Friday, August 7th, 2020

Party Like It’s 1999

This is not an original thought, but there is an odd symmetry between the circumstances surrounding this Tech boom and the previous one in 1999/2000. They were the both facilitated by a period of exceptional monetary stimulus in response to a bug – the Millennium Bug in 1999 and Covid-19 today. For those who don’t remember, the Millennium Bug was a theory that the operating systems of old computers would not be able to cope with the change of the century because their calendars were only set up with two digits for a year rather than four. It sounds ridiculous now, but the Fed was sufficiently nervous that it ran a very accommodative monetary policy into the turn of the century, despite that fact that GDP growth was over 4%.

At the time, there was the same rather fruitless debate about high valuations and paradigm shifts that we have now. We say “fruitless”, because it’s impossible to give a number ex-ante for the PE ratio or any other metric, which all investors would agree was too expensive. There are too many variables and too many unknowns. What we can do is observe the balance of risk and return and see whether the extra risk of investing in Tech is justified by the extra returns it generates. This allows to do two things: (1) identify a recommended weight relative to benchmark which precisely balances the probabilities of higher or lower returns; (2) track how this number changes from week to week.

The good news is that our model recommends a weighting in the US Tech sector which is 87% above benchmark. The bad news is that this figure peaked at 99% in early February and was above 90% for almost all of the period between December and mid-July. It is not impossible for Tech regain these levels, but the risk that it won’t is increasing. Last week, the recommended weighting for Tech sector broke down through its 52-week moving average, which itself is the highest it has ever been in the 25-year history of this model. There is no magic in this relationship, but by definition it doesn’t happen very often and it is worth paying attention to previous outcomes.

There have only been seven occasions, where the recommended weight has fallen below its 52-week MAV, while the sector was rated overweight and the MAV was peaking, (which it is now). Six out of seven times, the recommended weight did not bottom until it reached underweight territory. The exception is the most recent episode, in May 2018, where the sector just avoided a downgrade to underweight in January 2019. We see no reason to disregard these results. The best-case scenario is that Tech performs in line with the index for the rest of the year. The worst is that there is a period of extended underperformance, possibly prompted by a return to more normal monetary conditions as the Covid-19 pandemic recedes.

The Covid-economy theme is not just confined to Tech. The Communications sector (Facebook and Google) is currently rated at 38% overweight, just below its recent peak in May 2020. If the current retracement continues, it will break down through its 52-week MAV by the end of August – if not sooner. If these two sectors generate the same signal within a month of each other, the impact is likely to be cumulative. Tech would underperform because Communications was underperforming and vice versa. It wouldn’t be the end of the world – certainly not in comparison with the pandemic – but it would cause a lot of pain to a lot of portfolios.

Leave a Reply

Friday, July 24th, 2020

Signs of Life in the Eurozone

The new story in global markets is the rehabilitation of the Eurozone. For the last 12 months – and arguably for the last four years – it has been outshone by the US almost all the time, and when US investors did look elsewhere, it was either to Japan or Emerging Markets, rarely Europe and almost never specifically at the Eurozone. This may be changing.

Exhibit 1 comprises the standard charts we include in every asset allocation report. The one on the left shows the recommended weight of the Eurozone relative to its global equity benchmark and the one on the right shows the lead indicator, which is our estimate of how the left-hand chart will move over the next four weeks. Both charts have suddenly gone vertical, with the improvement dating from the last week in June, when the left-hand chart broke up through its moving average. The lead indicator shows that the bullish signal gets progressively stronger as we shorten the time period under consideration – in other words the pace of change is accelerating.

When we make calls like this, we are often asked whether they are justified by the underlying data. In this instance, we could reply by mentioning the new EU budget deal, the creation of a new risk-free asset for the Eurozone, Europe’s better recovery from Covid when compared with the US and the valuation gap with the US. But that’s not really the point. The point is that sentiment towards the Eurozone has improved significantly and our numbers suggest that this is likely to continue. Positive sentiment is a good thing, in and of itself, and should not be wilfully ignored.

It also helps if the macro and micro stories are aligned. When the Eurozone rally first started, we noted that there was not much evidence of a shift away from defensive sectors towards cyclical sectors like Industrials or Materials, or deep value sectors like Financials. This is beginning to change. We have just upgraded Materials to overweight, having had it as an underweight or neutral for the last two years. We could not get excited about it in June, because its lead indicator was not attractive, but this is now improving. We upgraded Industrials to neutral two weeks ago and it now has a set of charts, compared to the rest of the Eurozone, which look very like the Eurozone compared to the rest of the world.

But the key sector for the rehabilitation theme is Financials, which has always been the Achilles heel of previous rallies. We still have it as an underweight, but the lead indicator is now as good as that of Industrials and of the Eurozone itself. At the current rate of progress, the left-hand chart will break up through its moving average in early August, at about the same level as it did in October 2019. That rally ended with the onset of Covid in Europe, so there is a sense in which this rally is just the resumption of a pre-existing trend.

Clearly the world has changed since then, but if the Eurozone/EU really has embraced fiscal activism and abandoned austerity, the outlook for Eurozone banks may be better than the loan loss projections imply. We would argue that they are already included in the price as a result of the massive underperformance between March and May. A Eurozone rally without the participation of Financials will be anaemic at best. But if the region and the sector rally together, the result could be quite powerful.

Leave a Reply

Friday, July 10th, 2020

Rotation in the US

The period before the start of the Q2 results season in the US is often uneventful. Trading volumes have been higher than normal this year, but there have been no changes to our US sector ranking for two weeks and very few for the last two months. So, it may seem odd to argue that we are on the verge of a significant sector rotation, but that’s what we think.

There are three main reasons for this, all of which rely on data generated by our US equity sector model. First, the average weekly change in each sector’s weighting has fallen to a two-year low. The current rate of change is about half of what it was in June 2018 and 2019. This is unusually slow, even for a seasonally slow period. Second, the composite lead indicator for the whole index, is close to a 12-month high, having been well below its normal level from September 2019 to June 2020. This measures the amount by which the recommended weighting for each sector is expected to change (up or down) over the next four weeks. Thirdly, the level of conviction attached to this lead indicator has risen to a two-year high. This is measured by the goodness of fit (R-squared) between the trend-line (the lead indicator) and the underlying data.

In summary, the level of activity recommended by our model has been unusually low; the lead indicators predict significant change in the recommended sector weights and attach a high level of probability / conviction to this view.

For each sector there are two relevant pieces of information, the gradient of the trendline and the level of conviction. All sectors meet our definition of high conviction. The sectors with the most conviction are Materials, Consumer Discretionary and Communications, and those with the least are Healthcare, Financials and Industrials. However, all the conviction in the world counts for nothing, if the lead indicator is pointing sideways, rather than up or down. The sectors with the most powerful lead indicators are Energy and Materials (both positive) followed by Utilities and Consumer Staples (both negative). The sectors with the most sideways lead indicators are Technology and Consumer Discretionary.

The strongest combination of lead indicator and conviction level belongs to Energy. We have a very large underweight in the sector, but we expect to reduce this significantly. The next best combination belongs to Materials, where we expect the neutral position to be upgraded to overweight. The third and fourth positions belong to Utilities and Staples, where we expect to reduce our exposure. Utilities has already been downgraded to underweight and we expect Staples to follow soon. On the basis of these numbers, we do not expect much change to Financials or Industrials (both underweight) or to Technology and Consumer Discretionary (both overweight). The outlook for Healthcare and Communications is unclear, but Healthcare’s lead indicator has deteriorated badly in recent weeks.

Some may argue that it is premature to talk about sector rotation if we are not forecasting a change of leadership. We agree, but with two reservations. First, a change in laggards provides the same opportunity for generating alpha as a change in leaders. Second, these are just the signals we can identify now. Others may develop as we move into the results season. Rotations often start small and then get bigger as investors realise that their current positions are growing stale and that there is money to be made elsewhere.

Leave a Reply

Friday, June 26th, 2020

Buying Dips & Selling Bounces

There ought to be a special place in hell for the broking analyst who first came up with the statistic that, if you had missed out on the n best weeks of stock market performance during the last n years, your return would only have been x instead of y. Like all really damaging statistics, the problem is that it is true. However, it is only one side of the story, is easily misinterpreted, and doesn’t actually tell you what to do. We thought it would be a useful exercise to review all of the evidence to see what works, and what doesn’t, so that if there is a second wave of the pandemic and a second sell-off, we can have a slightly better-informed reaction function.

We have looked at the data for over forty countries, since the beginning of 1995, based on the weekly total returns in US dollars. We summarise on the basis of the ten best and worst weeks during this period, but if anyone wants the data on another basis, we are more than happy to share it. To qualify for entry in the US, the weekly return needs to be slightly better than 7%, or slightly worse than -7%. For other countries, the median 10th best week is 10.7% and the median 10th worst week is -11.2%. There is a table which shows the detail for each country. For discussion purposes, we concentrate on the 15 most important markets, including the US, and the aggregate for the world ex US.

If you had missed out on the 10 best weeks in the last 25 years, your annual total return for the world ex US would have been 1.7%. instead of 5.4%. It’s the sort of number that ends discussion immediately, but it shouldn’t. If you had missed out on the 10 worst weeks, the equivalent number would have been 10.7%. The uplift from avoiding the bad weeks is far bigger than the penalty for missing out on the good ones. This is also true on a net basis. If you missed out on both best and worst, your return for the world ex US would have been 6.8%, instead of 5.4%, an uplift of 1.4%. We reach the same conclusion, but with different numbers, for each of the 15 major markets we look at – apart from China, which wasn’t a major market until sometime after the 2008 crisis.

Most professional investors should be aware that global equity market returns, in aggregate and in individual countries, are dominated by tail events and that these are negatively skewed. Of course, missing out on good periods is bad, but avoiding bad periods is better. Unfortunately, none of this helps with the really important question: what to do if you have just been through one of the 10 worst weeks in the last 25 years.

We looked at what would have happened if an investor had sold at the end of one of the 10 bad weeks and bought back in one week later. Conventional wisdom is right; it’s normally a mistake to sell at this time, but the result is far more nuanced than one might expect. First of all, it’s not always wrong. For the 15 major markets, it’s only a mistake in 62% of our observations. The best places to apply the rule (of not selling) are the UK, Australia, Canada, Japan and Switzerland, but the rule is wrong more often than not, in France, the Netherlands and India. The biggest penalties for selling out come in Korea, the UK and China, equivalent to an annualised return of about 2%. However, for the US, Hong Kong and Brazil, the penalty is less than 20bps per year, which is easily justified on the basis of risk efficiency.

Second, for the world ex US, investors would have realised a small uplift, equivalent to 21 bps per year, if they had sold immediately after one of the 10 worst weeks. The difference is down to different sample periods; a really bad week in one country is not necessarily a really bad week for the world and vice versa. The result is not significant, certainly not good enough to be the basis for a successful shorting strategy, but it directly contradicts the buy the dip mentality. On a global basis, one very bad week really is often followed by another bad week.

This result is so interesting that we extended the experiment to see what happened two weeks after each of the 10 worst weeks. What happens if we sell at the end of the first week after the crash and buy back at the end of the second? The answer is much clearer. For the world ex US, the cumulative uplift in returns is a massive 1.7% a year. The strategy works 6 times out of 10, and there are two occasions when the penalty for selling is less than 2%.

The strategy works in aggregate in 14 out of the 15 major markets. In some cases, like Germany, Hong Kong and the US, the result is marginal and it doesn’t work in Japan. But for others, like Korea, the UK, South Africa and Australia, the uplift is greater than for the world ex US. The median uplift for the major markets is equivalent to 1.0% per year, and the strategy works for 59% of our 150 observations. Waiting a week and then selling the bounce is a successful strategy.

If we look at the two weeks after a sell-off on a combined basis, we find that investors get an uplift in returns if they sell at the end of the crash week and don’t buy back in until the beginning of the third week. For the world ex US, the annualised uplift is equivalent to 1.9%.

In 9 out of 15 major markets there is an annual uplift, for which the median is 40bps p.a. The strategy does not in the US, but it works surprisingly well in the UK and Australia, and the only really bad result is in China (where the penalty for missing out is equivalent to 2.0% per year).

Our conclusions are follows.

  1. Avoiding the really bad weeks is more important than participating in the really good weeks – always and everywhere.
  2. Really bad weeks are not automatically followed by really good weeks. Sometimes they are, but in aggregate the result is only just positive and cannot be relied on as a trading strategy.
  3. The second week after a really bad week is normally negative. Avoiding this week by selling and buying back later is a successful strategy in aggregate and works for most individual countries.
  4. Avoiding the two weeks after a sell-off is normally better than buying the dip. The only country where FOMO is justified is China. In the other major markets, the penalty is either non-existent or is very small and can be justified on the basis of lower volatility / risk efficiency.
  5. If you buy the dip in week one, don’t forget to sell the bounce in week two.

Leave a Reply

    Full Blog Archive:

  • Search Blogs by Category