At Inukshuk Capital Management, our focus is on building lasting relationships rooted in trust and collaboration. By focusing on active risk management and long-term value creation, we help families and institutions achieve financial sustainability.
Stay up-to-date on the latest developments by following us on LinkedIn here.
In this issue:
- Global Equity Market Performance
- The Ten Percenters
- Clusters
- Context
- May So Far
- Wrapping Up
Global Equity Market Performance
All four equity market indexes we follow shrugged off a disappointing March and rallied impressively. The S&P 500 was up 10%, just edging out MSCI Emerging Markets by five basis points. The S&P/TSX 60 gained 4.3%, lagging the S&P 500 by just under 6%. It closed the month ahead on the year versus the S&P 500 by around 3%—although it has given back most of that so far in May. After a weak March, MSCI EAFE continued to lag, up 2.6%.
As of this writing—the second week of May—our systems are indicating positive momentum in all four equity markets although there are still some signs of caution in the sector signals.
If you would like to stay current on our measures of trend and momentum in the markets we follow, please click here.
A one-month ten percent rally by the largest equity market in the world is not something that happens very often. This month we take a look at the history of ten percent rallies.
The Ten Percenters
A +10% performance in any month for the S&P 500 has happened only 32 times. Interestingly, the opposite—a -10% move has happened 31 times. Going back to 1927, there have been 1,181 months of S&P 500 price data. So that translates to a 2.7% and 2.6% chance, respectively, of any month with a change greater than 10%.
Note: We are using Macrotrends data, based on Robert Shiller’s dataset. Prior to March 1957, this reflects the S&P 90 Index (the predecessor to the modern S&P 500). The series are combined for continuity.
Let’s look at the distribution of monthly returns over this timeframe.
Most of the monthly changes are centred around the average, a gain of 0.65%. The distribution resembles a “bell curve”, or normal distribution. Or if you want to be technical, it follows a Gaussian distribution—though not precisely. Modelling markets on the assumption that their returns match a perfect Gaussian has played a role in some of the more epic financial market meltdowns in the past. That’s the topic of several books.
Seventy-four percent of all months fall between -5% and +5%, with 29% down versus 45% up. Months worse than -5% and months better than +5% represent 11% and 15% of observations, respectively. Regardless of the size of the move, across all observations there are more that are positive than negative. There’s a 60% chance any month will have a positive return.
Both findings might be expected, given the index has risen from its 17.66 close in December 1927 to 7,209.01, as of April 30 this year. A bit of a surprise is that the largest change was a gain of 42.2% in April 1933 versus the largest loss of 29.9% in September 1931.
Notice the dates are very close. Let’s look at these returns over time.
Clusters
It’s clear that extreme moves over short periods of time cluster together, whether up or down.
You can see that in the chart below.
Some highlights and lowlights:
Between 1927 and 1940 the market rallied +10% in one month, 18 times. That’s 56% of all instances.
The market was up 10.4% in June 1929, it fell 19.9% in October and 13.4% in November. The end of the Roaring Twenties kicked off the start of a very volatile Thirties.
On September 1, 1939, Germany invaded Poland, effectively starting WWII. The market rallied 16.5% that month. That’s after a 10.9% rally in July. Which followed a 13.5% setback in March.
Surprisingly, the market was relatively calm from the 1940s to 1970s.
Mister Market doesn’t always seem to get too upset about wars.
In October 1974 the market was up 16.3%, after an 11.9% fall in September. That was the start of the next cluster of volatility that lasted for roughly ten years and included seven one-month +10% rallies.
1987 was interesting. It started with a 13.2% rally in January. Nine months later in October, it crashed 21.8%.
The most recent +10% gain before April was November 2020, up 10.8%. That was after the pandemic panic of March and April which saw the S&P 500 fall 12.5% then rally 12.7%.
Do these +10% months tell us anything other than they are associated with general volatility?
Context
An idea to test whether any one +10% month could be predicted or be an indicator of future performance is to put it in the context of what was happening prior to and after the event.
Before
First, we calculated the three, six and twelve-month returns of the index before each +10% month. We then compared those averages to the returns across all such periods in the full dataset.
On average, the returns were +2.0%, +4.0% and +8.1% respectively. However, prior to +10% months all three averages were negative—underperforming all-period returns by 4.6% to 10.5%.
Is there a useful signal when all three before periods are negative that predicts +10% moves?
Averages have limitations, so we counted.
There were 211 instances when all three were negative, or an 18% chance. In the following year, the index produced at least one +10% month 125 times, or 59% of the time. In other words, if at any time three negatives are observed there’s a better than even chance that over the next twelve months a +10% will occur, and maybe more than one.
The problem: there’s a 66% chance of at least one -10% month over that same period.
The percentages add up to more than 100% (59% + 66%) because these events are not mutually exclusive. Also, the data shows that the three negatives can persist over consecutive months, so the forward-looking windows can overlap. And, during high-volatility periods, the same 12-month forward window can contain multiple +10% and −10% monthly moves.
The chart above that highlighted +/-10% months shows this.
After
Everyone wants to be able to predict the future, so we looked to see what happened in the months after a +10% month.
After a +10% month, all three average returns were positive, but they underperformed all-period returns by 1.0% to 4.8%.
There were 546 instances when all three periods were positive across all the data, or a 46% chance. Over the following year, the index was +10% in a month 126 times, or 23%. That’s not a very reliable signal, considering there were 40 -10% months over the same timeframe.
Now
At the end of March, the three, six and twelve-month returns were -4.6%, -2.4%, and +16.3% respectively. April then delivered a +10% gain. This mix does not fit neatly into either an all-negative or all-positive prior period.
We only have prior return data for 30 of the 32 instances of +10% months since the first two occurred too soon after the 1927 start of the dataset. Of those 30, 14 (46.67%) were preceded by all-negative returns, five (16.67%) by all-positive returns and eleven (36.67%) were neither.
We can’t draw many conclusions from this either. It only suggests a somewhat higher likelihood of a +10% month following periods that were uniformly positive or negative.
Since only the 12-month prior return is currently positive, let’s look at that period only. Of the 30 historical cases, 14 were preceded by a positive 12-month return and 16 by a negative return — essentially a coin toss.
Finally, we looked at how the market performs in the year after a +10% month. We know from the ‘after’ analysis above, returns tend to underperform on average. If we count, of the 29 instances (we lose one because the last one is now) 15 were up and 14 down. No real insight there.
The ‘before’ analysis indicates that observing a month where all three periods are negative shows the next twelve months could be volatile. The ‘after’ analysis seems to show that forward returns have historically been positive but underperform. The ‘now’ analysis does not add anything further.
May So Far
On May 7, Barron’s reported that xAI was being dissolved and combined with SpaceX. The speculation is this is in preparation for a SpaceXAI IPO. Subsequent reporting suggested this could be the largest IPO in history, raising $75 billion and valuing the company at more than $2 trillion.
On May 8, the Wall Street Journal reported that the University of Michigan consumer-sentiment index fell to a record low of 48.2 in May compared to 49.8 in April.
On May 12, CNBC reported that U.K. borrowing costs are at their highest level since 2008.
As of May 15, the S&P 500 is up 2.8% on the month while the S&P/TSX 60 is down 0.8%.
Wrapping Up
There’s no magic formula to predict the future. It’s not possible. We spend our time studying historical data to build models for managing risk. We do not use specific percentage return thresholds as a mechanism to build robust systems. They are historically unreliable.
We don’t use averages either. But in presenting a lot of data in order to put it into context they can be useful and hint at what to look for next.
An arbitrary cutoff line of 10% either way gets headlines, but in developing systems they’re not useful.
Wall Street calls a 10% pullback a “correction”. Which is funny when you think about it. As in: Was the price 10% ago incorrect?
A 20% drop is labelled a “bear market”. Big round numbers are what people can relate to and remember.
It’s hard to avoid these things. And it’s interesting to look into the numbers to see if they mean anything.
In this case, ten-percenters are rare. As far as we can tell, the most important takeaways are: They cluster with volatility and below-average returns.
There hasn’t been much of that recently in the broader market if you constrain your analysis by +/-10% moves.
We know trends in price exist. They appear to persist in volatility as well.
Stay tuned.
Have a question? Contact us here
Challenging the status quo of the Canadian investment industry.
