentration.
- Maximum position & max-to-equal-weight
- The single largest holding's weight, and how many times larger it is than an equal-weight position (equal weight = 100% / number of holdings). A 5x ratio is a very high-conviction single bet.
- High-conviction bets
- The count of holdings held at more than twice their equal-weight share - positions the manager has deliberately oversized.
- Position-weight standard deviation
- The dispersion of position sizes; higher dispersion indicates a more selective, conviction-led book.
Benchmark comparison
How different a fund is from its designated SEBI category benchmark index.
- Active Share
- Active Share = 0.5 x the sum of the absolute differences between each stock's fund weight and its benchmark weight. Labels: Closet Indexer (< 60%), Moderately Active (60-80%), Highly Active (> 80%).
- Weight Overlap
- The sum of the minimum of (fund weight, index weight) per stock - the percentage of portfolio weight shared with the benchmark. It is the complement of Active Share.
- Cosine similarity
- The directional similarity between the fund's and the index's weight vectors (100% = identical direction, 0% = completely different). Lower values indicate a more differentiated, active portfolio.
- Jaccard similarity
- The overlap of the two stock universes regardless of weight: stocks held by both, divided by stocks held by either. High Jaccard with high Active Share means the same names at very different weights.
Sector exposure
Where a portfolio is positioned at the industry level.
- Sector weights
- Portfolio weight aggregated by the disclosed industry / sector of each holding.
- Sector concentration (sector Gini)
- The Gini coefficient applied to sector weights - how concentrated the fund is across industries rather than individual stocks.
AUM & capacity stress
Whether growing assets are quietly degrading how a fund is run.
- AUM (approximate)
- Assets under management approximated each month from the sum of disclosed holding market values. Reported in rupees crore; treat as an approximation, not the official AUM.
- Capacity stress score
- A composite of Spearman rank correlations between AUM over time and four signals: falling conviction (Gini), rising holdings count (forced diversification), compressing top-5 weight, and shrinking average position size. Labels: No Stress Detected, Early Signs, Moderate Stress, High Stress.
Crowding: consensus & contrarian
Where a whole category agrees, and where individual managers break away.
- Consensus pick
- A stock held by at least half the funds in a SEBI category. Heavily crowded names carry herding risk if sentiment turns.
- Contrarian bet
- A stock held by exactly one fund in the category - where a manager is taking a differentiated position against peers.
Composite scorecard
A single, peer-ranked score combining conviction and activeness.
- Conviction score
- A percentile rank within the SEBI category, weighting Gini (35%), Top-5 weight (25%), high-conviction bets (20%) and HHI (20%). 0-100, higher is more conviction-led.
- Activeness score
- A percentile rank within the category, weighting Active Share (50%), Cosine similarity (30%, inverted so lower cosine scores higher) and Jaccard (20%, inverted). 0-100, higher is more differentiated from the benchmark.
- Composite score
- The average of the conviction and activeness scores (50/50). All ranks are computed within the fund's own SEBI category, never across categories.
Data & filters
The disclosed data the analytics are built on, and how it is cleaned.
- Source data
- Monthly mutual fund portfolio disclosures (per holding: fund, date, ISIN, stock name, industry, weight and market value) plus NAV history. Figures reflect the latest disclosure month shown on each page.
- Filters applied
- Derivatives are excluded (futures flag), zero-weight lines are dropped, and Indian securities (ISIN starting "IN") are used for equity analytics. Weights are normalised per fund-month to sum to 100%.
- Market-cap classification
- Large / mid / small cap follows the AMFI classification (ranks 1-100 large cap, 101-250 mid cap, the rest small cap), refreshed as AMFI updates its list.
- Benchmark proxy
- Each SEBI category is compared against its designated index using a tracking index fund's disclosed holdings as the benchmark, in the same format as active-fund holdings.
Run these metrics yourself
Every definition above powers a live analysis in the Gajamudra Analytics Engine - conviction reports, vs-benchmark analysis, the composite scorecard, AUM stress, consensus/contrarian and category leaderboards. New to the terms? Start with the knowledge center.
For research and education only - not investment advice. Mutual fund investments are subject to market risks; read all scheme-related documents carefully.