I should have put this out a few days ago, but better late than never. The portfolio has moved pretty aggressively over the last few weeks away from bonds, and it’s worth discussing.
Correlations between bonds and stocks are quite high right now. As we know, treasury bonds aren’t paying out much with the 30 year yielding just over 2 percent. The E/P ratio on stocks is not that much higher, but it is higher. And surprisingly, stock volatility is actually about the same as the long treasury bond volatility.
This puts the stock/bond relationship on a geometric frontier, to something like this.
Stocks should should have more return and similar volatility to bonds, and bonds aren’t providing much rebalancing benefit due to the high correlation. They are fairly useless right now. If you want to maximize return, maximize Sharpe ratio or any partial Kelly portfolio on the geometric frontier, the choice is some combination of stocks and all cash.
The Great Portfolio Bond Debate
There’s been a debate lately in the investment world about bond’s place in the portfolio with yields so low. People are looking to potentially drop bonds and hold other investments as “protection”. My position is they still deserve a meaningful place in the portfolio as long as they continue to provide diversification to stocks. But if the negative correlation goes away, then bonds don’t provide the same benefit at low yields.
Geometric Balancing will happily hold bonds at low yields as long as they show a negative correlation to stocks. But when that goes away, the strategy drops them. The bond position has been low in the portfolio for a few weeks due to higher correlations. But yesterday the position went to zero, which is rare.
I’ve never seen this happen with real trading. The last time it occurred in the backtest was spring of 2007. It’s not common. In 1979 however, it happened 10 times. Shades of 70’s style markets returning? We’ll see.
If the lower correlations return (or stock volatility spikes), bonds will come back to the portfolio. On June 10th the portfolio rebalanced to:
88% SPY , 0% TLT , 12% GLD
See the Portfolio page for charts and tables.
*All investing strategies come with the risk of loss, including this one. This portfolio may not be appropriate for your investment goals and requirements, and it is not investment advice.
If the 30 year bond has a yield of just over 2%, then how does the 100% bond portfolio have a geometric expected return of 2.5%?
It wouldn’t. The chart wasn’t meant to be a perfect reflections of todays market, just similar.
This is curious. I thought I’ve copied your principles but my algorithm just moved … 100% bonds.
I thought you don’t use the ‘historic’ return in any way for stocks, only a prospective estimation of ‘expected’ return from E/P ratio. Was I wrong?
This is the best place to quick check. The live sheet in here isn’t my portfolio, but it will get you in the ballpark and shows the data for a two asset portfolio.
https://breakingthemarket.com/reflecting-at-the-milestones-implementing-the-partial-kelly-strategy/
Thanks! I’ll definitely need to refresh myself on the underlying logic (didn’t check back for a couple of months but the drastic divergence from your portfolio made me wonder).
Off the top of my head I can vaguely guess that as geo returns of two assets come close to each other and correlation approaches 1, two assets effectively become indistinguishable. So the tiniest fluctuations in volatility could sway portfolio composition drastically.