The one sentence that kills a Canadian odd-lot trade

We bought the Air Canada odd-lot trade on 10 September. We sold it on 11 September, at a loss, before the offer had even closed. This issue is about the sentence that made us do it, because it applies to every Canadian buyback you will ever be offered — and because two more Canadian bids are already in the pipe behind this one.

Most of the 241 is the same story retold by twelve wires, plus bond tenders, interval funds and non-traded REITs. Two survived: a UK trust tender and a US cash bid.

The trade we killed

Air Canada, and Section 11

Air Canada is buying back C$800m of its own stock in a modified Dutch auction, C$29.00 to C$33.00, about 27.6m shares or 9.8% of the count, expiring 24 September at 5pm New York time. The circular carries the standard odd-lot clause: holders of fewer than 100 shares who tender all of them are taken up in full, ahead of proration. The stock had been trading near C$28, below the floor, because ordinary holders expect to be prorated on a 9.8% bid. Odd-lot holders are not prorated. At 99 shares the whole spread looked capturable: 3.4% at the floor, 17.7% at the top, over two weeks. We bought 99 at C$28.04.

Then Section 11, “Income tax considerations”. A Canadian company buying its own shares outside the open market creates a deemed dividend equal to the purchase price minus the paid-up capital of the share. Air Canada puts its paid-up capital at approximately C$10.60. For a non-resident that dividend is withheld at 25%, reduced to 15% under most treaties.

The gross spread at the floor is C$0.96 a share. The tax at the floor is C$2.76 a share. For a non-resident the tax is nearly three times the spread, and the break-even clearing price is C$31.12 — 7% above the floor. Air Canada’s 2025 bid, range C$18.50 to C$21.00, cleared at C$18.80: 1.6% above its floor, with roughly 99% of tendered shares taken up. Apply that to this one and you get about C$29.50.

The circular says it in plain language: holders who want capital-gains treatment should consider selling in the market instead of tendering. So we did. Sold 99 shares on 11 September at C$27.80: a realised loss of C$24.86 after both commissions, about 0.9%. It closed that day at C$27.53. The trade we avoided was roughly five times worse.

Who can actually run a Canadian buyback

We are in group 4. That is the whole story of the Air Canada trade.

The clean one

Arbutus: same country, opposite answer

Arbutus Biopharma is a Canadian company too. It is running a modified Dutch auction between US$5.00 and US$5.75 for US$230m of its stock, on Nasdaq, expiring 29 September at 5pm New York time. Holders of fewer than 100 shares who tender all of them are taken up in full, ahead of proration.

Its offer document states paid-up capital of at least C$8.42 a share — above the top of the US$ range once converted. No deemed dividend arises. Same country, same structure, and the tax sentence goes the other way. That is why the sentence is worth finding rather than assuming.

From Friday’s $5.12, 99 shares costs $507. Clears at $5.00 and you lose 2.3%; clears mid at $5.375 and you make 5.0%; clears at the $5.75 top and you make 12.3%. Cash around 6 October, so the middle case annualises near 83%.

The election is where people lose money. You can name a price inside the range, or elect “No Specified Price”. Naming a price feels careful and is expensive: name $5.75, watch it clear at $5.25, and your shares are not bought at all. No Specified Price cannot be excluded on price — you take the clearing price, whatever it is. In a Canadian bid the same box is called a Purchase Price Tender; do not go looking for an NSP label.

Two ways to ruin it. Buy 100 shares instead of 99 and you lose the priority and join the prorated queue. Sell in the market after electing and you are short against a tendered position.

The situations

Two things this table will not do. It will not quote you a Japanese price: our broker sells us no Japanese data, so the Kakaku.com bidding war — EQT has raised to ¥3,680 and extended to 29 September — sits in the watch list unpriced rather than in the table with a stale number. And it will not carry a row whose spread is gone: Austriacard closed Friday at exactly the €10.00 offer, so there is nothing left in it.

New this week

A 19.5% discount, and a way to be paid for a fifth of it

The Lindsell Train Investment Trust published its tender circular on 11 September. It will buy back up to four million shares, 20% of the register, and every holder may tender up to 20% of their own holding with that basic entitlement satisfied in full. The price is net asset value at the close on 9 October, less 5%, with the tender costs added back.

The shares trade at roughly a 19.5% discount to NAV, against a five-year average of 9.1%. Buy at a 19.5% discount, sell a fifth of it at a 5% discount, and that fifth is worth about 18% — roughly +3.6% on the whole position, in a month, with the remaining four fifths still owned at whatever discount the market then chooses.

What you are buying is the discount, not a price. The tender price floats with NAV, so a falling market takes the price down with it and leaves the 14.5-point capture intact. The vote is on 8 October and the board recommends it; elections close at 1pm London on 9 October, which is 08:00 New York, and the broker’s cutoff comes earlier than that. Cash on 16 October.

And a plain one. Copart agreed on 10 September to buy ACV Auctions for $10.50 a share in cash, $1.9bn, by tender offer. The stock closed Friday at $10.41, 0.9% under. The tender has not started yet — the offer document sets the clock — so this is a row to watch rather than a deadline to meet. At 0.9% it needs to close inside two months to be worth the settlement risk.

The execution table

New permanent section. Not what a deal pays — what the broker actually does with it. Every cell here was seen in a live account or read in the offer document; nothing is estimated.

The pattern, after five deals: the broker’s cutoff is not the offer’s expiry, and the gap runs from four hours to two days with no rule behind it. The published expiry is not the date that binds: the broker’s own cutoff is, and it can come hours or days earlier.

The dates that matter

Killed, and why

Air Canada, for us and for anyone else in group 4 — the tax is nearly three times the spread. It stays live and good for a Canadian registered account.

Virtus NFJ’s odd-lot shortcut. We looked for the usual clause and found the opposite, in terms: “this tender offer will not have any special pro ration provision for odd-lot tenders, which means that all odd-lot tenders (including Shareholders who own fewer than 100 Shares) are subject to proration.” Buying 99 shares of NFJ buys you nothing extra. The row survives on the discount alone.

Docebo. A Canadian buyback that closed on 8 September at a fixed US$25.00, with up to US$70m authorised. Holders tendered 99,332 shares — 0.4% of the company, US$2.5m of the US$70m. When a fixed-price bid sits at or below the market, almost nobody tenders, and the odd-lot clause is worth nothing. A reminder that the size of an offer says nothing about the size of the opportunity.

Austriacard — Friday’s close was exactly the €10.00 offer. VineBrook Homes, and the week’s crop of interval funds and private credit vehicles — real tenders, no listed shares to buy. Solidcore’s US$1.2bn on-market tender — the listing is not one a retail broker reaches.

And the standing one: JARA. A 75.3p liquidation distribution against the last printed price looked like +7%, until we asked the exchange for trading status rather than price: suspended, on the day the liquidation was entered. A stale last price is not a tradable price. Every row above is checked for status and volume, not just quote.

The book — our own money

Open: ABUS 99 at $5.05, tendered at No Specified Price · ARI 300 at $6.92 · Delivery Hero 50 at €37.04 · Oma 300 at €17.07, elected for cash at €17.20.

Closed this week, both at a loss. Air Canada: bought 99 at C$28.04 on 10 September, sold at C$27.80 on 11 September, −C$24.86. Zevia: 2,000 shares sold on 10 September at $1.272–1.285, −$322 — a speculative satellite that never became a special situation and was cut in a general clean-up of the book.

Two losses in one week, printed at the top of the letter rather than the bottom. The Air Canada one is the expensive kind to avoid and the cheap kind to take: C$25 to learn a rule that will be worth more than that on the next Canadian bid, and there is one every month.

Sources

Added 17 September 2026: the primary filing or company release behind each row. Times converted to New York time on 17 September 2026, which also corrected two day names (29 and 30 September).

Correction, 17 September 2026: the Air Canada buy filled at C$28.03; the C$28.04 printed above included commission. The −C$24.86 result is unchanged.

99 Shares is information, not advice. We hold or intend to hold positions in situations we write about, and we say which. Every number traces to a primary filing or a broker quote, and we print the date when the number is not from today. Nothing here is a recommendation to buy, and tax treatment depends on where you and your account live — the whole point of this issue.

99shares.org · hello@99shares.org · Unsubscribe at any time.