TSX Trails the S&P 500 by Four Points in September, in Either Currency
The S&P/TSX Composite closed Friday at 35,800.89, up 0.26% on the day and 1.295% below where August finished at 36,270.50, measured in Canadian dollars. The S&P 500 gained 0.75% over the same stretch in US dollars. Put both indexes in the same currency and the distance holds either way: the TSX trailed the S&P 500 by 3.99 percentage points measured in US dollars and by 4.07 points measured in Canadian dollars. A Canadian who held US equities unhedged made about 2.77% in Canadian dollars in September while the home index lost 1.295%. September’s problem is a Canadian one rather than a seasonal one.
| Basis | S&P 500 | S&P/TSX Composite | Gap |
|---|---|---|---|
| In US dollars | +0.75% | -3.24% | 3.99 points |
| In Canadian dollars | +2.77% | -1.295% | 4.07 points |
Index levels, sector proxies and share prices in this piece are daily closes from Yahoo Finance, measured from the August 31 close to the September 25 close.
Three sessions of September remain, Monday the 28th, Tuesday the 29th and Wednesday the 30th, since the exchange trades through the September 30 statutory holiday. Closing the month flat takes 1.312% across those three sessions, so finishing higher takes more than that, on a price basis that excludes dividends. Anything less ends five consecutive positive months for the index. The month peaked early, at 36,633.10 on September 3, and Friday’s close sits 2.27% below that high.
September daily closes of the S&P/TSX Composite against the August 31 close of 36,270.50. Source: Yahoo Finance.
The five-month streak exists in Canadian dollars and not in US dollars
The run being defended is a Canadian-dollar run. In local currency the index gained in each of the five months from April through August, after a 4.58% loss in March.
| Month | Return in CAD |
|---|---|
| March 2026 | -4.58% |
| April 2026 | +3.65% |
| May 2026 | +2.37% |
| June 2026 | +0.25% |
| July 2026 | +1.06% |
| August 2026 | +2.96% |
| September 2026, 18 of 21 sessions | -1.295% |
Translate the same monthly closes at the Bank of Canada’s own daily USD/CAD rate and one month changes sign. June 2026 was a gain of 0.25% in Canadian dollars and a loss of 2.65% in US dollars.
| Month | Return in USD |
|---|---|
| April 2026 | +6.05% |
| May 2026 | +1.08% |
| June 2026 | -2.65% |
| July 2026 | +2.36% |
| August 2026 | +4.18% |
| September 2026 to date | -3.24% |
So the streak is five months long in Canadian dollars and two months long in US dollars. For anyone holding the index unhedged in US dollars, the run a Canadian investor is watching ended three months ago. The whole of the difference is currency: USD/CAD went from 1.3866 on August 31 to 1.4145 on September 25, which is the US dollar gaining 2.01% and the Canadian dollar losing 1.97% over the same stretch, two numbers describing one move. The rate is the Bank of Canada’s Valet FXUSDCAD series, the Bank’s own daily publication. What is pushing the loonie is a story in its own right, and we set out the mechanism behind it on September 25 in our piece on the Canadian dollar’s slide.
A streak that survives in one currency and died in another is a reminder that the currency a portfolio is measured in is a position, not a background condition. How much of that position to carry is a separate decision from what to own, and Canadian Depositary Receipts are one of the instruments built around exactly that question.
Closing a gap this size in three sessions is possible and uncommon
We measured every completed month of the index from July 1979 to August 2026, 566 months in all, taking the return from the close four sessions from the end to the final close. A final-three-session gain of 1.31% or better happened in 106 of those 566 months, or 18.7%. The median final three sessions returned +0.30% and the mean +0.25%, with extremes of +9.80% in November 2008 and -10.39% in August 1998.
Condition it on the hole September is in and the count tightens. Of the 49 months that entered their final three sessions down between 1.0% and 2.0%, seven finished the month positive, or 14.3%: August 1983, February 1986, December 1991, July 1993, June 2012, August 2019 and February 2022. Their median final three sessions returned +0.26% and their median full-month outcome was -1.05%.
Those 49 months needed anywhere from about 1.01% to 2.04% to get back to flat, and September’s 1.312% sits in the easier third of that band, so the conditional 14.3% understates its case while the unconditional 18.7% sits nearer its actual hurdle. One number to carry away: roughly one month in six, somewhere in the 15% to 19% range. That is a probability, not a forecast.
The return over the final three sessions of all 566 completed months since July 1979, with the 1.31% September needs marked. Source: Yahoo Finance.
September’s damage came from two separate engines
Sector moves below use iShares TSX sector ETFs as proxies.
| Sector | September to date |
|---|---|
| Technology | +2.05% |
| Financials | +0.77% |
| Consumer staples | +0.33% |
| Energy | -1.27% |
| Real estate | -3.06% |
| Materials | -3.17% |
| Gold | -3.46% |
| Utilities | -3.87% |
The first engine was a rate move. On the Bank of Canada’s own benchmark bond yield series, the Canada 2-year went from 3.01% to 3.39% between August 31 and September 24, up 38 basis points, and the 10-year from 3.73% to 3.97%, up 24 basis points. That is what hit the income names, and at the sector level it shows up as utilities down 3.87% and real estate down 3.06%. At the name level it is starker: TELUS fell 10.51%, BCE 9.15%, Enbridge 5.82% and Hydro One 4.74%. When yields reprice a whole cohort of income stocks in three weeks, the question that follows is not the index level but whether the payouts behind those prices are covered, which is the test we rank Canadian REITs on.
The second engine has nothing to do with bond yields. December gold fell 3.58% from August 31 to September 25, and gold equities fell 3.46% alongside it, which is why the gold and materials sectors sit in the same part of the table as the rate-sensitive ones for an unrelated reason. Kinross was the worst performer among the 37 Canadian large caps we screened, down 16.45% on its own 2027 guidance cut, so that one is a company event rather than a sector read.
Technology’s +2.05% is the most misleading line in the table, because it is one stock. Celestica gained 24.25% in September while Constellation Software fell 8.47%, CGI 8.16% and OpenText 7.95%, with Shopify down 1.50%. That is the same concentration we documented on September 25 in our look at Canada’s technology ETF, and it is worth knowing before reading a green sector line as a broad advance.
The banks are the reason the index is only down 1.295%. Scotiabank rose 4.31%, BMO 3.27%, CIBC 2.17%, TD 1.80% and Royal Bank 0.79%, with National Bank the only one lower at -0.21%, and financials as a sector finished up 0.77%. Manulife at +5.63% and Dollarama at +4.86% were the other large caps that held, against Brookfield at -8.64% on the other side of the ledger.
Three sessions with one macro release and a quarter end in them
Sangoma Technologies reports after Monday’s close, with a conference call at 5:30 PM Eastern. The only scheduled Canadian macro release inside the window lands Tuesday at 08:30 ET, when Statistics Canada publishes real GDP by industry for July. Its June GDP by industry release put June growth at 0.3% month over month with 13 of 20 industrial sectors contributing, and the second quarter at 0.9% after 0.1% in the first, with 17 of 20 sectors expanding. That release also carries StatCan’s own advance estimate for Tuesday’s number: “Advance information indicates that real GDP was essentially unchanged in July. Increases in real estate and rental and leasing and professional, scientific and technical services were offset by decreases in retail trade and manufacturing.”
Wednesday is quarter end, and Q3 closes with the index up 2.71% for the quarter as of Friday.
What the market has priced into the October 28 decision
The Bank of Canada’s policy rate is 2.25%, where it has sat since October 30, 2025 and where it was most recently left at the September 2 announcement. The next scheduled announcement is October 28, per the Bank’s key interest rate page. On the September 25 settlement, Montreal Exchange one-month CORRA futures carried 11.63 basis points of increase into that decision, an implied rate of 2.37%. That is 47% of a quarter-point move, which is less than half of one. The 38 basis point move in the two-year yield through September is not in tension with that small figure: a two-year yield discounts a policy path over two years while the October contract prices a single meeting, so what repriced in September was the path rather than the October decision. A move higher is priced as possible, not likely, and no change remains the single most likely outcome on the day.
A broken streak has carried little information about the year that followed
There have been 24 completed runs of five or more consecutive positive months in the index between 1979 and 2026, the most recent being the ten-month run from May 2025 to February 2026 that broke in March. Comparing those against every month in history is not a fair test, because every one of them is a negative month that arrived after a long advance and the average month is neither. The comparison that isolates the break is five-month runs that continued into month six against five-month runs that broke in month six, measured from the close of month six in both cases, set beside two reference groups.
Median return, and share positive, from the close of month six:
| Horizon | Streak broke, n=24 | Streak continued, n=38 | After any negative month, n=220 to 221 | All months, n=556 to 567 |
|---|---|---|---|---|
| 1 month | +2.39%, 58% | +0.66%, 66% | +0.45%, 53% | +0.88%, 61% |
| 3 months | +4.66%, 75% | +2.84%, 82% | +1.85%, 61% | +2.52%, 66% |
| 6 months | +3.24%, 75% | +4.33%, 68% | +3.39%, 64% | +4.75%, 68% |
| 12 months | +4.03%, 57% | +0.14%, 52% | +8.03%, 65% | +8.45%, 69% |
At twelve months the first two columns fall to 23 and 33 observations, because the March 2026 break and the most recent continuations are not yet a year old, and the two reference columns shrink for the same reason as the forward window runs past the end of the data. The 38 continued observations also come from 13 distinct runs, since a long advance contributes one window for every month it survives.
Read across those rows and the sign of month six barely matters. The bottom row is the finding: both streak columns trail both reference columns over the following twelve months, and neither streak column is distinguishable from the other.
What the two columns share is more informative than what separates them, and what they share is five months of gains already behind them. The advance is the part of this pattern that carries information. The break is not. Dispersion inside the broken-streak set is wide enough to respect in either direction: the worst three months among them followed the May 1998 break at -27.13%, and the worst twelve followed the September 1989 break at -19.87%, while the best twelve returned +44.61%.
A down September would sit inside a quarter and a year that are both up
Whatever Wednesday’s close does to the monthly count, the proportions are worth holding onto. The index is 1.295% lower in September, 2.71% higher for the third quarter and 12.89% higher year to date in Canadian dollars, or 9.39% higher in US dollars. The median outcome for the 49 months that entered their final three sessions in the same hole was -1.05% on the month. A September that ends there costs the index a line in a table of consecutive monthly gains, a line that only ever existed in one of the two currencies this market is measured in.
Disclaimer: The content on bestcanadianstocks.ca is for informational and entertainment purposes only and does not constitute financial advice. Past performance is not indicative of future results. Always consult a qualified financial advisor before making investment decisions. Index and sector levels are S&P/TSX Composite and iShares TSX sector ETF daily closes from Yahoo Finance. The monthly, quarterly and year-to-date returns, the base rates over the final three sessions of a month, and the forward returns after a run of five or more positive months ends are computed by us from that daily close series, which begins in June 1979 and runs to the September 25, 2026 close. The September 22, 23 and 25 closes were checked against the closes verified first-hand for earlier pieces before any of it was drawn. US dollar translations of the index use the Bank of Canada’s own daily USD/CAD series, FXUSDCAD, from the Bank’s Valet API. The June GDP figures and the July advance estimate are quoted from Statistics Canada’s June 2026 release of gross domestic product by industry. The policy rate and the October 28 announcement date come from the Bank of Canada’s key interest rate page, and October rate pricing is derived from Montreal Exchange one-month CORRA futures settlements of September 25 with spot CORRA from the Bank of Canada.
