One of the biggest challenges with investing is to take the emotion out of it. Don’t fall in love with a company. Revisit your rationale and see if it’s still intact, including the valuation. There is a price when it becomes clear to buy, and a price when it becomes clear to sell.
In between, the gap can be wide, and the temptation to constantly do something, high. Therefore, over the almost 15-year tenure of the fund there have been only 14 companies held, sold and subsequently bought back again, and only one where that has happened more than once. Generally, we made a profit during the first holding period but stepped away when valuations became too stretched, only to step in again after a new pull-back in shares.
In February, a second company made it into the fund for the third time: RELX, previously known as Reed Elsevier. The first occasion was early on. We bought the shares in March 2012 at a forward price to earnings ratio (FWD P/E) of 10x² and sold them in April 2015 at 18.5x FWD P/E². In 2020, the exhibitions business was affected by COVID, and concerns arose that Open Access to scientific journals would damage RELX’s business model. The valuation fell to 15.5x FWD P/E² and presented another opportunity to invest. Both issues were eventually addressed and the shares subsequently rebounded strongly as the market fell in love with higher growth and margins from new digital products. We eventually sold our holding in July 2024 when the shares traded at 25x FWD P/E².
After a peak in the early summer of 2025, RELX’s valuation had begun to fade as the potential for new competition from AI startups emerged. We decided to continue to wait and put RELX on our To Do List. In February this year, Anthropic announced legal plug-ins for its ClaudeCowork product. Panic took hold and RELX shares unusually sold off 15% in one day².
We were again interested. The first step was to assess Claude’s capabilities compared to RELX’s legal product LexisNexis. Our conclusion was that while Claude can automate routine contract review tasks, it is unable to come up with a detailed analysis that fully captures a lawyer’s style, past data and current legal issues³. In the last 15 years RELX built an enormous database to help lawyers with complex needs. Importantly this is not just based on public information but also proprietary content. LexisNexis can source from behind paywalls, captchas and from restricted regulators4.
With over 650 data scientist and engineers and over 3,000 technologists to double check the data, broken feeds can be fixed imminently, hard copies can be translated into the relevant format and lawyers can be guaranteed high quality data4. When we spoke to Claude users, hallucination was highlighted as a real problem. In an industry where quality and correct information and data are of upmost importance this presents a major hurdle.
While we believe it would be difficult for any new Large Language Model to take away market share from RELX, we closely considered this along with the associated margin headwinds as a worst-case scenario. Even then, we felt the shares had become oversold. With one of Warren Buffet’s mantras front of mind: “Be fearful when others are greedy, and greedy when others are fearful.”, we initiated a position for the third time in the fund’s life in RELX at around 16x FWD P/E².
This concept has also worked well in other sectors. Several years ago, we had to make another choice. The fund had invested in Heidelberg Materials in 2017, and despite higher profits by 2022, the shares had fallen to a level valued at only six times forward earnings². Had we made a mistake? The short-term picture was of course difficult: higher energy prices, low demand and an implied view that cement and aggregate prices were too high.
We re-checked our assumptions and spoke to industry contacts. They talked about the rest of the decade: new regulations, pent-up demand and less competition – all of which would strengthen market leaders and leave them with excess cash flow. The medium and long-term won out, and Heidelberg became a top ten position.
Similarly uncomfortable was the first investment in ArcelorMittal in 2024. Making steel is a difficult activity at the best of times, but persistent low-cost competition from China had decimated profits across the world. However, fast forward a few years and Arcelor is in rude health thanks to new custom duties and quotas in Europe5, improved costs and smart acquisitions and disposals.
A second small addition to the fund follows in the same footsteps. LATAM Airlines is the South American regional market leader that has emerged strongly out of its post-COVID restructuring. Thanks to the combination of structural cost reductions, impaired competition and resurgent demand for premium travel, LATAM has gained market share and is in a sweet spot to serve a middle class expected to grow from around 400 million to 490 million people by 20426.
Connectivity in South America needs a lot of improvement. Intra-regional traffic is ten times smaller than that of Europe7. The average number of trips per capita is 0.67, compared to levels many times that other parts of the world. The key question is: can LATAM grow profitably?
We think so, and thankfully it is not too heroic an assumption given that in 2025 LATAM already achieved a 16.3% underlying margin, among the highest in the global industry. It benefits from strong hub positions, a well-balanced order book and the support of an unusual combination of owners and joint venture partners: the families involved in privatising and founding both the original LAN (Chile) and TAM (Brazil) airlines respectively, Delta Airlines and Qatar Airways.
Once again, a short-term challenge; higher jet fuel prices in 2026, opens an opportunity for the medium and longer term - where a combination of the best management teams and the strongest balance sheets can add the most value. Over time, we expect obtaining and maintaining aircraft will remain difficult, and yet demand for higher end travel will continue to advance. LATAM is able to meet this need, and at the same time generate substantial free cash flow. Company guidance currently points to more than $1.6bn in 2026 and more than $1.8bn in 2027 at previous more normal fuel price levels8. The current market value implies something vastly worse.
We financed the new positions through the exit of Evonik. The structural headwinds of a chemical company being based in Germany have unfortunately remained in place. Despite a solid effort to cut costs, management felt it prudent to reduce the payout ratio and cut the dividend, a final sign that any potential recovery from oversupplied markets would be many years away.
1 Warren Buffet
2 Source: Bloomberg
3 Source: Bloomberg Intelligence
4 Source: Fireside chat with Mike Walsh, CEO of RELX’s Legal Division, hosted by J.P.Morgan’s Daniel Kerven; 24.09.25
5 New measures to protect EU steel market from global overcapacity | News | European Parliament
6 Latin America commercial aircraft services market value set to double by 2042 | Airbus
8 LATAM Airlines Investor Day December 2025