Trial DataTrek Morning Briefings for Free

Thousands of investors and financial journalists rely on Nick and Jessica’s newsletter every day for their thought-provoking work on markets, data and disruption. See why for yourself by starting a 2-week FREE trial below.


Correlations Show Crowded Trades

By admin_45 in Blog Correlations Show Crowded Trades

We’re dedicating today’s Data section to asset price correlations, something we’ve done in one form or another every month for 12 years running ever since the depths of the Financial Crisis in October 2008. While not as widely discussed as valuation metrics or performance data, correlations are important for 2 reasons:

  • Volatility and correlations have a self-reinforcing dynamic. When markets get choppy, it’s very hard to find financial assets that can provide real shelter from the storm. Investors therefore reduce risk, putting further pressure on stocks and bonds.

    That makes correlation analysis a useful signal of market bottoms.
  • Drilling down to investment styles, asset class and sector correlations, you can see turning points in market psychology that are less obvious when just looking at daily/weekly performance. Moreover, history shows that extremes in correlations often signal when popular trades have gotten overly crowded.

    That makes correlations useful for considering changes to portfolio weightings.

We’ll focus on the latter point today – looking for crowded trades – with 3 examples:

#1: US Momentum stocks versus US Value stocks. On Monday MTUM (an iShares ETF which tracks the MSCI Momentum Factor index) underperformed IVE (iShares S&P 500 Value index) by 705 basis points. Given that the Momentum Index has been a rock star for years while large cap Value has gone nowhere since 2017, that was a remarkable relative move. For readers with a statistical bent, 700 bps is 10 standard deviations from the mean based daily return data since 2015.

The following chart puts Monday’s move into a more complete historical context, showing the 60-trading day (about 3 calendar months) correlation of daily price returns between MTUM (Momentum) and IVE (large cap Value).

Two points here:

  • You can see we’re in an unusual spot in MTUM/IVE correlations, at 0.44. That is more than 2 standard deviations (0.15) from the mean (0.77).
  • History shows that when MTUM/IVE correlations are at 0.40 or lower (very close to where we are now), markets are strongly convinced that either the global economy is about to pick up (2017 low for this correlation series) or it is about to weaken considerably (October 2018 trade war, Jan 2020 virus outbreak). The latter pushes capital into Tech and Health Care stocks, and these are routinely at the top of the Momentum Index.

Takeaway: Even though we like some flavors of Value (e.g. Industrials, which we’ll get to in a minute), the last time Value disconnected this much from Momentum by outperforming things didn’t work out for Value stocks. That was back in Q1 2017 (first large trough in the chart above), and MTUM went on to beat IVE/Value 3:1 over the rest of the year. We are not saying this reversal will occur quickly – there may well be a few months of Value outperformance left – but the correlation data says we should be “renting” Value rather than thinking it’s our forever-home.

#2: The S&P 500 versus the Russell 2000, or basically US large caps versus small caps. The chart below shows the same time period as the previous one (5 years, 60-day trailing correlations) using the IVV ETF for the S&P percent return data and IWM for the Russell:

Takeaway: we’ll cut to the chase here and simply say there’s nothing unusual about the current 0.82 correlation between US large caps and small caps. Mathematically we’re well within 1 standard deviation (0.07) from the mean (0.86). Therefore, unlike the MTUM/IVE analysis, capital markets aren’t swinging strongly from betting on big companies to small ones. Even that recent low point visible on the graph (right around 0.75) is not at 1 standard deviation from the mean. We still like US small caps and this analysis tells us that markets have not overly tilted to that point of view, a good thing in our book.

#3: US Large Cap Industrials (using the XLI ETF as a price proxy) vs. US Large Cap Technology (using XLK). Same format/time series chart here:

Takeaway: the current correlation relationship between large cap Industrials and Technology shows that markets have certainly disconnected the 2 sectors more than usual, but not as much as Momentum/Value in point #1. At present the 60-day correlation is 0.51, which is basically spot on one standard deviation (0.18) from the mean (0.72). This indicates enthusiasm for the Industrials>Tech trade is noticeable, but not so stretched as to make it crowded. We’re still in the camp that says there is more potential earnings leverage in Industrials than Tech going into 2021 and therefore still like the group.

Summing up: these 3 pairs are hallmarks of the push-pull dynamic playing out in capital markets between “virus fears” and “economic reopening”, and what we’ve shown you today is that market sentiment is not evenly distributed among them. Small caps remain the easiest group to like, followed by large cap Industrials. Value is, for all its current buzz, more problematic given it is already skewed so far from Momentum.

Trial DataTrek Morning Briefings for Free

Thousands of investors and financial journalists rely on Nick and Jessica’s newsletter every day for their thought-provoking work on markets, data and disruption. See why for yourself by starting a 2-week FREE trial below.