Is the leveraged All Weather Portfolio worth it?

[Bostrom]

I found this article testing a 3x leveraged AWP: https://www.optimizedportfolio.com/how- ... he-market/

The CAGR has been 18% with a maximum drawdown of 46%. Much better return than S&P and with a lower volatility(10,4 CAGR, 51% max DD). Is this realistic? Is there anything you should consider before you create this portfolio with real money?

[jacob]

Is there anything you should consider before you create this portfolio with real money?

"Past performance is no guarantee of future results" would be the primary concern. I'm not sure how the algorithm came about, but in order to avoid model dredging the designer will generally hold back about half (there's a fancy way of computing how much to optimally hold back, it's related to the "stopping problem") their historical data and then test actual performance on the other half(*). This avoids overfitting and more importantly it avoids creating "false positive"-type strategies, i.e., thinking there's 'gold' when there's not.

(*) Another way would be to test the model on the data of another country. Arguments or presumptions have to be made that their economies and financial markets are similar.

You've probably heard of indexers talk about sequence of return risks. The reason they're doing [out of necessity] this is because they didn't create their model with the above in mind. To be fair, it's easy to do the above IFF your turnover is high. E.g. if you're in and out of a position in 1 year, you can use 1980-1990 for your science data and 1990-2000 for your test or even better 1980-1985+1990-1995 for your science and the remainder for your test. Whereas if your turnover is 50 years, you probably can't because the underlying trends could very well have changed in the mean time. (This is a different kind of problem to solve.)

TL;DR - If you actually start running this, the biggest drawdown in the future data stream might turn out to be 146% instead of 46%.

[candide]

(my) TL;DR the correlations between long-term bonds and stocks might not be the same going forward as they were in the back-tests.

I was doing an implementation of the non-leveraged AWP for a while, but at some point I couldn't wrap my head around having that heavy of a position in long-term bonds in a zero-interest rate world [1].

You could argue that the Fed has now bought itself some slack, and long term bonds could turn around soon, and that's fine as far as that goes, but for my money (literally) we are in a rising interest rate environment until proven otherwise, which has a high likelihood of stocks and long bonds going down at the same time. And that seems like a way AWP could test maximum draw-downs, or exceed them.

Just food for thought when it comes to leveraging this up.


[1] While the nominal U.S. rates weren't actually zero, they were getting below the inflation rate, so in real returns we were there. And unless we can bring inflation down, we're still there.

[xmj]

So a twitter acquaintance runs a leveraged Permanent Portfolio and blogs about it at https://demonetizedblog.com/, do please have a look at how that's working out for him in H1/2022 and how he reacted to the drawdowns in March'20.

If you do decide to run it, make sure your sizing is such that your regular cash inflows can cover any margin/capital calls.

[WFJ]

Please don't use leverage, it always ends badly. Read you margin agreement and understand that all securities can be moved from margin to non-margin account by risk manager in an instant. All ex-post research on margin discard this as a possibility but happens in every downturn, forcing individuals to sell at the worst possible moment.

During severe market moves, risk managers will move the most widely margined securities into non-margin in order to quickly reduce risk. This means VTI, VOO, SPY, QQQ, TLT will likely be moved from margin to non-margin during these time periods and wreck individuals accounts. I was on the "emergency margin call team" in 2000-2002 and stocks like GE, INTC, IBM were moved from margin to non-margined because they were widely held and easier to quickly reduce risk to the firm.

Depending on your stats background, watch this. If you don't understand the risks and this short presentation, don't use margin.
https://www.youtube.com/watch?v=91IOwS0gf3g