I bought O’Shaughnessy’s What Works On Wall Street because I wanted to understand the purely statistical drivers of stock market returns. I was fed up with the subjectivity of investing and believed only Warren Buffett had the edge to value qualitative factors such as competitive moats and pricing power accurately. The whole thing seemed incredibly over-engineered. The fact that active managers, who demanded higher fees and marketed their professionalism, were incapable of beating the market over the long term supported my viewpoint. This is true both in the United States…
and in Europe…
According to O’Shaughnessy, the dismal performance of fund managers can be attributed to their human nature. It is not that these professionals cannot find exploitable investment opportunities – they simply have difficulty sticking with sound strategies through thick and thin (periods of over- and underperformance). Indexing, on the other hand, never varies. It eliminates the human element entirely and thereby sidesteps flawed decision-making, inconsistency, and a host of other limitations that plague active managers. The thought that a strategy as simple as buying the stocks that make up an index – such as the S&P 500 – has such enviable performance relative to active management says it all: markets are inefficient as a result of human psychology.
Sure, investing in an index fund is an excellent choice. However, if we know that the driver of index funds’ long term performance is simple consistency and a defined set of quantitative rules that minimise human influence, we can take it a step further. Indeed, through identifying the correlations between a wide range of valuation ratios, yields, various composite factors, value-momentum combinations and stock returns by means of backtesting, we can figure out which quantitative strategies have beaten or trailed the market historically i.e., we learn What Works On Wall Street.
O’Shaughnessy used two datasets in his studies: the Standard & Poor’s Compustat Active and Research Database (1963-2009), and the Centre for Research in Security Price (CRSP), which covers 1926-2009. This mountain of data was filtered and analysed in the following way:
- Stocks with inflation-adjusted market caps below $200 million were excluded
- Delisted stocks (mergers, bankruptcies, etc.) were included to avoid survivorship bias
- To ensure only publicly available data were used for calculations, quarterly and annual financials were time-lagged by three and six months, respectively
- 12 separate portfolios were formed every year for each strategy, one every month, and the results were averaged to counter seasonality effects, and the rebalance period for each monthly portion of the portfolio remained one year
- Transaction costs and bid/ask spreads were excluded, as these vary too much between investors
- All returns, both compounded and absolute, are real (inflation-adjusted)
Valuation ratios, yields, and composite value factors
The graph below compares the compounded returns an investor would have achieved between 1927/1964 (depending on data availability) to 2009 through holding stocks which ranked either in the lowest decile based on common ratios (P/E, EV/EBITDA, etc.), or the highest decile regarding certain yields (dividend yield, buyback yield, etc.). In other words, these are all stocks which represented ‘good value’.
Interestingly, all strategies outperformed the market. EV/EBITDA seems to be ‘king’ – stocks which ranked in the lowest decile outperformed stocks ranked in the lowest decile based on all other ratios. We could speculate that this may be due to the ratio being a fairer and more generalisable assessment of value as it ignores differences in capital structures, capital expenditures, and taxes. In contrast, the P/E and P/CF ratios, which are close behind in terms of returns, do not ignore these factors. P/B value seems to be the ‘worst’ ratio, as stocks in the lowest P/B decile had the poorest returns in the group. Perhaps book value has become a poorer indicator of value over time, with asset-light technology companies rising to prominence.
Knowing the importance of compounding, let us visualise these findings in absolute terms – how much would $10,000 invested in 1964 have turned into with these rates of return, after adjusting for inflation?
From this graph, it’s evident that minor differences in returns can produce vastly different outcomes, illustrating the importance of both compounding and picking the right ratio. Although all of the above information is useful, O’Shaughnessy didn’t stop there. He also analysed the performance of stocks which ranked in the highest decile based on these same ratios and those which ranked in the lowest decile based on yields – stocks which are perceived as being of ‘poor value’.
We can see that stocks with the highest ratios generally made for terrible investments, and all of them underperformed the market. That shouldn’t be surprising – we have a tendency to assign unreasonable expectations and valuations to companies that they cannot possibly fulfil, leading to bubbles that inevitably burst. The 10% of stocks with the highest price-to-sales ratios performed worst of all, whereas the top decile of P/B value and dividend yield stocks actually performed relatively well. Again, this is logical: in the modern era, book value has become less important, and many high-growth companies (which could produce excellent returns) choose to reinvest capital instead of paying dividends.
However, questions remain unanswered. Would it make sense to pick a basket of stocks based on EV/EBITDA, and expect such a portfolio to outperform not just the market, but all other ratios and yields over the long-term (10 years plus)? Although this may seem like a minor concern when we consider that stocks in the lowest decile of ratios and highest decile of yields already produce market-beating returns, it is still worth exploring. In fact, O’Shaughnessy states that “an ongoing horse race” exists between these factors, leading him to develop and backtest three ‘Composite’ Value Factors that combined various ratios.
- Value Factor 1: PB, P/E, P/S, EV/EBITDA, and P/CF
- Value Factor 2: PB, P/E, P/S, EV/EBITDA, P/CF, and shareholder yield
- Value Factor 3: PB, P/E, P/S, EV/EBITDA, P/CF, and buyback yield
The performance of stocks that ranked in the lowest decile of individual ratios, the Composite Factors, and the highest decile of yields is shown below – these are the ‘good value’ groups.
As it turns out, the Value Factors outperformed all of the other ratios and the market, demonstrating the power of combining different ratios. Again, vice-versa, let’s show the performance of stocks which ranked in the opposite deciles (i.e., the ‘poor value’ groups).
All stocks which ranked in the highest deciles of financial ratios and Composite Factors, and the lowest decile of yields underperformed the market. Stocks with the lowest Composite Value Factor scores performed worst of all, as they were the most holistic indicators of value.
The Trending Value portfolio
Besides analysing valuation ratios, yields, and value composites, O’Shaughnessy tested dozens of other strategies, which are beyond the scope of this article. Nonetheless, I thought it would be interesting to discuss the Trending Value strategy, which came out on top in terms of risk-adjusted return (Sharpe ratio). This strategy picked stocks based on the following criteria:
- Stocks were in decile 1 of the Composited Value Factor 2 (the ‘best value’ stocks with the highest 10% of composite scores across all ratios)
- Buy the 25 and 50 stocks with the highest six-month price appreciation
This strategy produced a real return of 21.19% between 1964-2009 for the 25-stock portfolio, turning $10,000 into an astronomical $69,098,587. It beat the market in all 5 and 10-year rolling periods and had the highest Sharpe ratio of all tested strategies. The 50-stock portfolio produced a return of 19.85%, which would have snowballed $10,000 into $41,411,163. It beat the market in 99% of 5 and 10-year rolling periods. All characteristics of the strategy relative to the market are shown below.
We can see that the 50-stock portfolio swapped marginally lower returns for less volatility (beta). Finally, let’s visualise all of the returns described in this article.
“Finally, the data prove that the stock market takes purposeful strides. Far from chaotic, random movement, the market consistently rewards specific strategies while punishing others. And these purposeful strides have continued to persist well after they were first identified.”Jim O’Shaughnessy, What Works On Wall Street
So what can we learn from all this? That depends on your experience and goals. Personally, I learned the following:
- EV/EBITDA, P/CF, and P/E are the three best individual ratios, and should be a criterion or criteria in every stock screen
- Combining ‘Composite Value’ scores with momentum produces excellent risk-adjusted returns over the long-term which exceed those of the market and all single ratio portfolios
- Beating the market does not require a complex strategy – it can be done with the consistent application of a defined, simple set of quantitative rules over the long-term (decades), as was made clear in The Little Book That Still Beats The Market (magic formula), The Intelligent Investor (net-nets) and Deep Value (acquirer’s multiple)