Last week I introduced ETFs that offer some hope to turbulence-scarred equity investors. It really is possible to reduce volatility, and the amount of return we have to sacrifice may not be nearly as much as theorists would lead you to expect. Today, I’ll go a step further and build, step by step, an actionable low-volatility Smart Alpha stock strategy you’ll be able to follow for free in the Portfolio123 Ready-to-Go platform.

Why Not Stay on the Sidelines

That is a perfectly fair question. If equity-market risk is inconsistent with your situation, then by all means don’t let anybody talk you back into the stock market. But I would urge that whatever decision you make, you do so in a realistic context. It’s easy to envision getting out in times of stress expecting to jump back in when conditions are right. Accomplishing that is much more difficult. Stock prices discount future expectations. That means you’ll need to re-enter the market when panic remains the order of the day. If you wait for the “right” time, you’re likely to find yourself standing at the terminal looking sadly at a now-empty gate while the flight you missed is settling into cruising altitude. So either way, one’s stance regarding equities should be about a thoughtful consideration of risk and reward, not dreams of timing wizardry.

The Case For Low Volatility Stocks

These are for those who believe in and want exposure to the equity markets, not merely on a trading basis but due to general conviction regarding the asset class, the mega-trend package consisting of population growth, rising standards of living, productivity and education, the resulting rising economic activity and profits, and finally, rising share prices (something that can even happen as interest rates rise if those increases are offset by rising profits).

Mega-trends make for fabulous rah-rah talk. If only we could just ignore 2000-02, 2008, August 2015, etc. We can’t. But if we can find a way to better position ourselves to benefit form mega-trends while doing a tolerable job of making the occasional calamities a bit less calamitous . . . that might work. Today, I’ll try to provide a sense of what’s possible (and an implementation for those who so choose).

The Low-Volatility Secret Sauce

You may be aware of Beta, the numeric measure of volatility relative to the market. It’s useful in evaluating portfolio performance, but it’s an erratic-at-best and sometimes horrendous tool for telling you how much relative volatility a particular stock is likely to experience in the future. That’s because Beta is computed solely on the basis of historic share price data. It pretends volatility is a statistical phenomenon.It’s not. Stock volatility (relative to the market) is the end result of various company characteristics, generally related to the volatility of the profit stream (which in turn, is influenced by the nature of the company’s business, its size (i.e. how internally diversified its operations can be), its balance sheet (more debt means more interest expense and hence a more volatile net-income stream), etc.

Regardless of a company’s fundamental profile, oddball things can and do happen. So it’s possible for very stable companies to have very high betas, or even for the most speculative barely-surviving companies to have very low, or even negative betas (as often happens if the stock bounces around wildly, but in ways that don’t correspond to or may even be negatively correlated with the overall market – recall that beta measures volatility relative to the market). Since, as I discussed in my last post, past performance really, truly and genuinely is not predictive of future outcomes, relying on stock price trends from the past cannot justify assumptions about future volatility – unless we do something designed specifically to support future assumptions. That’s what this model is all about.

The Strategy

I’m going to use Beta, but to enhance the probability that the data I’m seeing comes from substantive fundamental characteristics (the kind more likely to persist into the all-important future), I’m going to measure Beta over three different periods. (There is no universal formula for Beta; anyone can choose to measure it against any statistically meaningful period.) I’ll use monthly returns relative to the market (giving me an opportunity to smooth out choppiness that could result from use of daily or even weekly returns) and do so over the course of the last year, the last three years, and the last five years. This combination factor — I’ll invent a cool name for it, MultiBeta1-3-5 — isn’t perfect. But it’s more likely to result from fundamentals than any single-period inquiry. But just to make sure . . .

I support MultiBeta1-3-5 by also using a pre-set company-quality ranking system I created for Portfolio123 based on margin, turnover, return on invested capital, and financial condition. Strength in characteristics like these should, I believe, lead to moderate volatility in the future.

Finally, I should mention that I chose to run this model against the S&P 500. I want the large-cap effect, the non-statistical phenomenon to which I referred recently that gives blue chips a better chance than small fry of achieving the more stable profit stream we seek. Helping, too, is a cultural phenomenon that makes the S&P 500, in effect, the world’s safe-haven equity exposure, much the way the U.S. Dollar is seen as a safe-haven currency. If panicky capital flows into the S&P 500 in times of stress, this would likely mute the “drawdown” and is exactly what we’re hoping to achieve.

So here’s the Buy protocol:

• Start with a universe consisting of S&P 500 constituents
• Limit consideration to stocks that rank better than 80 (on a zero to 100 scale) under the Portfolio123 Quality Ranking System
• Further limit consideration to stocks with tallies below 0.9 under each of the three beta computations that comprise MultiBet1-3-5
• If more than 20 stocks pass these tests, rank them from low to high based on MultiBeta1-3-5 and choose those with the 20 lowest tallies
• Invest equal amounts in the stocks that make the final list
• No more than 35% of the portfolio can be concentrated in any single sector (this allows for more sector concentration that is typical of general equity portfolios, as should occur if we want to focus on companies with below-average business volatility, but keeps it from getting exorbitantly concentrated in a singe area)

Refresh the model every three months. If any stocks are to be sold pursuant to the following sell rules, reinvest the proceeds equally (in however many positions need to be established to bring the portfolio back to 20 holding, or as close thereto as possible) in the most favorably ranked MultiBeta1-3-5 stocks (those with the lowest tallies). The sell rules are:

• The stock’s Quality rank falls to below 70, or
• The one-year component of MultiBeta1-3-5 rises above 0.95, or
• A sale needs to be made to bring an overly concentrated sector down to the 35% celling on sector exposure

The Buy thresholds are not carved in stone: Because we’re dealing with the unknowable future, which may not match anything that can be shown to have “worked” in any set of sample periods, I prefer thoughtful judgment (“heuristics”) combined with dress-rehearsals (looking at stocks that pass a potential model to make sure they’re obeying the sprit of the law) to a pretense that I can credibly predict the exact threshold that will work going forward. Also, that’s why I allow for some slackening in numeric thresholds before signaling a Sell.

Here are the (16) stocks that currently pass the model and make it into the initial portfolio:

Table 1

You can follow the progress of this model, and see portfolio updates, on Portfolio123 (assuming you register for a free membership).

Simulated Results

Recognition that past performance does not assure future outcomes is not a basis for avoiding backtests or simulations. In fact, these processes are extremely valuable as feedback on the efficacy of one’s translation of fundamentally valid ideas into language a computer can understand and process. The main keys to proper testing are (i) to consider only models that are based on sound financial theory that explains why they should be expected to work, and (ii) interpreting results reasonably (i.e. checking to see if the results are consistent with what one can reasonably expect in light of sample-period market conditions, rather than tweaking to try to drive simulated numbers as high as possible).

Here are the results of a 1/2/99 through 9/1/15 simulation (with slippage factored into assumed buy-and-sell prices based on a Portfolio123 protocol that assumes variable slippage rates based on a stock’s liquidity).

Figure 1

Figure 2

Realized beta, during the course of the simulation period, came in on target. Check!

Drawdowns during significant market downturns were significantly less than those experienced by the benchmarks despite the portfolio having been fully invested. Check!

The strategy was way worse than the benchmark in 1999. But we know what happened to those who got it right back then and it was not pretty. Check!

Consistent with financial theory, realized returns came in . . . hmm . . . contrary to what theory suggests, the tradeoff we received for having insisted on lower risk was higher, not lower, return. Huh?

Actually, this is not unique. There’s something out there known as the “Betting Against Beta” anomaly. Details are in the academic paper you can get by clicking on the link, but in sum, capital market theory assumes investors implement return expectations by investing in a market portfolio and leveraging up (margin), to whatever extent their risk tolerances allow. In the real world, many can do just that. But many (many more, perhaps) can’t for one reason or another (such as constraints applicable to an ordinary mutual fund or pension account). So to pursue higher return, portfolio managers have no choice but to get as much as they can from the basic equity pool, so they pursue higher risk (higher beta) stocks. Higher demand (relative to supply) of higher beta stocks causes prices to rise. And in the stock market, paying higher prices translates to realizing lower returns. The result is low return on high beta stocks. Conversely, there’s low demand (relative to supply) for stodgy low-beta stocks. Yes, you guessed it: Low demand translates to lower prices which in turn leads to higher return. So theory is turned on its head by non-financial workaday portfolio management constraints.

We have every right to wonder if sooner or later, the big-money funds will figure this out and start chasing low-beta stocks resulting in the end of the anomaly (i.e., the possibility it may get “arbitraged away”). All I can say is that the anomaly has been active for a long time (and according to one paper, is a factor, along with low-cost margin based on capital from by Berkshire’s insurance units, that contributed to the alpha earned by Warren Buffett) and has remained valid even when the papers were published in 2013. If things change in the future, we’ll deal with it. We’re on the gravy train now, but we can’t be greedy: Below-market returns would be acceptable for this strategy so long as they aren’t so low relative to theoretical risk-adjusted expected return as to drive alpha consistently negative. For now, though, let’s continue to enjoy our bets against beta.