Tax-loss and tax-gain harvesting
Realising a loss to reduce tax is well known. Realising a gain to reduce tax sounds like nonsense until you notice that a retiree in a low-income year can often do it for almost nothing — and reset their cost base permanently.
Loss harvesting
Selling a holding that has fallen realises a capital loss. That loss offsets capital gains realised in the same year; anything unused can be carried back against gains in the three previous years, or carried forward indefinitely. The proceeds are then reinvested, and the portfolio’s exposure is broadly restored.
What harvesting a loss actually buys is deferral, not elimination. The reinvested position has a lower cost base, so a larger gain waits in the future. That is still worth doing if the tax saved today is at a higher rate than the tax paid later, and it is worth much less if the rates are the same.
Gain harvesting
The opposite move: deliberately selling an appreciated holding in a low-income year and immediately buying it back. The gain is realised at a low rate, the cost base steps up permanently, and future gains from that point are smaller.
The thirty-day rule does not apply, because it only denies losses. Repurchasing immediately after realising a gain is entirely permitted, which makes this the cleaner of the two operations mechanically.
The window is the one the meltdown episode identified, with one extra condition: the years between the last paycheque and the first benefit, when taxable income is unusually low, bracket space is going unused and — before 65 — there is no GIS for a gain to claw back.
Where Priya’s window is, and where it is not
Priya — 54, Ontario, single. $118,000 salary, $410,000 in her RRSP and $88,000 in her TFSA, planning to retire at 63.
Priya holds $79,394 in a taxable account at retirement — the only account where any of this applies. Her marginal rate while working is 37.2%. Only 50% of a gain is included in income, so a realised gain is priced at roughly half the rate of ordinary income.
Her ledger shows where the cheap years are. She retires at 63 living on her TFSA, so her taxable income is zero at 63 and again at 64: two years of bracket space, and no OAS or GIS yet to lose. From 65 she receives GIS, which a realised gain claws back, so those years are not cheap. And by 70 her non-registered money is spent, so there is nothing left to harvest. The window is the two years before 65 — not the retirement years in general.

The planner cannot price a harvest — it models a return per account, not the holdings inside it — but it can show the window. Against a terminal return where the whole accumulated gain is realised at once, a gain realised in the window is a meaningful difference on the same dollars.
Why this is advisory rather than measured
A projection models a return per account, not the individual holdings inside it, so it cannot know which positions carry gains or losses in any given year. It can identify the low-income window where gain harvesting is cheapest — which is the useful part — but it will not report a dollar figure for a technique it cannot actually simulate.
That is a deliberate limit. A number produced by a model that does not track the underlying mechanism would be more convincing than it deserves to be.
The ledger shows taxable income for every year of the plan, which is where the harvesting window is — the years between the last salary and the first mandatory withdrawal, sitting in plain sight.
Next: what happens when forced income exceeds what you actually need.
GlidePathEngine is an educational planning tool — not financial, investment, tax, or legal advice.