Systematic Alpha: Intelligently Cloning Top Investors with Olympus

TLDR: Here is a simple, appealing idea for making money in the stock market (demonstrated to outperform the market by almost 4% a year in this research paper by the London Quant Group).

It involves (1) the selective identification and (2) mirroring of the conviction picks of long-term fundamental investors. Naturally, the devil is in the details of (1), which we hope to break down in this article.

The Concept of Copytrading

Some people are very good at picking stocks. They are famous for it. They have teams of analysts and decades of experience and they manage billions of dollars. You and I are, let's be honest, probably not one of those people. But what if you could just do what they do? What if you could peek at their portfolios and buy the same things?

There is, it turns out, a form for that. It is called Form 13F (in the US, as other jurisdictions have different names for these disclosure forms).

Every quarter, big-shot money managers who control more than $100 million have to file one with the Securities and Exchange Commission, listing all the U.S. stocks they own. To be clear, you can actually go on the SEC's EDGAR website and see what stocks Warren Buffett's Berkshire Hathaway or Bill Ackman's Pershing Square held as of the end of the last quarter.

The whole thing feels like a cheat code. The answers to the test are right there. Just buy what they buy, sell what they sell, and wait for the Brinks truck to show up. Right?

Well…

The Flaws of Public Reporting Disclosures

Of course, it's not that simple. If it were, we'd all be rich and I'd be writing this from a much nicer chair. The 13F, and any public reporting disclosure in fact is a flawed document, a blurry snapshot from a moving car, and trying to drive by looking at it is a good way to end up in a ditch.

For one thing, the data is old. Managers have 45 days after the quarter ends to file their 13Fs. They tend to use all 45 of those days, because why would you tell the world your secret recipe before you have to? This means that when the filing for the first quarter (ending March 31) comes out around May 15, you are looking at a portfolio that is, at best, six weeks old. The manager might have already sold the stock you're getting excited about.

For another, the picture is incomplete. The 13F only requires managers to list their long positions in U.S. stocks. It doesn't show short positions. It doesn't show international stocks. It doesn't show bonds or commodities or private equity or how much of the portfolio is just sitting in cash. High-profile short-seller Jim Chanos has pointed out this limitation. He explained a scenario where he was short Hewlett Packard (HPQ) as his core position. To hedge his exposure to the enterprise segment of HPQ's business, he took long positions in Microsoft (MSFT) and Oracle (ORCL). His 13F filings would only disclose his long positions in MSFT and ORCL, giving the impression that he was simply bullish on those stocks. However, without knowing his larger short position in HPQ, an investor would get an incomplete and potentially misleading picture of his overall strategy and intent, missing the fact that the long positions were primarily a hedge.

Oh, and also, the data might just be wrong. The SEC's own Inspector General took a look at the 13F process and found that the commission doesn't really have a system for checking if the filings are accurate. Subsequent academic studies have found "widespread and economically significant errors" in the data. Sometimes the corrections are less accurate than the original filings. It's a whole thing.

A Modest Proposal: A Guide to Systematic Alpha

So, blind copying is a terrible idea. But what if there was a system? What if you could turn this messy, delayed, incomplete data into something that actually, you know, works?

Many academics in the investing field have taken a look at this and concluded that to systematically beat the market from 13F filings, you have to answer two questions correctly: "the who" (which managers to follow) and "the how" (which of their ideas to follow). It turns out that if you get the who and the how right, a strategy of cloning hedge fund picks can outperform the S&P 500 by a decent margin. Here is a guide to doing that.

Step 1: Choose Your Shepherd (The "Who")

The first rule of intelligent cloning is that you can't clone just anyone. You want to follow managers whose strategies aren't ruined by the 45-day time lag. This means avoiding the high-frequency traders and quant funds (Renaissance Technology, we're looking at you). You want the opposite: boring, long-term investors who think in years, not nanoseconds.

The research paper on this topic has a name for these people: "Fundamental Equity Hedge Funds" (FEHF). These are the managers who do old-fashioned stock picking. They buy businesses, not tickers. Crucially, they hold on to their stocks for much longer (a median of 12 months), which means a position you see in a filing from last quarter is much more likely to still be in their portfolio today. This simple filter—focusing on the patient, fundamental investors—is your first and best defense against the data's flaws. The research is clear: the "who" is just as important as the "how."

Step 2: Find Their Best Ideas (The "How")

Once you have your list of suitable investors, how do you find their best ideas? Just picking their biggest holding is a start, but we can be more systematic. The paper suggests a two-part filter based on Conviction and Consensus. Our free filter for this can be found here.

Conviction: This is about how much a manager really believes in a stock. Since hedge funds aren't trying to beat an index, the best measure of conviction is just the raw size of the position in their portfolio. A stock that takes up a huge chunk of a fund's assets is a high-conviction bet.
Consensus: This is about how many other smart managers agree. When multiple, independent, high-quality funds all have high conviction in the same stock, that's a powerful signal.

The magic happens when you combine them. The paper found that a portfolio of stocks held with high conviction by a consensus of FEHF managers generated statistically significant alpha. Here's the recipe:

Filter for Conviction: Go through your chosen managers' 13Fs and keep only the stocks that make up at least 7.5% of their portfolio. This gets rid of the small, speculative bets and leaves you with the ideas they are truly willing to stake their reputation on. For example, a look here shows that Warren Buffet holds 25% of Apple, 15% of AMEX and 11% of Coca-Cola.
Rank by Consensus: Take that filtered list and rank the stocks by how many different FEHF managers hold them as a high-conviction position.

For example within the Olympus universe, the top consensus pick is MSFT (10 FEHF managers holding).

Build Your Portfolio: Take the top 50 stocks from that consensus ranking. That's your shopping list.

For example within the Olympus universe, the top 3 stocks that display both conviction and consensus are Berkshire, Alphabet and Moody's.

A strategy like this, according to the paper, outperformed the S&P 500 by an average of 3.80% per year and delivered a Sharpe ratio of 0.75 over a 15-year period. It's not a guarantee, obviously, but it's a system. It's a process for turning noisy data into a potentially profitable signal.

Step 3: Do the Homework (The Actual Work Part)

This is the most important step. The systematic screen gives you a list of 50 interesting ideas. It does not give you 50 buy orders. The guru has done their work, and now you have to do yours. As Peter Lynch said, "Know what you own, and know why you own it."

You can't borrow someone else's conviction; you have to build your own.

Case Study 1: Coca-Cola

In 1988, after the market crash, Buffett's 13F filings would have shown him buying over a billion dollars' worth of Coca-Cola. Blindly copying would have worked out great. But the why is what matters.

What the 13F Told You: Buffett is buying KO.
What You Needed to Know: Coca-Cola has an unassailable brand and durable pricing power, and the 1987 crash created a rare opportunity to buy a wonderful business at a fair price.

Case Study 2: BYD

In 2008, Berkshire's filings on the Hong Kong exchanges showed a $232 million investment in a Chinese battery maker called BYD.

What the 13F Told You: Berkshire owns a stake in some Chinese company you've never heard of.
What You Needed to Know: BYD had a technological edge in batteries, a visionary founder, and a strategy to build affordable EVs for the mass market, positioning it to overtake competitors like Tesla.

Case Study 3: Apple

For years, Buffett avoided tech. Then in 2016, Berkshire's 13Fs started showing a massive, growing position in Apple.

What the 13F Told You: Buffett is buying AAPL. A lot of it.
What You Needed to Know: Buffett realized Apple wasn't a tech company; it was a consumer products company with the world's stickiest ecosystem. He saw that people would rather "give up their second car" than their iPhone, which gave Apple incredible pricing power and a recurring revenue stream that the market was undervaluing.

The Bottom Line

In summary, blindly copying 13F filings is a bad idea. It's like trying to assemble IKEA furniture using instructions for a different, slightly older piece of furniture that are also possibly wrong.

But using a system? That seems more interesting. A systematic approach based on finding a consensus of high-conviction ideas from the right kind of long-term managers is a plausible way to extract real signal from all that noise.

It turns the 13F from a gossip column into a quantitative screen. It's still not a magic bullet. You still have to do the homework. But it's a much better place to start.