Generally, I’ve found a well diversified basket (similar to what Bchad mentions) is good for almost double the sharpe of a 60/40. I believe a sustainable sharpe of ~1 can be achieved with this kind of basket.
…you can also just look at Bridgewater’s returns, they’re pretty much the embodiment of the idea.
Historically you can do much better if you make some weighting/risk decisions, but it’s pretty good for a set it and forget it type approach. m2bps.
I am going o take a different tack and ask - what is wrong with diversifying into highly correlated assets?
Correlations can break down over the years. It’s not like EMH guys really know.
Conversely, expecting less correlated assets to stay uncorrelated is also too optimistic.
Ergo, measure risk in terms of magnitude of negative returns and not in terms of volatility. (See Warren Buffett’s rule #1.)
Volatility is important especially if you need money at a moment’s notice, underlining the fact that volatile assets like stocks are not the appropriate vehicle for your need. Otherwise, all that matters is how likely you are to not lose money (including opportunity costs.)
Sorry Greenman, not addressing your OP, but did want to get this of my chest. I don’t invest in bonds. I don’t need to.
“There some things only cash can buy. For everything else there are stocks.”
When $hit hits the fan, correlations break down when you least need it to.
Not sure if I understand. Just because they can break down and they can change into more correlated doesn’t mean the math changes from attempting to limit violatility until those things actually happen. But the not just the correlation but the expected magnititude is important to know if its worth it to me.
Real risk is how a company plans to implements its capex. How industries can benefit from a shift of consumer preference. Or correctly predicting how economy will perform. Covariance, beta, correlation. All voodoo stuff really. Nothing holds.These numbers are all historical and just tell a tale on a person’s investment strategy at that given time. That beind said, they do offer insights if strategy is consistent with decades of data to back it . The last 3 decades we had are amazing in terms of data, we experienced a real bull run in the 80s-90s, a real credit/liqudity problem in 2000’s, and its nice to see how current investment managers are reacting. Gartman who basically goes back and forth on his bull and bear thesis. Marc Faber, who is doomsday man. Or David Tepper who is bullish (slightly nervous). Lotta fear. WB mind set is prolly the best since he’s been doing it the longest.
Seems clear to me that the traditional statistics - beta, sigma, alpha, and such - are useful for managing day-to-day decisions. But then when the stilt hits the fan, you just need to switch to a completely different paradigm. Sitting there saying that volatility and correlations do not matter seems to me as short sighted as the people who think they are the only things that matter.
BChad,
even under normal conditions (not any extreme bubbles or pops) , do correlations matter? Have you observed them to have predictive power for returns?
My gut feeling (worth nothing) is - perhaps in superquant domain like Renaissance Technologies. I don’t see how they would, for an average 401K. But if you have real-world data, I am all ears.
Correlations do not predict returns - they are supposed to predict volatility.
I’d add (or substitute) portfolio volatility.
To some extent, they can predict returns, conditional on knowing other sets of returns, but that is splitting hairs.
Right, OK, but why do they matter? You don’t get more money with stocks with high beta, but with high returns.
EMH says one implies the others but that’s not demonstrated. Is it your or BChad’s position that it actually is?
Analogy: A ski lift takes you slowly and steadily 500 ft up in 10 minutes. A roller coaster takes you 500 ft up in 1 minute, maybe dropping you 100 ft in between ups and downs. Who is the winner?
Cuz they offset with low covariances
good point. As a portfolio manager who focuses on asset classes it prolly does matter more. But from a specific stock analyst perspective, its garbage. I’ve personally tried the whole capm model bit. Utterly useless.
It’s not the # of asset classes that matter but what they actually are.
http://www.efficientfrontier.com/ef/0adhoc/harry.htm
If you can find 4 assets that tend not to be positively correlated with each other, you’re set. Just remember to rebalance every few years.
Although I only lurk on Bogleheads, there’s a long thread on this (search “Permanent Portfolio” there)
If I ever get to sit on the committee that designs the CFA curriculum in my lifetime, I’m gonna do everything I can to have this included in the curriculum…
Greenie, stop worrying about this stuff and focus on something more productive… you think David Swensen worries about this crap?
Get your health in check, spend time w/ your family, and grow your business… you’re not going to be any happier if you figure out the answer to this.
Greenie is going through a mid life crisis.
too many baby mamas
US, Eurozone and Japan returns were similar in the past. That’s different from expecting them to behave similarly in the future. Some markets had multi-decade negative returns before (Japan’s 1989-who knows is still going negative).Bogle may have screwed a few japanese investors in the last few decades. That’s a roller coaster that went mostly down, with some volatility along the way.
As BChad pointed out before, the US stock market also had a really rough time in long periods of the 1800s. We don’t know what will happen in the future, and I’m all for having stocks as a majority portion of many long term portfolios, since they tend to beat the alternatives in most markets and in most periods of time. But buying 30 stocks in a single country forever and calling it a day is certainly not optimal.
What if your country will be the Austria of the next century? The Austrian market recovered the losses from the first world war about the time the Berlin Wall fell (page 37): https://publications.credit-suisse.com/tasks/render/file/index.cfm?fileid=88F22B53-83E8-EB92-9D555B7A27900DAC .
This is an unlikely scenario, not an impossible one - I think it is dangerous enough for people to avoid invest their entire lfe savings in a single stock market, if they can avoid it.
Some asset management firms show their Strategic Allocation’s historical results against a “dumb” stock + bonds portfolio. That’s proprietary data though.
If I recall correctly, this is all over your CFA3 books. They would often link to studies in the text or the footnotes.
But you don’t really need fancy math to have a reasoning. If investment A gets screwed by Risk X, and investment B doesn’t, you’ll better have some B in your portfolio.
For instance, some frontier markets barely have any correlation at all with US stock and bond markets. This is not due to some statistical coincidence, but just because local factors are responsible for most of those companies’ results, and those countries have very little integration to global finance.
Also, check the link I posted before. For an extreme exampke, a “dumb” single country austrian stocks + austrian bonds portfolio would still be bleeding today even if invested in 1900.
Remember that correlation primarily boils down to the *timing* of gains and losses, not the long-term returns. To merit going in to a portfolio, both assets need to have positive expected excess returns. It just means that when one asset performs better than expected, the other performs worse than expected, and vice versa. But the “expected” that we are comparing results to should still be positive. Uncorrelated simply means that the returns arrive somewhat out of phase.
A correlary to this is that the degree of correlation can change depending on your sampling frequency. Stocks and bonds may well have a higher correlation when measured annually than they have when returns are measured weekly or monthly.
The point of using correlation to reduce your portfolio’s variance is that most investors’ ability to invest is (or should be) limited by their ability or willingness to take risk. If you can reduce your portfolio’s variance, that ultimately means you can (barring tail events that also need to be monitored, but may be best managed via derivatives) invest in more of it. So if you have half the volatility, it means you can (other things equal) invest in twice as much of it than you could otherwise, and still stay within your risk budget.
If you are trying to neutralize liabilities (which means you are targeting a given return, rather than a given risk), having half the volatility means you can meet that return with half as much worrying about values in the interim.
Tail events are another issue that requires management, and levering adds its own risk (primarily the risk that margin calls are almost guaranteed to come at inconvenient times, and risk blowing you up if not properly managed). Many times you may have to go for the higher return asset if you are not able to lever, but that doesn’t mean that diversifying and levering isn’t appropriate when possible (particularly if the leverage applied is relatively mild).