Written by Summary
ARK Invest has published their main investment themes for each year since 2017, including numerical projections for how these themes will grow and outperform.
I checked how accurate their 2017 projections were over the last five years and was surprised by what I found.
I also looked at how their projections for 2022 compare to their 2017 projections in terms of theme and growth rate.
Cindy Ord/Getty Images Entertainment Thesis
Although I’m invested in many of the same individual companies as ARK Innovation (NYSEARCA: ARKK ) and am therefore bullish on the fund by proxy, I share many investors’ concerns about whether an actively managed fund with seemingly whimsical day trades, little concern for valuation and profitability, and a high expense ratio could outperform an individual’s buy and hold strategy. After looking at their annually published predictions going back to 2017 and how they panned out, I was shocked by the consistency, accuracy, and valuation-sensitiveness of their projections and now believe the fund is better managed and better positioned for outperformance than most investors realize. Auditing 2017’s Big Ideas
ARK’s first big ideas report was published in 2017 and included seven ideas. This section covers their predictions from that report and how accurate they ended up being five years later.
Contrary to the popular belief that ARK’s predictions are emotionally generated pie-in-the-sky numbers, their approach is actually highly quantitative. It uses Monte Carlo Simulations (which they publish) and is based on mathematical ideas like Wright’s Law (similar to Moore’s Law). To keep this article at a reasonable length, I won’t discuss their methodology for coming up with the following predictions, instead focusing on whether they have been accurate. Deep Learning
Prediction Accuracy: A
Deep learning is a type of machine learning modeled after the brain. ARK flagged Tesla’s (NASDAQ: TSLA ) autopilot and a variety of Google (NASDAQ: GOOG ) (NASDAQ: GOOGL ) products like AlphaGo, Google Photos, and Gmail Smart Reply as products that use deep learning. They also mention usage at Facebook (NASDAQ: FB ) and Baidu (NASDAQ: BIDU ). As someone who was working on Google Photos in 2017, I definitely remember it being an exciting time for machine learning, and I believe it’s still very exciting today.
ARK’s forecast for deep learning was very long term, estimating $17 trillion in market cap by 2037. It’s certainly too early to say whether this forecast was accurate or not. But the four aforementioned companies they flagged as having good deep learning products returned an average of 527% since then and represent over $3.5 trillion in market cap, so I’d say they’re off to a very good start. The caveat being that much of that return may not have come from deep learning. Mobility As A Service
Prediction Accuracy: B ARK expects that before 2020 fully autonomous vehicles will become commercially available, enabling the rapid rise and growth of autonomous taxi networks. This was a very bold prediction but it did end up being technically accurate. Waymo and Baidu were among a small group of companies that had driverless taxis available in some cities in 2020, and Tesla rolled out a limited version of autopilot.
However, their prediction was too aggressive, since even in 2022 most people have never been in an autonomous taxi and it still seems like it could be a long way away. In particular, they said that hailing an autonomous taxi in 2020 would be cheaper than driving a personal car. But according to Baidu’s 2021 projection, this likely won’t happen until 2025.
Part of their forecast included EVs becoming much cheaper, and this part turned out to be fairly accurate. A 2022 Nissan Leaf (OTC: OTCPK:NSANY ) – the cheapest electric car – will cost $28,375, compared to ARK’s estimate of $22,400 (down from $41,800 in 2016).
They were also chillingly accurate with their prediction for EV sales. An estimated 5.6 million EVs were sold in 2021 (leading to nearly 10 million in 2022, with China alone expected to cross 6 million) compared to their estimate of 17 million in 2022. Against just 500,000 EVs sold in 2016, 17 million was an insanely bold call against a consensus of just 2 million. It ended up making their Tesla investment very profitable since they were much more accurate than the analyst consensus despite overshooting by around 50%.
ARK also projected that oil demand and auto sales would peak around 2022 as a result of trends in EVs and robotaxis. It is still too early to say for sure […]
source How ARK Invest’s ‘Big Idea’ Predictions Performed From 2017-2022
Protected by Patchstack