US vs Canada vs Europe

General
1

Here's a question I had this week. If you divide your universe of stocks into four groups--US listed stocks, Canadian stocks, European stocks, and OTC stocks--which groups behave alike and which behave differently? My guess was that European stocks would behave differently and that the other three would have a lot in common.

So I created ten different ranking systems, each with five randomly chosen factors. I ran them on the four different universes. I noted the lowest bucket return, the top bucket return, and the slope for each ranking system and each universe, giving me 120 different data points.

I then looked at the rank correlation of the ranking system performances by those three measures for each of the four universes. In other words, did the ranking systems that did the best according to the top bucket in Europe also do well in the US? Did the top three ranking systems in one region end up in the bottom three in another?

It was clear from this experiment that Canada and Europe are similar to each other and very different from the US, while OTC stocks behave much like other US stocks. The average rank correlations on my three measures were:

EUR and CAN: 0.64
EUR and USA: 0.33
CAN and USA: 0.37
OTC and EUR: 0.01
OTC and CAN: 0.20
OTC and USA: 0.48

So US stocks do perform quite differently from non-US stocks.

In my experiment, the top bucket (top 10% of stocks) averaged 7.53% greater than the universe average in Canada, 6.97% in Europe, 18.7% in the OTC markets, and 5.1% for listed stocks in the USA. Slopes and bottom buckets had similar numbers. US stocks were, overall, less sensitive to ranking than any of the other three groups.

Spurred by this, I decided to look at my actual performance. In my personal accounts, if you add up all the realized trades I've made in non-US stocks since 2019, I've made 8.05%. Ditto in US stocks, 4.98%. (This is derived from summing up the proceeds from all the trades and dividing by the costs, so it weights more recent trades, which are larger, more heavily.) The average non-US trade delivered 6.64% while the average US trade delivered 5.63%. (The first two numbers are money-weighted; the second two are just plain averages, so include very small trades.) The numbers from my hedge fund, Fieldsong Investments, show the same thing: my non-US stock picks have been outperforming my US stock picks.

Two very different explanations can be offered here. One is that the formulas we use in ranking systems have all been arbitraged away in the US, while they have not in Europe and Canada. The second is that investors in the US behave illogically and ignore value and quality, while investors in Europe and Canada have more sense and reward sensible factors. According to the first explanation, changes in stock prices in the US are harder to predict because they already reflect all available information; according to the second, it's because they don't make any sense.

5 Likes
2

I looked at it before and concluded that it by large can be explained by size factor.

Below is the Market Cap for Easy Trade USA and Europe. Canada Primary with same filter as Easy Trade Europe.

MarketCap Median Average
USA 1,862.14 3,188.89
Europe 606.44 1,164.94
Canada 38.39 114.86

That said, in my strategies Canada and Europe outperforms USA as well.

1 Like
3

Size relates to outperformance, sure. That's one reason I included OTC stocks in this study: because they're extremely tiny. The fact that OTC stocks correlate far better with US stocks than with European or Canadian stocks says to me that different factors work in the US than in Europe and Canada, and you can't explain that only by size.

1 Like
4

I think these are important topics and I have been thinking about it over the last few years

The US has a greater proportion of volume traded by HFT strategies for example (although Canada is high too) and Hedge Funds specially much more than Europe with HFT percentage of volume being almost twice as large by some estimates.

Not saying this is the answer but rather an example of the answer. For me the answer would be that the strategies dominating each region are different. The investor ecosystem and their chosen strategies is different across the regions. Perhaps emerging markets will show their own pattern too if you perform the same test likely closer to Europe except at the distressed end of valuations due to economic collapses and fraud

Additionally I would add that a different behavior to me does not necessarily mean the prices are for sure efficient in the US but that there is more competition in the same strategy type which depending on the strategy can actually make prices worse. Both market efficiency and inefficiency can make things challenging for fundamental investors -depending on the context of course. In other words, there is no reason at least for me that both answers you propose can’t be true at the same time depending on which trading/investment strategies are having the largest market impact

For example, hypothetically, crowded momentum trading and passive flows can make other approaches relatively worse but result in higher valuations and dead short funds resulting in greater success for the approach but not necessarily efficiently deployed capital in terms of actually allocating the nation’s funds for greater realized growth or calculating intrinsic value. Keynes wrote about this too as he noticed the US vs UK differences early on too

2 Likes
5

Thanks for sharing, Yuval. I really enjoy this kind of analysis.

I’ve been thinking a lot about regional exposure and the pros and cons of trade friction as it relates to broader global exposure.

I’m also interested in better understanding the trade-offs between running multiple geographic-specific models versus aggregating regions into a global universe and letting the rankings and rules decide.

1 Like
6

From Keynes’ 1936 General Theory of Employment, Interest and Money:

“If I may be allowed to appropriate the term speculation for the activity of forecasting the psychology of the market, and the term enterprise for the activity of forecasting the prospective yield of assets over their whole life, it is by no means always the case that speculation predominates over enterprise. As the organisation of investment markets improves, the risk of the predominance of speculation does, however, increase.

In one of the greatest investment markets in the world, namely, New York, the influence of speculation (in the above sense) is enormous. Even outside the field of finance, Americans are apt to be unduly interested in discovering what average opinion believes average opinion to be; and this national weakness finds its nemesis in the stock market.

It is rare, one is told, for an American to invest, as many Englishmen still do, “for income”; and he will not readily purchase an investment except in the hope of capital appreciation. This is only another way of saying that, when he purchases an investment, the American is attaching his hopes, not so much to its prospective yield, as to a favourable change in the conventional basis of valuation, i.e. that he is, in the above sense, a speculator.

Speculators may do no harm as bubbles on a steady stream of enterprise. But the position is serious when enterprise becomes the bubble on a whirlpool of speculation. When the capital development of a country becomes a by-product of the activities of a casino, the job is likely to be ill-done”.

-John Maynard Keynes, The General Theory of Employment, Interest and Money (1936)

IMG_1930
IMG_19301071Ă—947 65 KB

1 Like
7

Me too! I've been doing the latter the entire time I've been using Portfolio123. And when I did develop a Europe-only ranking system, its performance in Europe after a year turned out to be worse than my all-countries ranking system. So I am approaching this with some trepidation.

However, there's another trade-off that may make a large difference. I believe, though I'm not certain, that the best ranking system for the US will favor very tiny companies, and the best ranking system for Europe/Canada will be more lenient about company size. If so, that will pay off big-time in transaction-cost savings.

But there are other factors thrat weigh in favor of aggregating regions into a global universe. An especially strong one is that FactSet's coverage of Canadian and European companies pre-2008 is spotty, and I don't trust any backtests run on that data as it might suffer from survivorship bias. So if you want to take the years 1999 to 2008 into account, you just can't do it with a Europe-and-Canada-only backtest. Your best solution is to do a global universe backtest with a rule like Eval(AsOfDate < 20070701, ExchCountry("USA"), 1).

Lastly, for me personally, I want to try using two ranking systems, one US and one non-US, in my hedge fund because I'm going to be treating the US and the non-US portfolios differently in one important respect. Most of my non-US holdings are not marginable, while most of my US holdings are. So if I base my backtesting for non-US stocks on a simulation that doesn't use margin and if I base my backtesting for US stocks on a simulation that uses a good deal of margin, I may get results that are closer to what I'm actually doing in the fund.

But it is a very fraught decision, and your comment is forcing me to think hard about it before pulling the trigger. Thanks.

3 Likes
8

Yes, there are so many dynamics and variables — even more so for you running a hedge fund.

I’m hesitant to “hold all regions” in the same way I wouldn’t want to “hold all stocks.” Broad consideration makes sense to me.

Using momentum to determine which region-specific models are held at a given time feels like a reasonable alternative — though in practice I’m presently constrained by what’s available in the Marketplace.

1 Like
9

This makes sense to me as well, especially given that Europe and Canada share the same accounting framework. Europe reports under IFRS, and Canada has used IFRS since 2011, whereas the US reports under US GAAP. These differences materially affect the construction and interpretation of fundamental factors — for example revenue recognition, accrual discretion, treatment of exceptional items, and balance-sheet recognition of assets and liabilities.

1 Like
10

+1. I’d isolate price and fundamental signals and review the results before arriving at possible explanations.

11

I am running a US-only, a ExUs-only and a “global” startegy in equal proportion. That way I always have a fixed 1/3 allocation to both US and ExUS and the global strat can overweight regions opportunistically.

My main 2 cents:

  1. Country momentum, Country “Value”, FX (geo)political shifts etc. are in my experience far more difficult to predict than e.g. cross-sectional industry leadership within a region. An ExUS run like last year could happen this year in US smallcap or not, who knows. But if your startegy abandons a region entirely, you will miss exposure to such shifts entirely.

  2. universe composition has a lot of impact on the final cross-sectional multifactor ranking. That means that the top composition of the US ranking as satandalone does not have to be identical to the top US stocks in the global universe ranking. This way it is really hard to say “which approach is better” most of the time. In my backtests, the global strategy outperforms the regional standalones but not by a huge degree. But a book of the three startegies does way better on a riskadjusted basis

image
image802Ă—445 19.4 KB

PS: Looking forward to soon add Asian/Pacific to the mix.

4 Likes