The engine reproduces the insurer’s own illustration — to the rupee
Every insurance calculator on the internet claims to be accurate. This page says exactly what we compute, where the numbers come from, how the arithmetic is checked, and what we deliberately refuse to do.
How a policy is analysed
Four things happen between the brochure and the number on your screen.
- 1
The policy is data, not code
Each policy is a JSON configuration holding its own guaranteed addition rates, surrender value factors, paid-up rules and premium tables, taken from the insurer’s official brochure. Nothing about a specific policy is written into the calculation engine, so adding a policy cannot change the answer for any other one.
- 2
Every configuration is validated before it can be used
Configurations are schema-checked when the application starts. A malformed factor table stops the build loudly rather than quietly producing a plausible, wrong number — which is the failure mode that matters here, because a surrender value nobody can tell is wrong is worse than one that never appears.
- 3
Each quantity is computed by an isolated module
Guaranteed Surrender Value, Special Surrender Value, paid-up value, guaranteed additions, death and maturity benefit, loan eligibility, revival cost, inflation adjustment, opportunity cost, break-even and XIRR are separate pure functions composed into one analysis. None of them reads anything but its own inputs.
- 4
The engine is tested against the insurer’s own illustration
The benefit illustration printed in each brochure is copied verbatim into the policy configuration and asserted as a golden test: for the same inputs, our engine must reproduce the insurer’s published figures to the rupee. Any change to any calculation module has to keep those green before it can land.
What “to the rupee” means
Life insurers publish a benefit illustration in every brochure: for a stated sum assured, term and premium, here is the guaranteed maturity value, the surrender value at each year, and the death benefit. Those tables are copied into the policy configuration exactly as printed, and a test asserts that our engine, given the same inputs, returns the same figures.
Not “close to”. The same. If a change to any calculation module moves a single one of those numbers by a rupee, the test fails and the change does not ship. It is the only check we know of that distinguishes a correct policy engine from a confident one.
Keep, make paid-up or surrender
For a policy whose figures reproduce its brochure, the three paths open at your policy year are set side by side. Each figure comes from the same modules as the rest of the analysis: the Guaranteed Surrender Value, the proportionate paid-up sum assured plus the Guaranteed Additions accrued to date, and the maturity benefit.
Each path carries two returns. From the start is the IRR of every premium you have paid against what the path pays. From today treats the surrender value you would give up by not surrendering as money put in today, adds the premiums still due, and sets them against what the path pays. Premiums already paid are the same on every path, so the second figure is the one that differs between them. Neither is a recommendation, and nothing on the page marks one path as preferred.
Paid-up figures count only the additions accrued at the paid-up date, so they are a floor. The actual surrender value is the higher of the guaranteed and the special surrender value; the special value is set by the insurer and not published, so it cannot be shown.
Term cover plus investing the difference
The panel takes the policy's annual premium and splits it in two: a term plan premium for the same sum assured on death, which you enter from a quote you have (we name no insurer and suggest no figure), and the remainder, invested at the start of each premium year at a return you choose. The return starts at the policy's own IRR, so with the defaults the only thing that changes is that the cover and the saving are separated.
Where the policy term is longer than its premium-paying term, the term premium for the later years is taken from the investment, so the money you pay out is the same both ways in every year. The comparison is shown at maturity and, for a single-life policy, on death in selected years. For a joint-life policy it is shown at maturity only: the first death pays out and the policy continues with premiums waived, which the split does not reproduce.
Tax is not modelled: a policy's maturity may be exempt where an investment's gains are taxed, so enter the return you expect after tax. The result is arithmetic on your assumptions, not evidence that either choice is better.
Where we stand
BullTimes Insurance Lab is educational and neutral. It does not recommend buying or rejecting any policy — it presents the numbers so you can decide.
What the Lab does
- Show the guaranteed cash flows of the policy you actually hold, computed from its own brochure factors.
- State the annualised return those cash flows represent, as an XIRR over the real dates.
- Restate the maturity value in today’s purchasing power, at an inflation rate you choose.
- Show what the same premiums would have become at a return you choose, and say plainly that the comparison is only as good as that assumption.
- Set keeping, making paid-up and surrendering side by side at your policy year, with the return on each, without marking any as preferred.
- Say when a figure is set internally by the insurer and therefore cannot be computed — Special Surrender Value being the main one.
What it will not do
- Tell you to buy, keep, surrender or make a policy paid up. That is your decision and it depends on facts we do not have.
- Rank insurers, or present a "best policy" list.
- Earn a commission on any policy. We are not an intermediary, we hold no IRDAI registration as one, and no insurer pays us anything.
- Pass your policy details to an agent, a broker or a lead-generation service.
- Present the opportunity-cost comparison as evidence that investing would have been better. It is arithmetic on a return you supply — change the assumption and the conclusion changes.
Why the neutrality is enforced in code, not just promised
The Lab uses a language model to explain a policy in plain English. Models are agreeable by default, and an agreeable model asked about an insurance policy will drift into telling you what to do. So the system prompt forbids a recommendation, and every response is then passed through a filter that strips verdict language before it reaches the page.
Belt and braces, because in a market where nearly every insurance “tool” is the top of a sales funnel, a stated principle that is not enforced is worth nothing.
The honest limitations
- • Special Surrender Value cannot be computed. It is set internally by the insurer and is usually higher than the guaranteed value. We compute the GSV exactly and say so rather than inventing an SSV.
- • Bonuses that are not guaranteed are not projected. Reversionary and terminal bonuses depend on the insurer’s future performance; anyone quoting them as a number is quoting an assumption.
- • The opportunity-cost figure is only as good as its return input. It compares your premiums against a rate you choose. It is a scenario, not a finding.
- • Only the policies we have configured are supported. Analysis requires the brochure factors; we will not estimate a policy we do not hold data for.
See it against your own policy
Enter your sum assured, term and premium. You will get the guaranteed cash flows, the real annualised return, and what it is worth in today’s money — and no advice.
How figures are sourced, checked and corrected across BullTimes: editorial policy.