The use of style factors within a diversified investment portfolio is an efficient long-term strategy. Like all investing, there are positive periods and some negative periods but if investors trust factors for the long-term, then it has been proven to yield higher risk-adjusted returns.

Investors who stick with their style factors for long periods of time are harvesting the factor risk premium. By constantly being invested in factors, investors allow themselves the time to deal with any potential under performance whilst also continuing to benefit from the diversification that a set of low correlation strategies generates. Style factor investing is therefore not short-term trading and is instead part of an overall diversified, long-term portfolio.

Just like investing in the overall market, it is essential to hold these factor portfolios for long periods of time to ride out some factor underperformance that can occur and which could last for a few years. It is impossible to tell ahead of time which factors will do better in the future and we can only use the historical performance as a guide. It is also recommended to hold a blended combination of factors, which benefits from the diversification of various uncorrelated strategies.

Cumulative performance plots of some style factors compared to MSCI World Index. Source: MSCI

In reality, most investors care about risk versus return. The idea that an investor has a maximum amount of risk they would like to take is an intuitive concept and for some investors is as, or even more, important than return. Therefore, there are lots of different strategies competing for this risk budget. If there are ways to use uncorrelated strategies that will lower your overall risk, this is a very powerful concept.

Risk-adjusted returns scale the returns achieved by the risk taken to achieve those returns. A 20% annual return with a portfolio volatility of 40% annualised is not the same as a 20% annual return with a portfolio volatility of 10%. Scaling returns by the risk taken to generate those returns gives us a sharpe ratio. If we work with daily returns and assume a risk-free interest rate the formula for the sharpe ratio is

The rolling 1y sharpe ratio for the S+P 500 during the post global financial crisis period

If investors want to use factors to quantitatively invest, they have a choice between

• using the traditional construction of a long-short, market neutral portfolio which general has very low correlation to the overall market

• form a long-only portfolio (similar to a smart beta) to tilt their portfolios to the factor but maintain long market exposure.

Robeco produced a nice paper "when equity factors drop their shorts" analysing the performance difference between a factor investment strategy that is the traditional long-short versus one that is long-only. This is quite an important research piece as it identifies the successfulness of long-only factors, which are the majority of retail focused investments. Long-only factors are also popular with institutional investors who are benchmarked to market indices.

To create a more market neutral long-short portfolio, the paper also applies a natural pseudo-short of selling an index ETF, such as the SPY, to create a short side to sit alongside their long factor portfolio. Therefore this allows the creation of an overall market-neutral factor if an investor so desires.

Whether an investor wants to employ a long-only factor strategy or a long-short strategy depends on various criteria. Investors who employ long-short strategies are mostly concerned with

• generating absolute return, e.g a diversified strategy that can generate return independent of the overall market direction
• the ability to apply leverage to their portfolios
• are allowed to short by their broker and/or operate in countries with no short-selling constraints

Investors who focus on long-only factor investing are more numerous and widespread. They tend to focus on

• outperforming a benchmark index
• tilt their portfolios to concepts or factors they believe in
• always want to be invested so their risk budget is put to work

Some participants attempt to employ factor timing. Factor timing is where an investor wishes to significantly increase or reduce their exposure to a factor depending on how they think it will perform in the future. This can be employed via discretionary weighting schemes or they can employ a quantitative method to shift your factor weights according to some trusted model.

Factor timing is a difficult thing to implement correctly, even though it has very natural appeal and makes sense - much like you weight stocks in your portfolio with different weights due to your conviction, the same can be said of factors. A recent paper by Gupta and Kelly at AQR entitled "Factor Momentum Everywhere" showed that by timing factors based on their recent performance, investors could earn higher sharpe ratios.

Factor mean-reversion is one model that may offer a nice halfway house that avoids simply equal-weighting your factors and instead allows some tilting of your factor weights around some constant weighting. This can therefore allow factors that have performed poorly to be allocated less weight in your portfolio so less than the average weight but not zero weight. Alternatively, factors that are doing great are again weighted higher than the average but never allowed to grow to totally dominate your portfolio.