If you’ve watched Billions, you know that Bobby Axelrod employs a therapist to keep his hedge fund analysts cranking at maximum capacity.
I’m no therapist, but my job the last two years has entailed interviewing 100+ investors at the world’s best hedge funds. (Don’t worry investors, the confidentiality of the couch will be respected.)
I thought, along the way, I might learn the secret to become the next Bill Ackman. Or at least, I’d understand a big puzzle in investing: how incredibly smart people, armed with data from satellites and every database in the world, more often than not lose to index funds that buy whatever comes along.
But what I mostly learned about the “secret sauce,” is that it might not be sauce at all.
The Sauce
Bobby Axe is strangling his mouse. The robots are messing with him. He should be happy - his stock is up, while its competitors all plummeted. Quants employing a peer trading strategy had sold off rivals after some bad news, but his stock was doing fine. What did they know that he didn’t?
The “sauce” at any fund is the process. The website of any fund is full of words like “disciplined,” “research-driven,” and “bottoms-up,” meant to indicate that they don’t trade off random tips or hunches.
When I interned at a fund in business school, the process was strict, passed to my boss from his, back up the chain like a family recipe.
But no matter how good you are at following the instructions, everyone seems to throw in their secret ingredients.
A great example is sell-side data (Wall Street analysts). If you ask investors whether they consider the revisions by sell-side analysts important, the honest ones will say “usually, but not always.” “The sell-side is always behind,” every fund says, except when a downgrade tanks the stock, and except when an analyst talked to a buy-sider…How do you track all the exceptions?
Most will say experience. I often hear things like, “I like to see X and Y,” which are two things that have worked for them in the past. But add that to also knowing when to manually recalculate EBITDA, or when to use blended or simple NTM, or a dozen other bits of individualized flavor, and you have a process with a lot of personal judgement.
Expert-level investing, at times, is a bit like football coaching before the analytics revolution. Coaches knew a ton about football, but whether to go for 2 ended up being a mix of individual risk aversion and gut feeling. Now coaches have someone in their ear giving them a precisely modeled answer.
So we tried.
Going for two
Finding competitors is core to valuation, and so at Visible Alpha, we iterated on algorithms for determining peers for each company. We wanted to give users not just a set of peers, but the best peers. At some point though, users told us it wasn’t worth making it perfect - because everyone told us eventually the lines get blurred.
That took me by surprise. There had to be some hard and fast rules. But we kept running into the same questions: when looking at a tech company, should we include a rival in Canada? Some investors say they do, and some don’t. What about France? Everyone has a different answer. But crucially, each one thinks their approach is right.
We ran into other traps, too. If you want to value Amazon, you likely have to do a “sum of the parts” valuation where you put a dollar value on each of Amazon’s many businesses. But for Amazon’s cloud business, your best comps are Google and Microsoft….which also have many business lines. Circular reference, #REF?
That’s the hard part about an algorithm. We used methods with proven alpha, but investors often disagreed - for the best peers, it turns out that’s often part of the bull and bear debate itself. If the rules change constantly, then we can’t make it systematic. And if we can’t make it systematic, it’s hard to test. There just aren’t enough data points to measure against: price alone isn’t enough. So we’re back to coaches on the sidelines.
And everyone “knows” the signals that work - if I see multiple upgrades, or if the stock is a four out of five, or cheap compared to my custom index that I made.
The challenge is that if your process is full of little idiosyncrasies, how do you know that they all add up to a net positive? How do you know the sauce is actually better when you’ve added so many secret ingredients? The robots will calculate and backtest every individual piece of data to assign a consistent weighting - but the reality is Bobby Axe is creating a new recipe for every investment.
Does that mean I learned nothing?
No, not at all. Actually, studying investors led me down some very interesting paths. I do feel I’m a better investor than I was two years ago, even though I already had a degree focused on value investing.
But I also learned that a lot of the challenges have more to do with psychology as much as process.….but I’ll share more of that in part 2.
I do believe that funds with the right structure and tools can deliver uncorrelated alpha, and that active investing is a critical piece to functioning markets.
What are the right structures and tools in my opinion? Unfortunately…that’s not until part 2 either. But here’s a few preliminary points in the meantime.
No one agrees on anything…
There’s nothing magical about hedge funds. In our imaginations, these guys are trading off super hush-hush data that costs millions, but that data is often a supporting part of the thesis and rarely the center. When it is, usually in quant strategies, those returns often fade quickly and inexplicably. Most of it is old school business analysis, except no one knows what that means any more, since every industry is being “disrupted.”
…except the things they agree on.
Everyone “knows” the stock is driven by this one KPI, investors will always say (think vehicle deliveries for Tesla, or user growth for Robinhood).
The sell-side is less important, they agree, except that the sell-side usually helps you understand which KPI matters, how it’s calculated, and gives you the data. Now, there’s no gold standard for that data - some use Bloomberg, some use FactSet, some use VA - and setting up a bot to trade that KPI is hard because, given those metrics are unregulated, companies will change how they calculate them on a whim.
So we know what we agree on…but good luck trading it if you’re not focused on that number full time.
Sell side is useful
I’m a numbers nerd, so to me the fun part is finding quant studies to see how much of investor intuition is true.
It turns out the sell side does generate alpha, particularly in small caps like the Russell 3000. For that to happen is actually pretty good, because if investors are often trading against the sell-side, it’s almost a self-fulfilling prophecy that the sell-side’s accuracy will be very hard to measure.
Overall, learning the nitty gritty of investing was pretty fascinating. But some of the most fun stuff was the psychological games involved (next week’s post!). See you in the new year!