Notes on OSAM’s Factors from Scratch

How does the ‘value’ factor generate excess return?

  • Method for decomposing the return into earnings growth and multiple expansion
    • Limit to large caps (conservative, since the factor is weakest here)
    • June 1964 – Oct 2017
    • Cheapest quintile of stocks rebalanced annually at the end of June
      • Rebalance averages about 38% turnover
      • Turnover requires technical adjustments to normalize the decomposition: “rebalancing growth” and “unrebalanced valuation change” [details in the paper]
  • Findings
    • Value stock fundamentals do deteriorate in the holding period. This is reasonably expected since these are the cheapest stocks. But the prices turn out to be overly pessimistic, as the multiple expands during the holding period more than enough to compensate for the decline in earnings.
    • The excess return becomes highly diluted after 1 year as the bulk of the market’s re-rating of the stock occurs within the first year. 1 year maximizes the “gain to time invested ratio” and also strikes a reasonable balance with the costs to rebalance
    • The market’s re-rating of the stock higher proves to be vindicated as fundamentals do stabilize. The value factor is capturing the ‘overreaction’ discount and the rebalance sells the re-rated names into the market’s stabilizing bid.
    • Value is difficult or impossible to time since it only correlates reliably with future returns at market extremes (ie dot com era).
  • ‘Value’ in recent context
    • Values has underperformed since the financial crisis although the underperformance is not unprecedented
      • Value is even cheaper today on relative basis compared to the overall market but both value and the market are about 50% more expensive than historical averages
      • Value is not currently stretched despite underperformance
      • Value may be underperfroming since cheap stock’s implied underperformance turned out to be even higher than their subsequent realized underperformance.
      • Likely that this is bad luck as opposed to the market having become better at handicapping future performance.
      • From Asness: ‘In stock selection, value is still not super cheap (i.e., super-cheap would be if the cheap stocks were way cheaper versus the expensive ones than normal).78 It would be fair to wonder why not, especially given the poor long-term value returns. Well, with any strategy, you can lose because either prices or fundamentals move against you. Unfortunately, more of this current drawdown has been about fundamentals. Value, at least using the behavioral. explanations, is a bet that prices over-extrapolate current prospects. The better companies deserve to be priced higher versus fundamentals, but even so, they’re priced too high (and vice versa). However, sometimes prices are correctly reflecting this information, and sometimes they are actually underreacting to it (meaning what looks expensive is actually an ex ante good deal). Prices may over-extrapolate on average, that’s why value works long-term, but not all of the time. Value wins more than it loses, but when price differentials underdo it (meaning, unlike most of the time, cheap companies aren’t actually cheap enough versus the expensive ones) is a time that value fails.79 Importantly, we find no signal from this analysis for timing value going forward. Value is not predictably bad or good following periods where fundamentals move against it.’

How does the ‘momentum’ factor generate excess return?

  • Momentum factor constructs equal weighted basked of top quintile of names based on prior 6 month returns
  • Findings
    • Recent returns are a better predictor of earnings growth than simple expensiveness
    • Decomposing returns we find that the resultant earnings growth exceeds the size of the multiple contraction
    • Unlike ‘value’ which leads to excess returns for many years (albeit at a declining rate after year 1, the momentum strategy is mean reverting. The excess return actually overshoots in year one and subsequent years actually show underperformance.

Combining ‘value’ and ‘momentum’

  • Value converges to fair value after initial overreaction which leaves them overly offered, momentum diverges above fair value at the tail end of the holding period as shares become overly bid.
    • This makes them complementary timing-wise over the 1 year holding period
    • They work best in opposite market environments

Digging further into factors

The above describes how, in general, these factors work (distilled to their essence on average you are betting against overdone price declines in companies facing headwinds or trouble). [The strategy is innately convergent and supplies liquidity]. However, when digging into this general dynamic further, OSAM finds lots of dispersion under the hood. Since that is the case, it makes sense look for what differentiates the names which are favorably re-priced vs those which continue to underperform their price outlook.

To illustrate they show AAPL vs IBM from 2014-2018. Both stocks were in the cheapest quintile of P/E in 2014.

IBM earnings declined, AAPL’s grew and the stocks were predictably punished and rewarded respectively. They validate this is not an anomaly by looking at names historically that are priced similarly and then looking at their performance as a function of earnings growth over the next year. The names with faster-growing earnings outperform those with slower growing or declining earnings and the effects increase are amplified by the degree of growth (fastest growing, perform better than just faster growing on average etc). The paper’s appendix goes further by decomposing the stock’s returns into contributions from multiple expansion and earnings growth.

These findings unsurprisingly apply to the stock market in general — companies whose earnings grow faster, have share prices that grow faster. However, they find this dynamic to be much stronger in cheap (aka value) stocks meaning the rewards for being able to predict earnings growth are higher in the value arena.

This chart shows the ‘excess return vs historical average’ binned by rank of earnings growth. While the names with fastest growing earnings are the relative best performers, we can see how much volatility there is in the entire value factor with periods like the most recent 5 years and periods in the late 1980s being notably poor for the factor.

Zooming in:

How to capture this return with info available at the time?

“Is it possible to reliably identify the top-growing stocks in the value factor using presently-available information? The answer is surely no, especially if presently-available information is limited to price and financial statement data. The forces that determine future earnings outcomes in businesses arise out of complex, idiosyncratic chains of causality that are not fully captured in that data.”

A more reasonable and still very valuable goal is to “tilt” exposure towards the names which are more likely to be indicative of real value vs the ‘value traps’ which bring the value factor’s average down.

  1. Minimize selection bias by using a composite of measures to identify value
    • For example, P/E will be understated if a company takes a one time gain (for example if it sold a balance sheet item for much higher than its accounting value). Measures of value are vulnerable to any accounting variable which is not reflective of the ongoing business. By using a composite of measure the risk of a single accounting aberration having undue influence is mitigated.
  2. Addition by subtraction by removing the value-traps
    • Momentum: a measure of trailing total return; higher is better.
    • Growth: a measure of trailing change in earnings; higher is better.
    • Earnings Quality: a measure of accruals; lower is better.
    • Financial Strength: a measure of leverage; lower is better.

Because low scores in these indices have a disproportionately large impacts they choose to cut of say the bottom 10% as the best trade-off between the desire to avoid the worse outcomes reliably and the desire to maintain a large enough universe of names and diversification. Asymmetrical filter: Scoring poorly on those measures is a better predictor of poor performance than good scores are predictors of positive performance

  1. Create an equal weighted portfolio of the remaining top half of names: “value leaders portfolio”


The results as the portfolio becomes tilted to higher quality value names:

When the process above is used to filter value-traps and we further narrow the universe to the ‘value leaders’ we find that our equally weighted portfolio had much higher exposure to the faster-growing names than a strategy ranked according to simply cheapness (ie “Value:Top Quintile”)

This is the excess return to the traditional value factor in different historical periods:

This table shows how the tilt to the top quintile exceeded 20% in every period

This table shows the returns to strategies decomposed to multiple expansion (ie the point spread re-rating) vs earnings growth

The leaders strategy improves the generic value strategy by eliminating the names which drag on EPS growth (at the lesser expense of having less pronounced average multiple expansion)

The outperformance of the value leaders strategy is notable for three reasons:

  • First, it requires only a modest amount of intervention. The percentage of original value stocks retained in the final strategy–38%–is relatively large. Moreover, the strategy is rebalanced annually, rather than quarterly or monthly. These characteristics suggest that the strategy is able to accomplish more with less.
  • Second, it’s occurring entirely in the large cap space, a space in which factor signals are comparatively weak and, according to some, non-existent.
  • Third, it’s associated with a significant shift in allocation towards the value factor’s top growth bins, a shift that we know is efficacious, given the extreme levels of outperformance produced by stocks in those bins.”

OSAM notes that “In practice, we therefore use methods that are more focused and refined. We also take advantage of the benefits of concentration and size: factor investing is more powerful when applied in a concentrated manner and when used outside of the large cap space.”

One thought on “Notes on OSAM’s Factors from Scratch

Leave a Reply