Jason Buck did an incredible episode on the Infinite Loops podcast. On the back of it, Roger Nusbaum took the opportunity to recreate the Cockroach Portfolio this way:
Look at the backtest. No, really, look at it.
This isn’t a cockroach strategy scraping by on survival. This is something that would make a hedge fund manager quietly close his laptop and stare out the window for a few minutes. The risk-adjusted returns are that good.
So naturally, I tried to make it even better. And I failed. Repeatedly.
Every change I made, every “improvement” I was convinced would add edge, made things worse. Not marginally worse. Just… worse. The kind of worse that makes you feel a little foolish for touching it in the first place.
So let’s go through it, line by line.
Momentum
I was sure that swapping out plain equity beta, something like VT, for a momentum factor would juice the returns. It’s a reasonable prior. Momentum has one of the most robust records in the literature.
The problem? There’s no clean, turnkey portable alpha solution to actually implement it. No easy lever to pull.
Then I read Roger’s post and convinced myself this was finally the moment to play with box spreads. Get cheap leverage, express whatever factor I want in the equity sleeve, and if I ever forget to roll the options — which is a real possibility — the margin loan from IBKR is sitting right there as a backstop.
Then I found this post from Man Group that stopped me cold.
Testing portfolio ingredients properly requires a lot of data. And not just any single metric: drawdown frequency matters, but so does drawdown depth. If you’re trying to maximize CAGR, both dimensions of pain are relevant. Ideally, you’d stress-test every possible combination across a long enough sample to cover multiple economic regimes.
Roger’s analysis covers less than five years. That’s not a criticism, it’s just a constraint. Five years is one story. Markets tell many stories, and we haven’t read most of them yet. Many new stories can pop up in the future, and we should be ready for them as well.
CBOE
Roger uses this single stock as…a “better TAIL”? The logic is simple: if markets spiral, CBOE thrives because more volatility means more trading, more trading means more fees. Simple business model arithmetic.
I respect the logic. I just don’t trust it. Or better, I do not trust myself with it.
CBOE is a company. Companies change. A new CEO, a new board, a competitor with a better mousetrap, any of these can quietly dissolve the very property you thought you were buying. The “chaos hedge” you owned on Monday might be something entirely different by Thursday’s earnings call.
To be fair, I’m slightly more sympathetic when people apply this same reasoning to sectors rather than single names. You’ll find portfolios with dedicated sleeves for commodity miners, healthcare, consumer staples. The intuition is the same, embed a structural economic function, not just a stock. But the problem doesn’t disappear. Sectors drift. Classification systems change. What MSCI calls “healthcare” today might include businesses that bear little resemblance to the defensive cash flows you thought you were getting.
The strongest pushback against my position is obvious: if CBOE changes, you just swap it out. Which is true. And fine…if you have the time, the tools, and the discipline to notice the moment the thesis breaks.
Roger is a financial advisor. He thinks he can do that. I believe him.
But the way I think about portfolio construction, I want something closer to lazy. Not fully lazy, TBH plenty of alternative strategies demand regular attention, at least periodic check-ins. But there’s a meaningful difference between checking in on a factor or an asset class and developing genuine fluency in when a single business has quietly stopped doing the job you hired it to do.
That gap is wider than most people think.
BKLN
Anyone who reads Roger knows his feelings about fixed income. Especially the long-dated kind. Fair enough, duration has been a brutal place to sit for stretches of the last few years.
But bank loans as the substitute? I’d push back hard here.
The problem with BKLN is that it helps you precisely when you don’t need help, and abandons you precisely when you do. In a normal recession, the kind where central banks cut rates aggressively, bank loans fail you twice. First, the floating rate structure means you get none of the price appreciation that traditional bonds deliver when rates fall. The whole point of holding fixed income in a downturn is that rate cuts lift prices. BKLN misses that entirely.
Second, the credit component, the part that generates the attractive yields in calm times, behaves like equity when things get ugly. It goes down. Right alongside your stocks. The diversification you thought you had evaporates at exactly the moment you needed it most.
TFLO is a softer version of the same problem. No credit risk here, which is something. But the floating structure still strips out the rate-cut boost. You’ve neutralized one of the primary reasons fixed income earns its place in a portfolio.
Now, to Roger’s credit, this is a conscious call on his side. He recognized the hostile rate environment and shifted to floaters. And in the specific window he analyzed, when rates moved essentially in one direction, that call looked brilliant. But removing duration risk also removes a meaningful chunk of long-term expected return (you get why, right?). This trade only pays off if you successfully time your entry and exit from fixed duration. Get it right and you look like a genius. Get it wrong, or just get ‘unlucky’ with timing, and you’ve permanently reduced your returns for nothing.
I won’t tell you Roger can’t do it. But I’d strongly suggest you don’t try it at home.
AQMIX
This one made me laugh a little. In the exact window Roger chose for his analysis, AQMIX delivered roughly the same returns as DBMF with lower volatility. On paper, a compelling case.
Expand the window, and the picture changes completely.
AQR went through a multi-year rough patch that was painful enough that the team eventually went back and rebuilt the model. That’s not a small thing. When a firm of that caliber concludes that something in the engine needs replacing, you’re not just dealing with normal beta headwinds, you’re dealing with model risk on top of market risk.
For something sitting in the managed futures sleeve, I’d want a more “beta-like” proxy. Something that tracks the underlying risk factor more faithfully, without the idiosyncratic baggage of a single manager’s implementation choices. DBMF exists for exactly this reason. Even better, the trend following sleeve should have different models and managers (TBF this is something Roger says all the time).
EBSIX
There’s no clean ETF equivalent with the same track record. But two ETFs launched last year, HFGM and ASGM, and they’re outperforming it by a wide margin. The most likely explanation is a higher target volatility, which in a diversified portfolio context is usually a feature, not a bug. More vol from an uncorrelated source can actually improve your overall risk-adjusted returns, not hurt them. The math of diversification is counterintuitive that way.
So my working assumption is that both ETFs will serve better inside a portfolio than EBSIX does. Not because EBSIX is bad but because the structural characteristics of the ETF wrappers appear better suited for the role we’re asking this sleeve to play.
The correlation profile is what makes this genuinely exciting. EBSIX sits at 0 correlation with VT and -0.3 with TLT. If the new ETFs carry similar numbers, and early evidence suggests they will, the future will look bright 🙂
The rest of the portfolio is largely beyond dispute. BTC is the obvious wildcard: a deeply personal choice, and one that didn’t exactly flatter the backtest (TBF again, Jason has it in his portfolio). Take it out and the picture arguably gets cleaner.
But even with it in the mix: a 12% CAGR with a 1.4 Sharpe.
Sit with that for a second.
That’s not a number you should expect to replicate. That’s not even a number most professional allocators would publish without a disclaimer. Set your expectations honestly…and even with that honesty, what’s left is something that should make you look twice:
It is still an amazing portfolio…and yet again overstated because there is no fund that performs as SHRIX in the NatCat space.
What I am reading now:
Follow me on Bluesky @nprotasoni.bsky.social
0 Comments