Welcome back to the Wolf of Harcourt Street Newsletter.
Every month, I'll provide you with an update on my portfolio, including all of the transactions, the current allocation, and my buy list. In addition, I'll share a recap of the articles you may have missed from the previous month.
The Data Vendor That Talked Itself Out of a Win
ReviewSignal pre-registered a forecast on Shake Shack: at least four of the six customer-experience aspects it tracks would still sit below their 210-day baselines when re-measured on 28 July, two days before the scheduled Q2 print.
On grading day, ReviewSignal's own collection rotation had left the Shake Shack sample 18 days stale — and on that stale sample the forecast would have scored a HIT. Instead they re-scraped 45 locations and roughly 2,000 fresh reviews hours before grading. The fuller data showed a MISS, and that's what they published.
No cherry-picking. No rewriting history. The forecast, methodology and result are all available in the public ledger for anyone to verify.
That’s what makes this interesting for investors. Customer experience can change before it shows up in the financials, and ReviewSignal is trying to systematically measure those signals across 780,000+ reviews, 19,000+ locations and 79 consumer chains, updated daily.
You can explore the data yourself with a free sample pack, including a review-level sample, chain-day aggregates, data dictionary, methodology and limitations. No sign-up, no email gate, no call.
Transactions
Mercado Libre (MELI)
I added to my MELI position despite it already being the largest holding in the portfolio. The stock fell 8% after releasing its Q2 earnings, creating another opportunity to add.
Last month, I shared my updated financial model and valuation, which showed MELI was materially undervalued. Its Q2 results largely validated those assumptions, despite the market’s initial reaction.
MELI reported revenue of $10.2 billion, up 50% YoY, marking its first ever $10 billion quarter. Gross Merchandise Volume growth accelerated for the sixth consecutive quarter to 44%.
What is even more remarkable is that MELI is now growing at rates comparable to the Covid lockdown era. Back then, physical retail was effectively shut down and GMV was just a quarter of today’s scale.
Total Payment Volume also continued its upward trajectory, with growth accelerating for the fourth consecutive quarter to 56%.
So, why did the stock sell off?
EBIT margin came in at 7%, down from 12% a year ago. This was deliberate and has been well signposted by management, with MELI investing heavily behind the enormous opportunity it sees ahead.
Some of these investments include PIX discounts and take-rate cuts in Brazil, cheaper shipping, increased credit card issuance and further investment in its logistics network.
The engagement data management shared was also encouraging. Ecosystem users, those using both the marketplace and fintech products, are growing 37% and generate contribution profit that is a multiple of the combined contribution from single-product users.
In my model, I had forecast MELI to deliver revenue of $39.3 billion, up 36% YoY, and EBIT of $3.3 billion, an 8% margin.
The market now expects revenue of $41.7 billion, up 44% YoY, and EBIT of $2.9 billion, a 7% margin.
If MELI can continue to grow faster than I have forecast, the runway for growth is even longer than I originally anticipated.
Take-Two Interactive (TTWO)
Last month I thought I was done buying TTWO for a while. It turns out I was wrong, as the market presented another opportunity to add.
On 27 August, GTA VI: An Extended Look premiered on Netflix in a first-of-its-kind collaboration. Six hours later, the video was released on Rockstar’s YouTube channel, which you can check out below.
The viral nature of the game is becoming even more apparent following the Netflix feature. It flew to No. 1 on the Netflix movie list and according to, Sensor Tower, its viewership compared to Netflix’s previous Thursday audiences on mobile was up 50% and web visitors increased 125%.
This game is going to dominate internet culture like nothing we have seen before.
I still believe the market is massively underestimating the earnings potential GTA VI could create. TTWO has multiple potential revenue streams, including game sales, GTA Online, in-game spending, subscriptions and media, the latter of which Netflix showcased this week.
Allocation
I’ve been using Portseido to track my portfolio for years, and I highly recommend it. It consolidates all my transactions in one place, while its data visualisation and analytics capabilities guide my future decision-making and, ultimately, enhance my returns.
Sign up using my affiliate link here.
Buy Watch List
Rather than reference two existing holdings that are on my buy list this month, I am going to highlight two companies I have been researching recently but have not yet added to the portfolio.
Reddit (RDDT)
This community-driven discussion platform fits my preference for platforms, but the bull case has an interesting layer beyond the advertising business: proprietary human data.
Reddit’s forums represent roughly two decades of authentic, community-voted conversations. That data has proven highly valuable for training LLMs and is genuinely difficult to replicate synthetically.
Google has signed a licensing deal reportedly worth around $60 million a year to train Gemini, while OpenAI has separately licensed Reddit’s data. Reddit is now pushing for dynamic pricing based on how much its content actually improves model performance, rather than flat per-post fees.
It is also suing Anthropic over alleged unlicensed scraping, which reinforces the idea that this data has real and potentially defensible economic value.
The problem is that this remains mostly a narrative rather than a numbers story today.
Data licensing revenue grew 24% to $43 million last quarter, but that still represents only around 5% of total revenue. Google and OpenAI are the two dominant buyers, and Reddit has flagged uncertainty around renewing those arrangements.
Add in softer US daily active users and the decision to stop reporting logged-in and logged-out user metrics from Q3 onwards, and visibility into the underlying engagement trend becomes less clear just as the AI data thesis needs to start showing up in the numbers.
That combination of a compelling structural moat and an unresolved concentration risk is why I am still digging deeper.
AppLovin (APP)
This is the higher-margin, higher-beta idea.
The bull case is straightforward. The AXON model, an AI engine that continually improves ad targeting and bidding as it gets more scale and data, worked spectacularly well in mobile gaming. Management has now opened the same engine up as a self-serve platform aimed at e-commerce, a vastly larger addressable market.
However, the bear case is really a mirror image of the bull case.
The model only works if AXON keeps improving. When it doesn’t, growth can decelerate sharply. That is exactly what happened last quarter, when sequential growth slowed and management acknowledged that these AI gains cannot be scheduled on a predictable cadence.
What is notable is that this hasn’t actually been an earnings problem.
Consensus 2027 EPS estimates rose roughly 8.5% over the course of 2026, yet the stock fell about 48%. The difference is valuation. The market has cut the multiple it is willing to pay from around 35x forward earnings in December to roughly 17x by August.
Investors no longer trust that this growth will be smooth and predictable. As a result, the market is pricing AppLovin more like a cyclical ad-tech business than a compounding platform, even though underlying earnings power has not deteriorated.
Many investors are drawing comparisons with The Trade Desk. Whether that comparison proves accurate remains to be seen.
Next quarter is huge for this company. I expect the stock price to move. It is just a matter of which direction.
In Case You Missed It
Some of the articles you might have missed during the past month:
Final Words
After trimming ASML a couple of months ago, I continue to hold a modest cash position of just over 3%.
At the same time, I have continued to deploy capital into companies where the market has presented opportunities through multiple compression.
With all this going on, the portfolio has once again reached a new all-time high this month.
Times like this are when complacency can set in and you start cutting corners with your due diligence.
Market conditions might change, but your process is the one constant through all of this.
Stick to it.
If you'd like to support the work of an independent analyst, you can buy me a coffee. The proceeds will contribute to covering the annual running costs of the newsletter.
Join the community of informed investors – subscribe now to receive the latest content straight to your inbox each week and never miss out on valuable investment insights.
The Chat is a space designed to facilitate, real-time discussions, share knowledge and debate ideas with fellow investors. Join the conversation.
If you found today's edition helpful, please consider sharing it with your friends and colleagues on social media or via email. Your support helps to continue to provide this newsletter for FREE!
Happy investing
Wolf of Harcourt Street
Contact me
Twitter: @wolfofharcourt
Email: wolfofharcourtstreet@gmail.com
The most interesting line in the MELI section is the comparison between your model and the street's: the market now expects more revenue than you do, $41.7 billion against $39.3, and less EBIT, $2.9 billion against $3.3. That means the disagreement is not about the size of the opportunity at all, it is purely a theory of the margin trough, and a deliberate margin trough is the hardest claim to falsify in the short run because every weak print can be filed under investment. Which makes the ecosystem-user data you cite the number to watch: multi-product contribution profit is the one line that says the spending is buying economics rather than only volume.
What's your take on NIO currently? I have it on my watchlist too but they're not yet GAAP profitable, about to turn soon hopefully. Your allocation size makes me thing you're probably gonna size up slowly