The Builder's Edge
My investing journey, Part 4: 2017 to 2026
Sep 28, 2026 · @Deniz Erkan
In 2017, I came back to the market with money from the company Raed and I had built, and, for the first time, with a style I could explain. This is what the ten years since then look like.
Reconstructed time-weighted returns; Nasdaq-100 and S&P 500 price indexes (Yahoo Finance, FRED)
A dollar I invested at the start of 2017 was worth $30.38 by September 2026. For the last decade, I have compounded at about 42% a year. The same dollar in the Nasdaq-100 became $6.27, about 21% a year. In the S&P 500, it became $3.44, about 13.5% a year.
Year | My return | Nasdaq-100 | S&P 500 |
|---|---|---|---|
2017 | +18.7% | +31.5% | +19.4% |
2018 | −23.5% | −1.0% | −6.2% |
2019 | +16.0% | +38.0% | +28.9% |
2020 | +858.3% | +47.6% | +16.3% |
2021 | +17.1% | +26.6% | +26.9% |
2022 | −46.7% | −33.0% | −19.4% |
2023 | +66.5% | +53.8% | +24.2% |
2024 | +66.0% | +24.9% | +23.3% |
2025 | +31.2% | +20.2% | +16.4% |
2026 (to Sep 24) | +33.1% | +20.7% | +12.5% |
Annualized | +42.0% | +20.8% | +13.5% |
Growth of $1 | $30.38 | $6.27 | $3.44 |
The first three years were not easy. In 2018 I lost 23%, and by the end of 2019 I was still behind both indexes. Then the edge I had spent a decade building started to show, and it came from one source: what I had learned by building technology products, finally pointed at the stock market.
How building picked my winners
I read all of Peter Lynch's books, and one idea never left me: look around your daily life at the products you use. The best investment ideas are often hiding in plain sight, in the things you already understand from using them. For me, those things were technology products, because building them was my daily life.
Over the years, Raed and I built many products: SaaS, business software and consumer services. Building products means making a steady stream of technology choices. Which network do we serve users from? Which cloud do we run on? Which chips do we buy? Every one of those decisions is a small bet on which technology company is doing something better than its rivals.
Cloudflare came into my life as a practical choice. We first used it to replace our content delivery network, because it was faster and cheaper. Only later, using it every day, did I realize what it really was: not a delivery network but an edge cloud, a place to run software close to every user in the world. The investor in me saw it after the builder in me did.
Nvidia and AMD came the same way. We ran workloads on public clouds, on our own custom cloud and eventually in our own data centers, and every time we had to decide which chips to use: Nvidia, AMD or Intel. We felt the difference in our own bills and our own speed long before it appeared in anyone's earnings report.
That is the builder's edge. It is not secret information. It is the judgment you develop by choosing technologies with your own money and your own product on the line, and it told me which companies to research and which to shortlist.
Five companies
Company-level profit attribution, August 2016 to August 2026
Over the last decade, five companies earned more than my entire net profit: Cloudflare, Nvidia, Roku, AMD and Tesla. Everything else, taken together, lost money.
From 2018, I owned my strongest convictions with more force than before, using options mostly as synthetic stock: measured leverage, never heavy margin.
The year my record changed shape, 2020, was carried mostly by Roku, Square and JD.com. They look different from Cloudflare and Nvidia at first glance, but they came from the same place: they were all platforms. We used payment systems in our own products, and I watched closely how Square's leadership executed. Roku was building an operating system for televisions that looked like the TV platform of the future, and I liked its strategy and its CEO. JD.com, the platformization of e-commerce in China, was a little different, but it was a platform too.
The chip and cloud companies then carried the years that followed: Nvidia, Tesla and AMD in 2021, Cloudflare in 2023, AMD again in 2025.
Build things long enough and you start to see which companies will win.
Next in the series
Engineered compounding,
Deniz Erkan
Disclosure: This letter describes my personal investment history. Returns are reconstructed estimates from monthly account balances, are not audited, and do not reflect fees or taxes any other investor would pay. Company figures cover August 2016 through August 2026. This history is separate from the Runtime fund and its algorithmic strategy. Past performance does not predict future results. This is not investment advice.