Weights are liveThe weights are not fixed. They can be recalibrated, and doing so rewrites every score on the site.
How the BFI is calculated
The Bharat Friends Index is one number between 0 and 100, published for 194 countries. It is a weighted average of seven measured pillars, multiplied by a penalty where one applies, and truncated to a whole number. Nothing else goes into it.
Every figure below is the arithmetic the site actually runs. The worked example is Narnia, an invented country so that no real one’s numbers go stale here — you can follow it from the seven raw metrics to the published score without leaving this page.
Truncated, not roundedThe final step drops everything after the decimal point. A computed 82.90 publishes as 82, never 83.
The formula
score = truncate ( clamp0–100 [ Σ miwi × 0.95p ] )
- mi
- The seven pillar metrics, each 0–100 and each higher-is-better. Model-generated, refreshed regularly.
- wi
- The seven weights, summing to exactly 1.00. Set by the editors and read live, not fixed in the page.
- p
- The number of penalty conditions the country meets. p = 1 for an anti-India stance, and p = 2 if it also materially supports Pakistan against Indian interests, since the two stack. p = 0 for every other country — currently 1 of the 194 scored. Each unit of p multiplies by 0.95, removing 5% of the overall score — a proportion of the total, not 5 points off it.
Worked through: a country scoring 82.90 before penalties publishes 82 at p = 0, 78 at p = 1, and 74 at p = 2.
Seven inputs, one totalNo single measure decides a score. UN voting carries the most weight, and even that is only a fifth of the total.
The seven weights, and one country's arithmetic
They stackTwo penalties, each ×0.95. Both together give ×0.9025 — a 9.75% cut, not 10%.
The two penalties
After the weighted sum, two flat multipliers can apply. One is for a demonstrated anti-India stance; the other for material support of Pakistan against Indian interests. Each multiplies the total by 0.95, and a country can trigger both. They are the only part of the score that is not a weighted average, and they are large enough to move a country across a band boundary on their own.
An uneven ladderThe cuts are evenly spaced, but the bands are not: only the middle two span 20 points, and the ends are open. Nor are they evenly populated — Neutral / Developing alone holds 54% of the table.
The four bands
| Band | Score | Countries | Share of 194 | At or above the cut |
|---|---|---|---|---|
| Strategic Partner | 80–100 | 20 | 10.3% | 20 |
| Friendly Nation | 60–79 | 63 | 32.5% | 83 |
| Neutral / Developing | 40–59 | 104 | 53.6% | 187 |
| Watchlist / Caution | 0–39 | 7 | 3.6% | 194 |
Strategic Partner — Deep alignment across voting, trade and defence. Friendly Nation — Consistently warm ties with room to deepen. Neutral / Developing — Limited engagement, or a relationship still forming. Watchlist / Caution — Strained ties, or conditions Indian travellers should weigh.
What it cannot doA number is only as trustworthy as the account of its limits.
What this score cannot tell you
Which penalty fired
A country’s record notes only that a penalty applied, not which one. If a score was reduced, this site cannot tell you whether it was the anti-India stance, the Pakistan-support clause, or both — only that at least one applied.
Anything to the right of the decimal
The last step truncates. A country computed at 79.98 publishes as 79 and reads as a Friendly Nation; one computed at 80.01 publishes as 80 and reads as a Strategic Partner. The gap between them is three hundredths of a point.
How stable the inputs are
All seven metrics are generated by a language model and refreshed on a regular schedule. Where a refresh fails, the stored text reads “Metric is being updated” and the previous number stands until the next successful run.
That the weights will be the same tomorrow
The seven weights on this page are read live rather than fixed in the page. Recalibrating them recomputes all 194 scores at once, so any figure you cite from this site should be cited with the date you read it.
Machine first, vetted afterAutomation removes the ideological thumb on the scale; review catches what automation gets wrong.
Who publishes this
The Bharat Friends Index is a product of AksLabs AI, developed with our knowledge partner, FLAME University.
The methodology is deliberately automated. Explicit human judgement about which countries deserve to score well is exactly the bias the index is meant to avoid, so trade volumes, UN voting records and sentiment are read and scored by model rather than assigned by hand. Raw output without context can still mislead, so subject matter experts at FLAME University review the datasets and the algorithmic results for plausibility. Neither half is sufficient alone; the pairing is the method.