Rob Arnott

Rob Arnott has been in the money management business for almost 50 years. He launched Research Affiliates in 2002 while simultaneously running First Quadrant, eventually giving Research Affiliates his full attention when the business demanded it. The firm's business model is unusual: they don't run money directly. They create strategies and license them to firms like Schwab, Invesco, ProShares, and PIMCO. Through those partnerships, Research Affiliates indirectly runs $158 billion. They're best known for Fundamental Index, which weights stocks by the size of their business rather than their market cap, accounting for about $75 billion in ETFs alone. Schwab's FNDX versus iShares' IWD is a real-world test case, and Arnott says the value added has been "relentless."

On this episode of Behind the Ticker, Rob talks with Brad about NIXT, the Research Affiliates Deletions ETF. The idea is simple and counterintuitive: buy the stocks that just got kicked out of major indexes, because they've been systematically sold at their worst prices.

The Upside of Getting Dumped

When a stock gets deleted from the S&P 500 or Russell 1000, it gets hammered. Index funds have to sell. The stock is depressed, out of favor, and everybody who was forced to own it is now forced to sell it. Arnott's research shows that deletions outperform the market by an average of 7% per year for the first couple of years, tapering to about 3% after five years, for a total of roughly 2,800 basis points of cumulative outperformance over the five years after deletion.

He used a vivid analogy: "We've all had the experience of getting dumped. You can either wallow in self-pity for life, or you can pull your socks up, take some learnings, and move on. Companies that get dumped from the index face the same choice." Most of them pull their socks up. The white paper is titled "NIXT: The Upside of Getting Dumped."

Bayesian Methods, Not Data Mining

Arnott was emphatic about methodology. Research Affiliates built their own databases and analytic tools. They are "obsessive" about not data mining, which he says the quant community is addicted to. "If you build models that maximize historical back-test performance, all you're doing is maximizing historical back-test performance." Instead, they use a Bayesian method: start with a theory, use data to test it, but never use data to tweak and improve the back-test. "We don't go wherever the data leads us." The result is strategies that work in live markets, not just in backtests.

Brad connected this to what he's seeing with AI strategies: back-tests that J-curve beautifully but will never work in production because of curve fitting and optimization. Arnott agreed that strongness is the key, and his 20-year live track record on the Fundamental Index provides evidence that the philosophy works.

Index Construction: Filtering Value Traps

NIXT doesn't just buy every deletion. First, it constructs its own top-500 and top-1000 by market cap (avoiding licensing costs from S&P and Russell). Companies that fall more than 10% out of those ranges hit the deletions list. The team then filters on quality metrics: not just profitability (which Arnott considers too narrow) but debt-to-equity ratios, coverage ratios, and multiple measures. The worst 20% on quality get filtered out as potential value traps. The remaining deletions are bought and held for five years. If a stock gets re-added to a major index during that period, it exits the NIXT portfolio.

The data goes back to 1991 and shows consistent outperformance. Arnott emphasized the critical distinction: "We use data to test an idea, not to create the idea." The theory came first (forced selling creates mispricing), then the data confirmed it.

Value at Historic Discounts

Arnott placed NIXT in the broader context of value investing. Since 2007, value has underperformed growth by roughly 3,000 basis points. The relative cheapness of value versus growth, measured by price-to-book, hit a 9-to-1 ratio in 2020 (meaning the market was pricing growth companies to outgrow value companies ninefold). It's currently at 8.5 to 1. "Value is incredibly cheap," Arnott said, "which is something that I find very exciting." NIXT is very much a deep-value strategy, and Dave Nadig called it a "completion strategy" that fills in what indexes no longer hold by buying companies that have proven they can run a big business. The companies in the portfolio were recently in the S&P 500 or Russell 1000. They're just temporarily out of favor.

Key Takeaways

• Stocks deleted from major indexes outperform the market by an average of 7% annually for two years, tapering to 3% after five years, totaling about 2,800 basis points of cumulative outperformance.
• NIXT filters out the worst 20% on quality metrics (debt-to-equity, coverage ratios, profitability) to avoid value traps before buying deletions with a five-year holding period.
• Research Affiliates indirectly manages $158 billion through licensing partnerships. The Fundamental Index alone accounts for $75 billion in ETFs.
• Arnott uses Bayesian methods exclusively: theory first, data to test. Never data mining. His critique: the quant community is addicted to maximizing backtests, which produces strategies that are "rubbish" in live markets.
• Value is at historic discounts versus growth (8.5-to-1 price-to-book ratio). NIXT is a deep-value "completion strategy" buying companies recently big enough for major indexes but temporarily out of favor.

Listen to the full conversation on Spotify, Apple Podcasts, or YouTube.