How the numbers are produced

Assumptions and methodology

Effective 22 October 2026. The version of the tax data and projection assumptions applied to any result is recorded with it, under About in the app.

How your plan is calculated

Everything is calculated on your device, one year at a time, from what you enter. There is no market data feed and no connection to any financial institution.

Tax and benefit rules

Federal and provincial brackets and the main personal credits, CPP and OAS amounts, the OAS clawback, and RRIF and LIF minimums come from published government sources. The version in use is shown under About. The tables are refreshed every January, when tax brackets are adjusted for inflation and benefit amounts move, and again every July, because a province can change its rates mid-year.

What your investments earn

Vidda works out its return assumptions using data from the Projection Assumption Guidelines, published each year by the Institute of Financial Planning and FP Canada Standards Council. Canadian financial planners use the same guidelines, and they are set deliberately to be free from bias rather than optimistic or cautious. Vidda applies the risk profile you choose across the guideline asset classes, then subtracts a typical fee for the kind of investments you told Vidda are in each account or, until you say, the fee the guidelines use in their worked example. The rate you see is what you would keep. Guaranteed deposits skip the blend and earn the guideline short-term rate, because a GIC is not a portfolio. You can replace any rate with your own. While an account has no rate of its own, Vidda shows you the figure it is using, so you see it before you change it, and clearing the field again puts the account back on it.

Inflation, and how long your plan runs

Inflation is 2.1% unless you change it, also from the guidelines. Your plan runs to age 95 by default. That age is Vidda's rather than the guidelines': they publish a separate planning age for a man, for a woman and for each kind of couple, and 95 sits between the two single figures at 65. Outliving your plan is the one mistake you cannot fix later, so under “Money needs to last to age” Vidda shows you their table and you can take the age that fits you.

The shaded band on Home

Not a probability. It is the same plan run with every growth rate 1.5 points lower and higher, drawn as one band from the weaker market to the stronger one, to show how much your answer depends on an assumption nobody can know. The shift is deliberately large, and nothing else moves: the same spending, the same taxes, the same plan, with only the growth assumption shifted.

The stress test

It measures how much bad luck your plan absorbs, not the chance of anything in particular happening. A thousand possible futures, each with its own good and bad years in its own order, its own inflation, its own answer for how long you live and its own late life care and medical costs, reporting how many lasted. Every future also assumes you would react: spending drops 10% in any year the portfolio sits 20% or more below its peak, and there is no way to turn that off, so the share that lasted is higher than it would be for a plan that never flinches. The guidelines' addendum publishes, for information, how far each kind of investment has strayed from its average year to year; Vidda blends those historical figures for the stress test, because the guidelines set no figure for it. How much inflation varies, and how it moves with returns, are Vidda's own choices, and the Past years setting replays real years as a check that sits outside the guidelines rather than something they set.

How the math gets checked

Everything above describes what Vidda assumes. This describes how we check that the math sitting on top of those assumptions does what the published rules say it should.

A second version, written separately. There is a second version of the projection math, written from the government sources rather than from Vidda's code, and sharing no code with the engine inside the app, though it follows the same modelling conventions. It covers one person in Ontario with four kinds of account: an RRSP, a RRIF, a TFSA and a non-registered account. Everything beyond that, from couples to the other provinces, is checked in the other ways below. The same 144 test households are run through both, producing more than 22,000 individual year by year figures to compare. Where the two disagree, one of them has read a published rule differently, and we find out which.

Sixty-five worked tax cases. Every province and territory at four income levels, plus a pre-retirement case for each, with each one computed outside the engine and compared to it to within two cents.

A wide sweep for the things nobody thinks to test. More than 13,000 generated households, covering every province and territory, every marital status, ages from 20 to 75, every account type, every kind of debt and every kind of goal. Each one is checked for results that cannot be true whatever the inputs: balances that go negative, tax that exceeds income, money that appears from nowhere, an estate that does not add up.

Plans worked out step by step. Six small plans run at zero growth and zero inflation, so the key years can be worked out to the dollar from the published 2026 tax and benefit tables, one step at a time, without running Vidda. The engine has to match them every time its tests run.

Tests that are themselves tested. A check that never fails is not a check. So we plant deliberate faults in the engine and confirm that the right check catches them. That exercise has told us something uncomfortable and worth repeating. One planted fault, present in 447 test households, went unseen by all six checks confirming that the engine's own report on one feature was internally consistent. A single check asking whether the work had actually been done caught it in all 447. A coherent report about nothing is still coherent.

What none of this means

It does not make a projection accurate about your future, because nothing can. It means the math does what the published rules say it should. What Vidda leaves out is a separate question, and it has its own page.

The part worth telling you

The second version has gone quietly out of date twice. The first time, a mid-year tax update moved two figures in one copy and not the other, and for a while the two disagreed in a way that looked exactly like a real dispute about the rules. Nothing caught it until the comparison was next run, and at that time it only ran when asked.

There is now a test that fails when the published figures in the two copies stop matching. When it fires, the rule is to open the government page and retype the figure rather than copy it across from the other side, because a copied number is bound to agree, and so proves nothing. The test can force the visit. It cannot keep the independence for us.

The second time, what changed was a rule rather than a figure, which that test cannot see. From age 65, Vidda now counts money you withdraw from an RRSP as pension income, which earns the pension credit and can be split with a spouse, on the assumption that each year's withdrawal is first moved into a RRIF and paid out from there. The engine inside the app took the rule and the second version did not, and for several days the two disagreed on 46 of the 144 test households, every one of them holding an RRSP, in the years from age 65. Again, nothing caught it until the comparison was next run. The rule was then worked out for the second version from the Income Tax Act and the Canada Revenue Agency's page on the pension income amount, rather than copied across, and the whole comparison now runs with the app's other automatic tests, so a disagreement between the two now fails those tests.

We are telling you this because it is the most useful thing on the page. Any tool can say it is carefully built. The question worth asking is what happened the last time it was wrong.

Figures as at 22 October 2026.

What this is not

Estimates, not advice, and not a forecast. Vidda does not model everything, and the list of what it leaves out is published rather than summarised.

None of this is required of us. We publish it because a projection you cannot interrogate is a number you are being asked to take on faith, and the whole point of this product is that you should not have to.

What Vidda does not model→