Catching Moving Markets

Systematic Application

We now know that anomalies partly explain excess market returns by hedge funds and investors over both long time periods and different economic regimes. By using a simple factor based investing approach, Warren Buffet’s outperformance can for example be explained by something called the value and quality factors, which research has shown to be persistent anomalies explaining outsized investment returns in equity markets. Important to note is that we should not take away any thunder from Warren because he knew about these anomalies before anyone else. Other known factors include momentum, growth, small cap, term and carry and behind these factors are risk based and behavioral explanations. The logic of the former is that investors gets compensated for higher risks in the market and the latter is that we drive market prices up and down due to our human nature.

Mr. Buffet

Markets

A market is generally defined as the sum of buyers and sellers in a specific region, industry, or product, which means people will define the market differently depending on their perspective. Pensioners might define the S&P 500 equity index as the market, while advertisers probably consider the market as a segment of potential buyers and sellers of the specific product that they are looking to advertise. Investment professionals generally cares about their investments and therefore measure the portfolio success against a standard index. A bull market for an investor is when the standard comparison index is generating high returns. For example, a global sovereign fixed income trader might consider the market as all traded treasury bonds and a large cap equity trader would probably reference their market as all corporations with a market value in excess of 10 billion dollars. Both the fixed income and equity trader do care about drivers behind markets. Using different variants of factor analysis through applying statistical techniques is a common approach to decompose market drivers in both fixed income and equity.

NYSE - stock exchange

How a quant describes the market

Financial market prices move in response to supply and demand. When a product is in demand it will increase in price all else equal. The demand for playsation 5 in 2020 is a perfect example, as its online release in response to the Covid-19 pandemic meant that hackers were able to purchase multiple consoles using algorithms before any human. These hackers then resold the playstation 5 at a higher price on eBay and pocketed the difference. Behind this demand are popularity factors ranging from superior processing power to how many other people are using the system. Just like hackers are able to buy and resell the playstation 5 at a premium, talented quantitative traders are able to spot in advance the assets increasing or decreasing in price because of factors driving demand. Quants often rely on some type of system when making predictions of the future price of equities, bonds, commodities and currencies. This system generates signals corresponding to these underlying drivers or factors. The most important factor determining the price of playstation 5 is most likely the amount of other people that are using the system. The logic behind this conclusion is that gaming is more fun if you play with friends or against opponents. Figuring out the underlying drivers is slightly more difficult for a trader than a hacker because there is a lot of noise in financial markets. However, quantitative analysts can apply statistics to find the most significant factors driving the market.

Correlations between factors

What is a factor?

A factor is a common price driver shared by a number of financial instruments. Data scientists can find an infinite range of factors in the market, but many of them are not profitable to trade. To separate signal from noise a skilled data scientist must therefore decide what makes a factor profitable. Two professional quantitative researchers, Andrew Berkin and Larry Swedroe describes exactly that in their book “Your Complete Guide to Factor Investing”. They created a system consisting of five components to define factors. The first component is persistency, as a factor must work over a long time period and across different economic regimes. If buying small cap stocks only work when the tax rate is below 15% that is not a persistent factor. The second component is pervasiveness, which means that the factor must work in different markets across different geographical components. For example, if small cap stocks only outperform in European markets, but not in others, then the small cap factor is not pervasive . The third component is robustness meaning that the factor holds for various definitions used to measure the same thing. There are multiple performance ratios used to measure a firm’s value including price to earnings, book and sales ratios. For example, if low P/E ratios contradict low P/S ratios, then value might not be a significant factor. The fourth component is investability. If investing in a factor is impossible or if the trading costs makes trading expensive, then that factor is unrealistic. The fifth and final component is intuitiveness of the factor. According to Berkin and Swedroe, there must be a risk based or behavioral reason for why a factor exists (there are machine learning approaches that can find significant factors that doesn’t necessarily fly under a logical banner).

Example of a factor diagram

What are some well-known factors? Berkin and Swedroe together with the investment community have identified more than hundreds of factors meeting at least one of the five outlined factor criteria and this investment area has therefore been nicknamed the factor zoo. Just like a zoo has many different animals, a financial market zoo has many different assets with exposure to factors. However, Berkin and Swedroe argue that currently only eight factors fully meet the five factor criteria, as outlined in “Your Complete Guide to Factor Investing”. Some factors pass away with time, while others work over a special period, narrow band or regime. Finding profitable historical relationships that fails to continue into the future is relatively easy for investors relying on data science. Identifying these spurious relationships is also known as data mining. Data miners tend to overfit their models to data and their models are of poor quality. Another example of a bad factor is if it can be replicated by a combination of other factors. For example, past performance is related to market capitalization. If performance of a stock is good then there is a higher likelihood of a high market cap all else equal. Creating a weighted average of these two factors through different econometric techniques is one path to avoid collinearity. However, Berkin and Swedroe argues that only the following factors have stood the test of time.

What Factors Exist?

Figure 1

Outperformance numbers of known factors from Swedroe and Berkin.

Market Beta

In finance, market beta is how an asset moves in relation to the overall market. Mathematically, market beta is defined as the correlation between the asset and market return multiplied by the ratio of the asset’s volatility to the market’s volatility with volatility measured by standard deviations. The broad market has a beta = 1 and all other assets are below or above 1. Risky assets move more than the market and has a beta above 1. Some assets have a beta below 0 and these assets move in the inverse direction of market beta. Figure 1 shows that the market beta factor has a sharpe of around 0.4 and a risk premium of 8.4% annually. The market beta’s risk premium can be explained by its correlation to the economic cycle. In a recession, the market beta premium will underperform, as companies have a higher likelihood of default. According to Swedroe and Berkin, a second reason for a high risk premium could be that the stock market mostly consists of wealthy investors. In general, wealth has a diminishing marginal utility so the only way wealthy people will invest in stocks is if they get compensated with a high risk premium.

Size

The size factor most likely exists because investing in smaller firms involve taking higher risk. The risk based logic behind the size premium is that smaller firms have greater leverage, less profits, higher volatility, lower capital base and less liquidity. As a result, investors demand a higher return from investing in small firms. In good economic times, small firms tend to outperform, but in bad times they underperform due to a higher chance of bankruptcy. The size factor premium is therefore correlated highly to the economic cycle. The risk-adjusted (sharpe ratio) return ratio for the size premium according to Swedroe and Berkin is 0.24, which means it is on the lower side for a factor. A smart approach to extract a higher size premium is to adjust for firm quality when conducting the investment analysis by excluding junk firms. Research has shown that this type of junk screening lowers risk and increases the sharpe ratio.

Value

According to Swedroe and Berkin, the value factor has a relatively low sharpe ratio of 0.34, while the annualized value return premium is around 4.6%. The risk based logic behind the value premium is that value firms tend to have greater amounts of unproductive capital. For example, value companies might have factories that are not needed during times with low economic activity, which means that in bad economic times there is higher chance of value companies being distressed. Value firms also tend to take on more debt leading to a higher degree of leverage. Finally, value firms are more sensitive to interest rate. When rates increase, the value firms tend decrease faster in value than growth firms and the value premium increases drastically. In the end, stocks that do poorly in bad times should have a higher risk premium all else equal. The behaviorists on the other hand believe that people are overoptimistic of growth stories and as a result undervalues these boring value companies. Other behavioral reasons include biases such as anchoring and confirmation, which leads to investors to disregard contrarian information. When investors hear negative stories about a growth stocks they are simply more likely to ignore it. This leads to a value premium.

Momentum

Momentum has the highest annual return premium at 9.6% and sharpe ratio at 0.61 making the momentum premium the most well performing factor according to Swedroe and Berkin. The two types of momentum that exists are cross sectional and time series momentum. The former is a relative measure, while the latter is an absolute measure. In the cross sectional approach, the manager goes long assets with high momentum and short assets with low momentum. In time series momentum, the manager goes long assets in which prices trend higher and short assets for which prices trend lower over a certain time span. The logic behind momentum is mostly behavioral based, as investors overreact to recent winners and underreact to recent losers. This leads to a price a that trends away from its current efficient price. Another observation made by Yale professor Thomas Moskowitz is that people underreact to information travelling slowly into prices. The frog in the pan effect is commonly used to explain Moskowitz theory. If heat is turned up quickly the frog will immediately jump out of the frying pan, but if the heat increases slowly the frog marginally gets more comfortable until it is too late. The risk based explanation for momentum is that past winners are expecting higher growth with risky cash flows. If expected higher growth fails to materialize then that will negatively impact the price of these recent winners and momentum traders therefore get compensated accordingly. Two other risk based explanations relates to the liquidity risk associated with recent winners and mutual funds. Recent winners demand higher compensation since they have greater exposure to liquidity risk. Mutual funds are also sensitive to liquidity risk. When mutual funds experience outflows they need to sell assets at a discount price and vice versa for inflows, which can lead to momentum.

Profitability

Swedroe and Berkin defines the profitability factor as the ratio of gross profits to assets. The idea behind the profitability factor is that profitable firms generate higher future returns even if they have higher valuation ratios all else equal. The annual return premium is 3.1% with a sharpe ratio of around 0.33. With the generation of high revenues, profitable firms can expand quicker. This factor in combination with the value factor will lower portfolio volatility and increase returns, as they increase monthly returns of a portfolio to 0.71%., while only earning 0.31% and 0.41% separately. The risk based reason for a profitability premium is that much of the cash flow for growth firms is expected to be generated in the distant future, which is risky and thus requires a risk premium. The behavioral explanation for the profitability premium, according to research by Huijun Wang and Jianfeng Yu, is that profitable firms have an information uncertainty associated with them. This information uncertainty decreases investors ability to arbitrage trade and enhances overconfidence as well as other behavioral biases in the pricing of highly profitable firms.

Quality

According to research, firms of high quality generate higher future returns over those with lower quality. That is because high quality firms tend to have less leverage and fraud, while at the same time having higher management quality and diversity of stable revenue sources. The average annual return premium is 3.8% and the sharpe ratio is around 0.38. One of the largest users of the quality factor is Warren Buffet. He loves buying large, safe and profitable firms and the profitability/quality factor has therefore earned the nickname buffet’s alpha. The premium for quality firms is related to the profitability factor and most likely exist because of information uncertainty. Quality firms tend to do better in times of crisis due to being undervalued, as people tend to hold on to their growth favorites rather than investing in these boring firms.

Term Premium

Based on data from 1927 through 2015, Swedroe and Berkin has defined the term premium as long-term (20-year) U.S. government bonds minus one-month U.S. Treasury bills. This premium was statistically significant at 2.5 percent. The risk based explanation behind the term premium is intuitive, as longer duration means a higher probability of getting hit by unexpected inflation, which means that longer term bonds are riskier. The term premium is a great addition to an overall portfolio because it is negatively correlated to many other factors.

Carry

The carry factor is derived from borrowing in a low yielding currency and investing that amount in a high yielding currency. If the low yielding currency does not appreciate in value, the carry investor is profitable. However, sometimes currencies appreciate unexpectedly, especially in periods of financial distress, which is a risk based explanation behind the carry premium. For example, capital tends to flow to low yielding safe haven currencies in weak equity markets, which means low yielding currencies provide protection for most investors. At the same time this is a risk for carry investors. Carry is possible to extract from currencies, commodities, equities and fixed income and the return premiums with associated sharpe ratios can be found in Figure 2.

Figure 2

Carry outperformance

//Stay adaptive!!

1. Berkin, Andrew L.; Swedroe, Larry E.. Your Complete Guide to Factor-Based Investing: The Way Smart Money Invests Today BAM ALLIANCE Press. Kindle Edition.