Some people hate big losing years and they want to minimize that as much as possible. I asked Solver to come up with a "chicken" portfolio, the one with the smallest losing years. I run into two interesting options This one with -2.3% avg down year And this one with -2.23% Notice how Solver was forced to add gold to cut down on losing years. Also, it was forced to take some risk in stocks, which actually cut down risk
Looking at that previous portfolios with the highest Sortinos and Shape Ratios, the message from history seems to be "buy as much in stocks as you can safely hedge, but no more than that" The Sortino ratio seems to be superior to the Sharpe because the Sharpe punishes upside volatility (which people dont really care about). The Sortino focuses on downside vol and as a result, looking at that ratio only, that portfolio has 50% in stocks That was a lot more in stocks than I expected, and its a lot more than what Dalio recommends. So, right now this is the challenge to me, to understand this discrepancy. I'm sure Bridgewater has run the same numbers that I'm running now and maybe they even went further by 'simulating' the future by creating 500 years of future market returns (with inflation/deflation, high growth/low growth cycles in it) and calculated the best portfolio. Whatever they did (and I'm sure they spend millions and millions a year trying to improve of their All-Weather strategy) they were dissuaded from owning so much in stocks, at least on that Dalio public interview. Dalio only wants 30% and he is balls to the wall long high duration bonds.
I think I can understand the Dalio choice in the bonds a little more, Tony Robbins said the portfolio couldn't have leverage because people dont understand it/have access to it. One way to create 'synthetic' leverage is by juicing the duration of the bond portfolio. So maybe that was what he was doing. In his fund, he is perhaps long 120-130% in assets with the extra part coming from bond futures but in that public interview, that was not an option. So he did the next best thing, which is put a lot of duration in. But still, that doesn't explain why he owns so little in stocks, especially because he could have decreased some of the 15% allocation that he has in intermediate-term bonds (7-10 years) and increased it in stocks. It would still be consistent with the principle of 'buy as much in stock that you can hedge, but no more'. One could argue that his All-Season strategy is overly hedged, 30% in stocks with 40% in long bonds plus 15% in intermediate bonds plus 15% in commodities is very conservative. Maybe that's what he was going for? Minimizing max drawdowns as much as possible so the average person wouldn't panic? Perhaps. Looking at the data and using Solver, it does not look like Ray Dalio was going for the "efficient frontier" with that portfolio recommendation, it appears that he was going more for downside volatility control for a retail audience
I'm aware that one has to be careful with relying in history, there is the whole 'curve fitting' issue with systematic trading programs. However, if 50% in stocks 32% in 10y bonds and 18% in Gold is not the next 100 years Efficient Frontier in the US, I doubt it will be dramatically different from that. Maybe the right allocation is not 50% but 40% or 60%, but its unlikely to be 20% or 10% or 70%. Same thing with bonds, maybe its not 32% but its unlikely to be 15% or 55%. Gold, same thing. At least, that's how I'm thinking about the data. 1928-2016 is a pretty big sample, there was a lot of things that happened in that period
I run a test for the highest Sortino ratio portfolio that cannot use Gold vs It does worse in risk ratios (Sharpe and Sortino) as well in stuff like SD of Negative Returns, max drawdown (although that field I havent finished coding it and it bugs out sometimes, it cant be relied upon). Worst year, etc. % of up years is also worse
I meant the next 100 years Tangency Portfolio But I dont think trying to achieve the "Tangency portfolio", that one with the highest Sortino/Sharpe ratio, makes all the sense in the world. I'm not going to try to hug that point of the Efficient Frontier at all costs. Sometimes there might be some great opportunities and I'm not going to pass them because my Sortino ratio will drop. I think a mix of going for the Tangency and pulling a Buffett ("I rather make a bumpy 15% than a smooth 12%") is optimal. But I do think its nice to know where that point has been, historically, just so one can learn some principles of low volatility portfolio creation, which can then help you when building your own portfolio so you can pull a Buffett but more safely/conservatively I want the best of both worlds, to grow capital at good rates, while taking on resonable risk. The alternative to the Bufett's "take more risk even if it will increase volatility a lot" is to borrow and invest in that Tangency Portfolio. Its an interesting idea and I think Dalio does some of that in his All-Weather fund but its kinda hard to do because you need a safe source of leverage or you can run into trouble. Futures is a way so its something that I need to explore more. How to improve returns by using leverage while allocating the extra capital to the high Sortino ratio portfolio
Maybe what makes sense is going for is the Markowitz-Buffett point (which I just made up), somewhere in between the Tangency Portfolio (the highest Sortino portfolio) and the Buffett "all-in" portfolio (of 100% equities)
Perhaps something that can explain Dalio's small 30% weighting in stocks could be if Bridgewater is looking at monthly data instead of yearly data (like me). If they are, then in a month to month basis stocks will be punished a lot more in the Sortino Ratio. Month to month there is crazy volatility that gets dilluted once you smooth out things by looking at yearly data. As a result, when they calculated their Tangency Portfolio, it came out with less tolerance for stocks in it. Dalio then went and lowered even more to make it easier to to a retail crowd to deal with. This paper uses monthly data and the guy said Bridgewater gave him the data. http://www.advisorperspectives.com/newsletters12/pdfs/Why_a_60-40_Portfolio_isnt_Diversified.pdf That could explain it
Daal, I see you point about yearly data, but let's assume people will be uncomfortable with drawdowns within a year that get smoothed out on a yearly basis. What about taking a quarterly view?
Its another option, I will try to get that Bridgewater short-term data to see if I can run tests on it. Lets see if I can get a hold of that