Compounding Research

Before going into the article, something personally important. I’m studying for a Graduate’s Degree in Finance and Economics, in Boston. I’m about to start my second semester and, when the third one begins, in January, I’ll be legally allowed to do an internship. I’m looking for an asset management firm that might be looking for someone that performs equity research, preferably in Boston or New York. If you happen to know of anything, I’d be really grateful if you let me know.

E-mail: giulianomana@0to1stockmarket.com

Article

Building an investment framework is not easy. At the end of the day, 80-90% of investors underperform the market. Even though among overperformers there is probably a 0.1% who are brilliant guys like the ones we’ve covered here, I suspect the remaining 9.9% are normal people that simply follow solid rules.

I believe most of these people have learned from the 0.1% and are now able to buy good companies at decent prices. However, the problem arises when selecting them. One is supposed to dedicate hundreds of hours to researching a business and ultimately make a somewhat arbitrary call. Is the business I’ve researched a good one? For guidance in this regard, I wrote this.

We become masters of what we practice. I finished writing Zoetis research and thought it would be a good time to meditate on the research process. The key takeaway is that progress is only possible if we recurrently perform an activity. Each new research attempt stacks up from all previous ones. They compound.

My first and second research

Accenture was the first company I covered (I took it out of the blog). The article was published in September of 2022 and it’s ridiculously bad. It’s a thousand-words write-up with everything off, except the core, I believe. Length says nothing by itself, but it does give a vague sense of how much time was dedicated to the subject. Besides from the thesis, the business overview had a wrong focus, the financial analysis was merely three paragraphs describing what had happened, there was no detail anywhere and the MOAT analysis had no body.

One million mistakes: Too much ignorance on all fronts and poor writing

The second company I covered was ASML, two weeks after Accenture. A bit better than Accenture’s, but still worth 0. A noble attempt to go over photolithography’s complexity, but far outside my circle of competence at the time, which made the whole write-up unclear and miss-aimed. The idea behind the thesis and industry are good (semiconductors..), but the execution was not. The semi industry has one of the most significant tailwinds out there and it barely showed in the research.

Mistakes: Trying to explain something I did not understand. Ignorance on all other fronts.

Third, fourth and fifth articles

Texas Instruments might represent the last very poor article, but was one step above the others. Resourcefulness starts to show. Decent dive into the company’s history, first apparent dive into SEC filings, investor presentations, other sources and transcripts. Some good highlights like the more detailed business overview, now covering several elements, management’s approach to capital allocation, technical peculiarities and reproduced insights.

Mistakes: Still outside my CoC, can’t break down things to a clear level, the article seems like a collage which gives the sense of little thought dedicated to it, poor financial analysis, still looking at what everyone looks at and lack of reading.

The fourth research article was about Mercado Libre in mid-october, a 2,500-word article. It’s a relatively good surface analysis. It correctly covers the known things about Mercado Libre, what it does, its management history and exposes the immense secular tailwinds behind the company. First big step forward in analyzing a business’ competitive positioning, returns on capital and first time an article seems to make sense altogether.

Mistakes: No surface penetration due to lack of reading 10-Ks and transcripts. The thesis is in the right direction, but incorrectly articulated with poor and almost unrelated arguments. Ugly financial analysis.

The subsequent company I covered was Tesla in mid-November, 3.1k words. The structure is very similar to Mercado Libre’s research, but it adds a further sense of a narrative thread. Improved 10-K distilling and pieces of data obtained from non-traditional sources. First true industry and thesis articulation, which starts to show some thoughts dedicated to the topic. First attempt to provide unique insights and finally a new format for the financial analysis.

Mistakes: Still no surface penetration. Improved information quality and digestion, but little to no dots connected. Lack of exploration of Tesla’s unique manufacturing advantage. No dive into Tesla’s mentioned moat nor on its true competitive positioning. Still at a point of not being worth the read.

Sixth article, Microsoft

11k words write up written in December. First piece I think is worth reading, which is also when I started to take a first-principles approach to researching, trying to go over the actual basics of the business before the business itself. A necessary evil in my opinion. It took a lot of time, but the product was finally presentable. It had information nobody else had put together, finally personal insights and dots connected, a proper dive into the company’s operations, some more clearly written ideas, a sense of a narrative, though still far from good, and finally a personal and decently-argued thesis.

Mistakes: Long way to go with the format. Not well written overall. Poor financial analysis. No tangential exploration besides from Microsoft’s business units. Lack of diving into Azure’s competitive positioning and still a bad management analysis, or at least not a new one. No reading past the last 10-K and couple of quarters’ transcripts.

Seventh article, Visa

Visa’s write up was published in February and had 6.5k words. This was the first time I got to tell a story. Visa’s article has a detectable narrative behind it. The structure follows the essence and not the other way around, as prior. Finally an article that’s 50-70% clearly written and with thought-through ideas. Moreover, Visa’s article represented the company’s 10-K in a faithful manner, had occasional tangential exploration and had the business overview correctly articulated.

Mistakes: No meditated sentences, lack of clarity in some technical matters, no details in important regards like actual differentiation with competition. No distilling of the digital payments industry, which is crucial to understand the role Visa plays. Huge step downwards in terms of relevant details comprehended and apparently less thoughts dedicated to its thesis.

Eighth analysis, Google

I wrote Alphabet’s articles during March. Very much in line with Microsoft’s research in the superficial and structural sense. However, having the latter sorted out allowed this article to be much more profound, thought through, more clearly written, have a more elaborated thesis and several tangents explored.

Mistakes: No clear narrative. Still not going past 1 or 2 10-Ks and a few years of transcripts. Still far from good management analysis. Lack of reading of singularly-focused papers and articles. No expansion on Google’s moats ‘due to length’. The lack of reading still limits the output’s depth.

Ninth research, Zoetis

Zoetis research ended up being almost 15k words and a substantial step forward was taken with what respects to quality. I went through infinite sources and had never filtered as much information for a single company before. I read, or at least skimmed through, all of its 10-Ks, trying to analyze management’s narrative across time, I retrieved information from 5-10 events plus all conference calls, which might be 30-40. Moreover, I dedicated more time to each paragraph and individual sentence, which I believe gets to show when reading carefully. The articles follow a story and there is an implicit, but very much visible thread connecting everything. I think the product got to a decent point, although it took 2-3 months.

Mistakes: Even though I’d include it in a fourth article, I feel I still lack management analysis skills. Since the research is too recent, I’m not yet sure about the mistakes, though for sure there are many.  I generally realize the mistakes I make as months go by. Time allows me to meditate on these things.

Conclusion

I think astronomical progress was made and, interestingly, in less than a year. If you get 1% better every day for 365 days, when the year ends, you’ll be 37 times better. The human mind is not designed to comprehend nor calculate exponential growth. Dedicating a couple of hours per day to any activity over a prolonged period works wonders.

Personal commentary

I’ll be traveling back to Boston tomorrow and begin classes on Thursday. Once I start, I’ll get back to pure finance readings and articles will naturally follow. After finishing with The Innovator’s Dilemma, I’ll try to go over 1/5-1/4 of Michael Mauboussin writings. I think his write-ups will bring fascinating ideas to the newsletter. And after him, perhaps I resume with shareholder letters. Hope you enjoyed another unusual article.