Reflections – Winners and a Loser
PDF
A year of getting some right, and getting some wrong
We’ve closed out the financial year. So we thought we’d revisit three holdings we wrote about during the year. Two are standout winners; the other is yet to deliver the returns we were expecting.
The two winners are Marex (NASDAQ: MRX) and Silicon Motion (NASDAQ: SIMO). The loser is Artivion (NYSE: AORT).
Each has something to teach us.
Marex – When the market gets it structurally wrong
Clearing broker Marex, which we had started buying in October 2025, was our most recent Global stock write-up in May 2026 (link).
In May, the stock was trading at less than 10x earnings. The market was pricing Marex like a lower-PE-rating cyclical commodities broker. The shares had also been hurt by a short report from investigative research company, NINGI Research, that alleged accounting irregularities and off-balance-sheet entities.
We, however, had a different view.
Marex, as one of ~60 Futures Commission Merchants (FCMs) globally, sits in the middle of one of the most protected pieces of financial market infrastructure in the world: soliciting and accepting buy and sell orders for futures and options contracts and holding customer funds to facilitate the trade.
Rather than a questionable commodity, we saw a structurally protected compounder benefiting from three tailwinds, all working in the same direction:
- Growth in exchange-traded volumes, which have been growing at high single digits for years.
- Share gain from banks retreating after international banking regulations Basel III and IV made clearing structurally uneconomic for large bank incumbents.
- Consolidation of the fragmented non-bank tail, which is seeing larger specialists, such as Marex, consolidate the smaller end of the FCM market.
We met with the executive vice chair of one of their biggest peers. He personally explained why this is such an attractive market and why they would benefit. However, what was most instructive was how complimentary he was of Marex as a high quality peer that would also benefit.
This was the key reason why we did more work.
We then went through the short report ourselves and with sell-side analysts. On the day of Marex’s March investor day, we caught the red eye from Denver to New York. We arrived early enough to spend time one-on-one with senior management. From these conversations, we concluded the short thesis lacked substance.
Since we wrote about Marex in May, the stock has increased from around $58 to $63. Since we started buying in October last year, it has roughly doubled.
That’s been helped by strong recent financial results. Marex’s Q1 2026 revenue was up 48%, and adjusted profit before tax was up 59% year-on-year. That’s comfortably ahead of the guidance the company had provided at its investor day just six weeks earlier.
Why we still own it: At 10-11x forward earnings, we continue to see plenty of value in a business generating EBIT growth in the teens, and with a long runway for both organic and inorganic market share growth.
Peers like Interactive Brokers and CME are still trading at 20-30x. StoneX, the closest listed comparable to Marex, now trades at ~19x. The gap between those companies and Marex is closing, but it is still quite wide.
Marex also plays a unique role in our portfolio. It is one of the few positions that actively benefits from periods of elevated market volatility rather than being threatened by it.
What we take from this: Our ability to talk with industry leaders in global markets is a key edge. When the market is punishing a stock for something specific and testable – accounting allegations, a short thesis, a temporary earnings miss – the value in doing the work is high, but having our thesis supported and validated by trustworth and aligned industry leaders often limits our downside and ensures our efforts and energy is well directed. Structural business quality is the anchor. If the anchor is sound, the noise is an opportunity.
Silicon Motion – When conviction is rewarded, and then some
Silicon Motion was a name we highlighted in our SaaSpocalypse Now? strategy note in February 2026 (link).
Our core argument was that the AI dispersion between semiconductor stocks and software stocks was going to keep widening. That is, the market would continue to reward infrastructure stocks because of their earnings visibility. And it would continue to punish application-layer software companies because of the threat AI agents posed to their seat-based (per-user subscription fees) revenue.
We were expressing that view in the storage space by holding Silicon Motion, which we bought in June 2025.
Silicon Motion is a fabless designer of NAND flash controllers – the small processors that manage how data is stored, retrieved, and protected inside smartphones as well as the solid-state drives (SSDs) that are crucial for AI.
Our thesis was straightforward: as AI capex ballooned, memory and storage, of which Silicon Motion provides a vital component, were emerging as the next bottleneck. Nvidia’s Jensen Huang had said just as much at the Consumer Electronics Show (CES) on January 5 this year. Samsung and Micron were pushing through 40-50% price rises because demand was outstripping supply. And TSMC’s capex guidance, a bid to meet this relentless demand, was well above expectations.
We didn’t need to predict every twist in the AI cycle. We just needed exposure to the ‘picks-and-shovels’ (infrastructure layer) end of it, at reasonable multiples, with a business that stood to benefit from the memory upcycle.
Silicon Motion has since delivered.
The company’s Q1 2026 revenue was up 105% year-on-year; EPS beat consensus by nearly 25%; and the stock has been one of the best performers in the storage complex.
It is currently trading around $320, up from about $70 a year ago.
Why we still own it: The company’s underlying earnings growth is not a one-quarter phenomenon. Silicon Motion is riding a multi-year product-led growth cycle. Its expensive, high-performance chips – PCIe Gen5 controllers, enterprise SSD solutions, and new AI-optimised products – are all coming into a market where demand is outstripping supply on a structural basis.
We continue to see outsized earnings growth ahead for the company. While some of that is now in the price, we see further upside from current levels.
What we take from this: When you get a thematic call right we don’t want to let our sector exposure continue to grow aggressively. When the market is crowding into a thematic, discipline requires taking money off the table – but the key is to identify which stock has more to run compared with the one where more of the upside has been realised. The hardest trades to make are often the ones after you’ve been right.
Artivion – When the market derates faster than the earnings
We wrote up specialised aortic device business Artivion in April 2026 (link here).
Our thesis was that this was a high-quality medtech compounder. It is run by a management team we have known and trusted for years. And it has a product pipeline that could sustain double-digit growth to the end of the decade.
We bought it pre-COVID, sold at the 2025 highs, and re-entered in February 2026on a pullback that was overdone.
Since then, the stock has continued lower. It is now trading around $US22. It’s down from its November 2025 high of around $48 and down from ~$35 at the time of our write-up. A disappointing result in May contributed to the share price decline.
Here is the key point, though: Even though the stock has fallen close to 40%, the company only reduced its FY26 earnings expectations by mid-single digits. That multiple de-rating is difficult to reconcile with the fact that the issues driving the guidance softening were non-structural:
- Elevated capex in FY26 to fund the closing of the Endospan acquisition earn-out.
- Some conservatism on new product timing.
- A broader derating of medtech companies in an environment that has been driven by momentum rather than fundamentals.
None of this changes the shape of Artivion’s earnings power in 2027 and beyond, which should be accelerated by the launch of new products, AMDS and NEXUS.
We tested our thesis directly.
After the pullback, we met with two separate US cardiologists and a European competitor to check whether anything had changed at the customer, clinical, or competitive level. It hadn’t. Product adoption remained strong. Cardiologist sentiment toward Artivion continued to be positive. There was no material change in the competitive landscape.
Why we still own it: Our valuation of the business has not changed. Our confidence in Pat Mackin and the management team remains. The earnings power we are confident in for 2027 and beyond is still in front of us, and the channel checks confirm the recent issues are not structural.
What we take from this: When a stock derates by 30-40% on a mid-single-digit earnings revision, the market is expressing a view about the business’s long-term earnings power, not just the current year’s number.
The discipline in those moments is to check whether the market is right – to test the thesis with the customers, the competitors, and the clinical data – rather than either doubling down reflexively or capitulating. When the channel checks come back clean, as they did with Artivion, that gives us the confidence to stay invested.
Looking forward: Three vital characteristics of our holdings
We enter the second half of 2026 with the portfolio positioned around businesses we believe combine three characteristics:
- Structural tailwinds that don’t depend on any single macro outcome.
- Valuations that leave room for re-rating as well as earnings growth.
- And management teams we have known through multiple cycles.
Marex remains one of our largest positions. Silicon Motion has been trimmed, but we retain meaningful exposure. And we continue to hold Artivion hold with conviction – its earnings power has not changed, and the recent share price weakness has made it more attractive on our numbers, not less.