Smith Manoeuvre record-keeping, and the limits of a spreadsheet

Nearly everyone running the Smith Manoeuvre runs it in a spreadsheet, and for the first two years the spreadsheet is enough. This page is about what it does well, what it structurally cannot do, and what a draw has to carry with it if the interest claim is going to hold up in year fourteen.

What a Smith Maneuver spreadsheet does well

More than its critics allow. A competent sheet will:

amortise the mortgage correctly, including Canadian semi-annual compounding;
track the credit-line limit as it re-advances and the balance as it is drawn;
accrue interest on the line and total it by calendar year for the return;
apply an expected return to the portfolio and project a refund each spring;
show the whole thing against a baseline where you invested the same cash unlevered.

That is the projection, and it is arithmetic. If you want to check ours against yours, the same projection runs here with the assumption set printed under it, and the two strategy calculators on this site publish their engine version for the same reason.

What it cannot do

A spreadsheet holds numbers you typed. A record holds events that happened, with the documents that show they happened, in the order they occurred.

The difference is invisible while nothing is being questioned and total once something is. Consider the modest claim “the $9,412 drawn on 12 March 2019 bought 214 units of a dividend fund on 14 March 2019.” A spreadsheet row can assert that. What settles it is the credit-line statement showing the draw, the trade confirmation showing the purchase, and an account history with nothing in between that breaks the connection. Those are three different documents from two institutions, and the sheet is not one of them.

Four things follow, and each is a place a long-running sheet quietly stops being reliable:

It has no history. A cell that was edited last spring shows only what it says now. Which draw was reclassified, and when, is not recoverable from the file.
It records amounts, not uses. The interest claim rests on the use the borrowed money was put to, and a balance column carries no use at all.
It cannot see the account. A personal transfer into the investment account, a distribution that turned out to be a return of capital, a line payment made from the line itself — the sheet reports whatever was typed about them, or nothing.
It ages badly. Formulas break, tabs are copied, a laptop is replaced. The years you will be asked about are the ones furthest from the version you are editing.

What has to be on the record for each draw

The unit of evidence is a single draw, and each one needs six facts recorded when it happens rather than reconstructed afterwards:

the date and the amount of the draw;
the sub-account or credit-line segment it came from;
the investment account it went to;
what it bought, and when;
the document that evidences each of the two movements;
anything that later changed the answer — a sale, a switch, a return-of-capital distribution, a transfer out.

Six is not many. Multiplied by three hundred months, kept consistently while the rest of your life happens, and retrievable two decades later, it is the entire difficulty of the strategy and the reason the risks page puts bookkeeping alongside leverage rather than beneath it.

The events that break the chain

Not exotic events. Ordinary ones:

Interest capitalised onto the line during a month when cash was short, so the balance no longer equals the sum of the investment draws.
A draw that paused in a chequing account long enough to be mixed with salary before it was invested.
One convenient personal use of the line — a car, a renovation, a tax bill — after which one line carries two purposes and every subsequent payment against it has to be allocated between them.
A return-of-capital distribution reinvested without being recorded as what it was, quietly reducing the cost base and, with it, the borrowed amount still standing behind an income-producing use.
A sale, where the proceeds went somewhere other than a replacement investment.

What does a CRA review actually ask for?

A reviewer asks you to connect a specific borrowed dollar to a specific income-producing use, and to show it with documents from the time: the credit-line statement showing the draw, the trade confirmation showing what it bought, and the account history in between. Reconstruction after the fact is what fails.

If you would rather not maintain it yourself

This is the problem ACRU is built for, and it is worth being exact about what that means. The platform records each borrowed dollar against the use it funded as the transactions happen, keeps the working paper behind every figure, and flags the draws that break the chain. It describes the state of your records. It does not characterise your tax position, and it is not a substitute for the professional who does — what the platform traces, and where it stops sets out that boundary, and the paragraph the claim is made under is where the tests themselves live.

If a spreadsheet is working for you, keep it. The question worth asking is not whether the arithmetic is right today; it is whether the evidence will still be assemblable in 2041, and that question has the same answer for every one of these debt-conversion strategies.