Having a portfolio 100% invested in stocks will more than likely generate the best returns over time and can make sense for someone who is very young and can afford to take risk.However, as time horizons and risk tolerance decrease as the portfolio grows, it’s generally a good idea to diversify and cut down on the volatility in the portfolio. Moreover, a 100% stocks portfolio is rarely the most efficient approach to portfolio allocation. Unless one is an expert stock picker who is very adept at getting in and out of the market at the right time, it will generate solid returns over time but at the expense of an unnecessary degree of volatility. It’s fully possible that one can generate stock-like returns but at half the volatility of the S&P 500. The key is to accumulate various uncorrelated, (mostly) cash-producing assets that allow the portfolio to perform well in various environments – i.e., when economic growth is high (relative to expectations), when growth is low, when inflation is high, and when inflation is low.Let’s run through all four scenarios and what asset classes you can expect to outperform in each environment:1. High growth, high inflation = stocks2. High growth, low inflation = high-yield bonds, REITs3. Low growth, high inflation = gold (precious metals more generally), commodities4. Low growth, low inflation = safe bonds, such as US TreasuriesYou can get away with a portfolio that includes assets in 1, 2, and 4, and perhaps leave out 3 given gold and commodities will not be expected to generate real returns in the long-run, as they don’t produce cash. They are merely inflation hedges.However, including gold or precious metals in a portfolio can be beneficial for the sake of stamping out further volatility in a portfolio. If we do run into a low growth, high inflation period, everything in your portfolio in 1, 2, and 4 is only going to perform so-so, if not poorly. Gold will not generate “offense” over the long-run, but it will generate its fair share of “defense,” which means it possesses a fair amount of value from that perspective alone.For additional exposure, you can also toss in emerging markets and international bonds. I ran simulations for forward-50-year returns of various portfolios and compared their returns and risk over time. Simulations were run using past return and volatility information, with 10,000 portfolios simulated each time.Portfolio 1 I use data over a 1986 to 2017 timeframe.Over time, this generated 10.0% in real returns and generated volatility – as measured by standard deviation – of 8.7%. The S&P 500 have averaged around 10.4% real returns and 15.2% volatility since January 1986 to the present.Given this is so heavily concentrated in long-term Treasury bonds (50%), the max drawdown – actually a “worst year” reading (i.e., maximum losses over a one-year timeframe) – is listed at just 8.2%, with a standard deviation of 3.0%. Standard deviation largely isn’t a great volatility measure for the returns of financial assets, as it assumes the returns distribution is thin-tailed (it’s fat-tailed, meaning a wider range of expected future outcomes). However, if you multiply the standard deviation by three, you’ll see that ~99.8% of the time the maximum annual drawdown (over a 50-year period) shouldn’t be larger than 17.2% (8.2% + 3 * 3.0%), given three standard deviations covers this amount of the sample set.Portfolio 2 I use data over a 1995 to 2016 timeframe given data availability for emerging markets only goes back that far.If we include REITs and emerging markets, I break it down into this allocation:We obtain mean returns of 9.3% and a 7.5% standard deviation. So basically we’re slashing the volatility of the portfolio in half at the “cost” of just one percentage point in returns. This is a worthy sacrifice and overall a net gain over portfolio 1.Our maximum drawdown over one-year averaged 9.8% over this 50-year period – again because the portfolio is so heavily concentrated in long-term Treasuries – with a standard deviation of 3.6%. This means 99.8% of the time our maximum annual drawdown shouldn’t be more than 20.6% (9.8% + 3 * 3.6%). Portfolio 3Here we use the same asset types, but allocate just 40% to long-term Treasuries and increase the allocations to US stocks and high-yield.Average returns come to 9.5% with a standard deviation of 8.6%. Maximum annual drawdown, however, increases to 15.7% with a standard deviation of 6.0% (or maximum drawdown of 33.7% using the “three-sigma” rule used above). ConclusionDesigning a portfolio that can come close to the S&P’s returns at half the volatility requires building up a collection of mostly uncorrelated, cash-producing assets and holding them in the right proportion to generate stable, positive returns over time. There should be assets included that perform well when growth outperforms, when growth underperforms, when inflation outperforms, and when inflation underperforms. Accordingly, I believe its beneficial to own some combination of stocks, REITs (which are essentially stocks but have just a +0.57 correlation to the broader market), high-yield bonds, safe bonds, and a little bit of gold/precious metal exposure. Adding in exposure to emerging markets and international stocks can be beneficial as well. Below is an asset correlation table:You will not beat the market most years with this collection of assets (e.g., portfolio 2), but in very bad years, you will greatly outperform. Over time you will come close to matching the S&P’s returns and at a much lower volatility rate.