Investing
I firmly believe that financial literacy is a great way to build wealth. Common wisdom says that most people with a long investment horizon should just buy a low-cost index fund. But for those who, like me, simply enjoy researching businesses and learning their ins and outs, there are greater profits in stock picking (selecting individual stocks) and managing a portfolio.
This is my story. I started investing on Mar 14, 2024, 2 years and 6 months ago, with index funds and a lot of cash, and over time I moved to picking my own stocks. Below I tell why I bought each stock, why I sold it, and how my results compare with the S&P 500 (the 500 largest US companies). For the S&P 500 I use SPY, an index fund that follows it, with its dividends reinvested. Returns are in Canadian dollars, as of Oct 9, 2026. On purpose, I show percentages only, never amounts of money. This is not investment advice (see the disclaimers).
At a glance
My money grew +30.0% per year.
If I had put the same money into SPY on the same days, it would have grown +23.5% per year. Today my portfolio is worth 8.2% more than that SPY copy would be.
I did not put all my money in on day one: I added savings over time. So these are money-weighted returns (also called XIRR): the yearly growth rate of the money I actually put in, counting when I put each part in.
From index funds to my own picks
I started very conservative, mostly because I was afraid of my limited knowledge. In 2024, about half of my portfolio was cash and bonds, and my own picks were only about 7% of it on average. In 2025, my own picks were about 12% of the portfolio on average. As my picks proved that they could beat SPY, I trusted my research more, and I started to put a larger part of the portfolio into them. In 2026 I moved almost everything into my own picks: they averaged 56% of the portfolio over the year, and they are 97% of it today. The rest is index funds (3%) and cash and bonds (0%).
The chart splits my portfolio into three kinds of holdings. Cash and bonds are the safest part: account cash, CMR (a money market fund) and 60% of XCNS (a balanced fund). Index funds buy all the stocks of a market at once: the rest of XCNS, and funds like XEQT and VTI. My own picks are single companies that I chose after my own research, plus the IBIT and DRAM funds.
Against SPY with the same deposits
The chart shows what each dollar I put in was worth over time: 1.0× means I had exactly what I put in, and 1.5× means 50% more. The dashed line is the same money put into SPY on the same days. The blue band is the time when much of my money was in cash and bonds.
For most of 2024 and 2025 I was behind SPY with the same deposits, at worst by 14.5%, while about half of my money was in cash and bonds. Today I am 8.2% ahead: every dollar I put in is worth $1.46, against $1.35 in SPY.
This is the part I am most proud of. From May 28, 2024 to Apr 7, 2026, about 53% of my money sat in cash and bonds, and I still matched SPY. Each dollar I had put in was worth $1.15, against $1.16 in SPY. So I got the same return for much less risk. My portfolio went up and down much less than SPY (a volatility of 7% against 18%), and its beta was 0.26: when SPY moved 1% in a day, my portfolio moved only about 0.26%.
- My portfolio, 1.46×
- SPY with the same deposits, 1.35×
My own picks against SPY
This chart counts only my own picks, from the first one I bought. The money I paid for each pick goes in, and each sale and dividend comes out. The dashed line puts the same money into SPY on the same days. My picks grew 65.2% a year (money-weighted), against 30.0% for SPY. Today every dollar I put into them is worth $1.29, against $1.13 in SPY.
- My own picks, 1.29×
- SPY with the same money, 1.13×
Risk
| Measure | Portfolio | SPY |
|---|---|---|
| Volatility (per year) | 17.1% | 16.5% |
| Beta (against SPY) | 0.50 | 1.00 |
| Max drawdown | −20.3% | −19.4% |
Volatility is how much the value goes up and down: lower is a calmer ride. Beta is how much my portfolio moved when SPY moved: 0.50 means about half as much. Max drawdown is the largest fall from a high point to a later low point. All three come from daily values.
Calendar years
| Year | Portfolio volatility | SPY volatility | Cash and bonds (average) | Own picks (average) |
|---|---|---|---|---|
| 2024 (Mar 14, 2024 to Dec 31, 2024) | 6.9% | 13.2% | 50.7% | 7.2% |
| 2025 | 7.3% | 20.5% | 47.5% | 12.3% |
| 2026 (Jan 2, 2026 to Oct 9, 2026) | 29.1% | 13.7% | 16.9% | 56.4% |
Closed positions
Of my 8 closed own picks, 6 made money. Counting index funds too, 17 of 19 closed positions did.
"Per year" is a position's return scaled to one year. It is not a return I really made in a year: it is how I compare trades of different lengths, because I want the most return on the money I put to work, and a 30% gain in one month is not the same as 30% in ten years. "Max weight" is the largest share of the whole portfolio the position had while I held it.
This hides the index funds, the money market fund (CMR) and the balanced fund (XCNS) in both tables, so you see only my own picks. Returns include dividends. Weights are shares of the whole portfolio, also when index funds are hidden.
| Asset | Bought | Sold | Length | Max weight | Return | Per year |
|---|---|---|---|---|---|---|
| XEQT | Apr 2, 2026 | Oct 1, 2026 | 5 months 29 days | 25.7% | +13.1% | +28.1% |
| META | Jul 30, 2026 | Sep 24, 2026 | 1 month 25 days | 19.1% | +45.8% | +1,065.1% |
Thesis: The most dominant social network and advertising platform on earth, at its lowest price in three years on fears about AI spending and a pending lawsuit. I judged that the AI spending would pay off and that the lawsuit would be settled and the business would move on. I sold when Muse was announced: the stock went up so much, so fast, that I preferred to take the profit and rotate than to let it compound. | ||||||
| DRAM | May 5, 2026 | Jul 30, 2026 | 2 months 25 days | 39.0% | +3.1% | +13.8% |
Thesis: My AI thesis (see NVDA below), from the point of view of memory, the bottleneck of AI. Memory chip makers were the highest-quality leveraged bet on it. I closed the position after the Korean stock market fell 30% when leveraged trades got margin calls in a chain reaction. I made money, but for the risk I took, this trade underperformed. | ||||||
| XEC | Apr 29, 2025 | Jul 21, 2026 | 1 year 2 months 22 days | 38.3% | +52.1% | +40.8% |
Thesis: I overweighted emerging markets because their valuations were much more attractive than those of US stocks. | ||||||
| XEQT | May 26, 2025 | Jul 21, 2026 | 1 year 1 month 25 days | 25.7% | +25.4% | +21.6% |
| IBIT | Jun 6, 2025 | May 11, 2026 | 11 months 5 days | 3.1% | −28.5% | −30.3% |
Thesis: A small speculative bet to diversify a little away from stocks. I closed it together with LRN and SNWV in a portfolio reset, to free capital for the DRAM bet. | ||||||
| XEC | Apr 15, 2024 | May 11, 2026 | 2 years 26 days | 38.3% | +34.1% | +15.2% |
Thesis: I overweighted emerging markets because their valuations were much more attractive than those of US stocks. | ||||||
| CMR | May 23, 2024 | May 5, 2026 | 1 year 11 months 12 days | 64.9% | +3.1% | +1.6% |
| SNWV | Oct 6, 2025 | May 5, 2026 | 6 months 29 days | 1.4% | −59.5% | −79.1% |
Thesis: An excellent razor-and-blades product for wound therapy, with better results and cheaper unit economics than graft treatments. Then the government changed the reimbursement codes for graft treatments: most clinics that offered the treatment went bankrupt or became cautious and ordered less of Sanuwave's equipment. I overpaid for this business and mispriced the systemic risk of the reimbursement codes. My first mistake. | ||||||
| LRN | Jan 12, 2026 | May 1, 2026 | 3 months 19 days | 6.9% | +34.7% | +171.0% |
Thesis: An online school operator priced for low growth, whose stock fell 55% after fraud allegations and some temporary problems scaling its platform. At that price I could buy the healthy segment at fair value and get the segment with the fraud allegations for free, and the cash on the balance sheet easily covered the largest possible fines (if there was fraud at all). | ||||||
| XCNS | Jul 3, 2024 | Apr 2, 2026 | 1 year 8 months 30 days | 17.6% | +12.0% | +6.7% |
| PXT | Aug 1, 2024 | Mar 26, 2026 | 1 year 7 months 25 days | 7.5% | +58.3% | +32.1% |
Thesis: The largest private oil and gas producer in Colombia, near decade lows with an almost 10% dividend yield, a clean balance sheet and temporary operating problems. I sold after the Hormuz closure in the Iran war sent oil prices very high, at the same time as it announced the purchase of Frontera's Colombian assets. | ||||||
| TFPM | Jun 16, 2025 | Mar 25, 2026 | 9 months 9 days | 2.3% | +35.7% | +48.5% |
Thesis: A precious metals streaming and royalty company, which finances the miners in exchange for royalties on future gold or a share of the volumes, bought at a time of peak trade uncertainty, currency debasement and central bank gold buying. | ||||||
| XEQT | Sep 29, 2025 | Oct 14, 2025 | 15 days | 21.1% | 0.0% | +0.2% |
| MRE | Jun 16, 2025 | Sep 19, 2025 | 3 months 3 days | 2.4% | +33.3% | +201.8% |
Thesis: A Canadian auto parts maker at a 15-year low after Trump's tariff announcements: the price was so low that it traded at liquidation value. | ||||||
| XEF | Apr 15, 2024 | Aug 2, 2024 | 3 months 18 days | 19.9% | +1.3% | +4.3% |
| XIC | Apr 15, 2024 | Aug 2, 2024 | 3 months 18 days | 36.2% | +0.8% | +2.7% |
| XUU | Apr 1, 2024 | Jul 2, 2024 | 3 months 1 day | 45.6% | +4.6% | +19.3% |
| VTI | Mar 15, 2024 | May 28, 2024 | 2 months 13 days | 97.9% | +0.7% | +3.5% |
Open positions
| Asset | Held since | Weight | Return |
|---|---|---|---|
| NVDA | Mar 25, 2026 () | 28.2% | +26.4% |
Thesis: My AI thesis. At the company I work for, I saw engineers go from AI skeptics to spending thousands of dollars per month on AI credits. I bet on that demand from four points of view: NVIDIA as the enabler of the whole ecosystem, memory as the bottleneck (DRAM), Google as the spender with the clean balance sheet and the earnings power to sustain the spending, and TSMC as the neutral manufacturer that wins as long as the demand remains. Of the three that I hold, I estimate that NVIDIA is the only one that is undervalued (Google and TSMC are about fairly valued). The market fears that part of its demand is circular, because some buyers pay for the chips with debt or with financing from NVIDIA itself. It also fears a lower gross margin (71% to 72% at the lowest, from 75%) while the new Vera Rubin chips ramp up. I believe these fears are temporary: the demand that I see is real, because companies pay for the AI that their engineers use every day, and margins usually recover after a new chip ramps up. | |||
| TSM | Jul 31, 2026 () | 15.0% | +11.8% |
Thesis: My AI thesis (see NVDA). TSMC is the neutral manufacturer: it is the only chip maker with the scale, the technology and the customer relationships to make the most advanced AI chips, so it wins whichever AI chip designer wins, as long as the demand remains. I estimate that TSMC is about fairly valued today. | |||
| GOOGL | Jul 22, 2026 () | 13.8% | +1.1% |
Thesis: My AI thesis (see NVDA). Google is the spender: of the hyperscalers, it has the fastest-growing cloud business (about 82% growth in Q2 2026, against about 43% for Azure and 37% for AWS), and it has the clean balance sheet and the earnings power to sustain a spending race. I estimate that Google is about fairly valued today. | |||
| OTIS | Oct 1, 2026 () | 12.6% | +2.1% |
Thesis: Otis maintains about 2.5 million elevators and escalators, the largest service base in the world. Service is about 65% of its sales and 91% of its segment profit, and building owners seldom change a maintenance provider that works. The price fell in 2026 after two guidance cuts, because newly hired mechanics took longer than planned to become productive, the market for new elevators in China kept shrinking, and the CEO announced her retirement. I judged these short-term problems for a recurring, high-margin service business. The next test is the Q3 2026 report on October 28, 2026. | |||
| BN | Oct 1, 2026 () | 9.1% | +1.7% |
Thesis: Brookfield manages other people's long-term money: pension funds, insurers and wealthy families give it money to invest in real estate, renewable power, infrastructure, private equity and credit, and Brookfield earns fees on it. It also owns an insurance business and many of these assets itself. The market prices it below management's own estimate of its worth (plan value), because the structure is hard to value from the outside and higher interest rates lower the value of its holdings. I expect part of the gap to close as its earnings grow and its merger with Brookfield Wealth Solutions closes, expected by the end of 2026. | |||
| BN (TSX) | Oct 1, 2026 () | 6.7% | +1.9% |
Thesis: The same company as BN above, listed on the Toronto Stock Exchange in Canadian dollars. | |||
| MRX | Mar 2, 2026 () | 4.4% | +64.7% |
Thesis: Marex is a clearing broker: the regulated middleman between traders and the clearing houses for energy, commodity and financial futures. Regulation makes clearing too costly for big banks and does not let clearing houses serve clients directly, so a few non-bank brokers like Marex keep sticky clients, earn about 30% on equity, and sell those clients trade execution on top. I bought it because it was cheap, at under 10 times earnings. The market feared that its profits would not last, and that lower interest rates would hurt them more than I judged possible. I also bought at a time of peak market volatility. Marex earns more when markets are volatile, so it was good diversification for my portfolio. | |||
| NOW | Apr 22, 2026 () | 4.3% | +37.1% |
Thesis: ServiceNow crashed with the other SaaS stocks in the "SaaSpocalypse", when the market priced in business disruption, seat compression and the fear that "everybody will build their own tools". I judged seat compression a short-term risk: businesses can move to usage-based pricing, and because ServiceNow is a system of record, no sane customer would risk hurting its operations just to save a little on the bill. Of all the SaaSpocalypse names, I chose ServiceNow because it has one of the best distribution channels for AI workflow automation. Its AI products grew from about $600 million in annual contract value at the end of 2025 to more than $1 billion in Q2 2026, about 6% of its revenue run rate, and the company's long-term target is 30% of its contract value from AI. | |||
| XEQT | Sep 22, 2025 () | 2.9% | +17.3% |
| CLX | Sep 16, 2026 () | 1.9% | −1.4% |
Thesis: Clorox is a very popular household products brand, trading at the same price as ten years ago because of concerns about a weak consumer, high interest rates, a CEO change and a change of its ERP (business management) system. I expect the end of the ERP change and a new CEO to change the market's view enough for the stock to recover partly, and long-term interest rates and the consumer to improve slowly until the stock recovers fully. | |||
| SUGR | Jul 3, 2024 () | 0.9% | +20.7% |
Thesis: My first stock pick. Sucro imports cane sugar from Latin America and sells slightly refined sugar in North America. This sugar is great for customers who do not need it white or granular: they pay a lower price for the same commodity. The stock price did not include the growth in refining capacity that management had guided for the next few years, and the company was not profitable under GAAP because of the inventories it had built up, which I expected to change as soon as its first refinery started. In November 2024, Beta San Miguel, a Mexican sugar group, bought 15.9% of Sucro and got the right to make a takeover bid for all the shares: from May 2027 at no less than 9 times Sucro's 2026 earnings per share, or from May 2028 at no less than 8 times its 2027 earnings per share. The bid is Beta San Miguel's choice, and I am holding to see if it comes. | |||