2H 2020’s Most Important Datasets

By in
2H 2020’s Most Important Datasets

Earlier today we spoke on a panel hosted by Wall Street Horizon about the use of data in the investment process. The other speakers were both from the buyside and companies that provide data. It’s always impolite to ask someone’s age, but it’s safe to say there was at least 100 years collective investment experience among the 5 panelists, so this was a great discussion. There is a link at the end with the other speakers’ bios.

What really caught our attention was the last round of the conversation, which focused on what sorts of investment data will be the most useful over the next 6 months. Three points on what we took away from the group’s comments:

#1: Retail investor sentiment and activity in listed equity and options markets. The surge in small-money retail trading over the last year is truly new to the current market structure, which means it is having a novel effect on asset prices.

The last time we had this much growth in retail trading was in the late 1990s, but the way stocks trade now is very different. The SEC adopted Reg ATS (Alternative Trading Systems) in 1999, which led to an atomization of where and how stocks trade over the following decade. NYSE-listed stocks now mostly trade away from the floor, and the same goes for where NASDAQ-listed stocks trade. Keeping +50 ATS platforms and other trading venues in sync, tick by tick, is a technological challenge. Introduce a new investor constituency – retail traders – with novel approaches to allocating capital, and the system will go through a period of adjustment.

We went through the numbers last night, but the upshot is there’s something like 5-10 million new retail stock traders now versus 2019. Their sources of information are different from institutional investors (Reddit, etc), their approach to valuation is different (more momentum based), and as volatility has declined they’ve learned how to trade options. Again, none of this is “new”; listed options were a retail traders’ game when this product launched in the 1970s. But it is new to the current market structure.

Takeaway: markets are still coming to grips with the recent influx of retail investors, and hearing that there is still a data/information gap regarding their activities tells us that their influence will remain over the balance of the year.

#2: The pandemic has forced investors to think more thematically, and the increasing usage of data in the investment process has sped up the market’s ability to quickly discount these broad trends. One panelist mentioned credit card transaction data as his “most important thing to watch” in the back half of the year. Mastercard, for example, sells real-time anonymized and aggregated data on where cardholders are spending money. NPD sells point of sales data from +600,000 retail locations. This information doesn’t come cheap, but hedge funds and other asset managers are big users.

Takeaway: it is hard to overstate just how much more information is embedded in individual stock prices now than even just a decade ago. There are two reasons for that. The first is that the increasing digitization of everyday life leaves a data trail which is then sold to investors. The second is that current market structure, as outlined above, is highly automated and therefore pre-wired to receive large amounts of fundamental data inputs. It’s only a small-ish leap from coding an ATS to take data from other trading venues about price and bid-offer size to including company-specific data.

#3: Our answer to the “what’s important for 2H 2021” question was the two themes we come back to time and again: how the US consumer responds to inflation, and how urban unemployment develops.

Our approach to measure the former is to look at the volume of US Google searches for terms like “cheap” and “discount”. An uptick would signal rising concerns about inflation that could lead to declining consumer demand. As the following chart of search volumes since January 2019 shows, this has not yet started to happen. Volumes for each term remain stable, which tells us stagflation is not yet even close to being an issue for the US economy.

The US urban unemployment issue is important because it is large enough to affect the national joblessness data. As we reviewed earlier this week, New York, Los Angeles, and Chicago represent 13 percent of national unemployment but only 5 percent of the American population. Most of the issues keeping the jobless rate high in major US cities have nothing to do with interest rates. Rather, they are pandemic related (continued work from home, tourism, etc.). If the Federal Reserve only looks at national unemployment and sets policy based on those numbers, they will miss that fact.

While not the sort of tech-enabled “Big Data” that we discussed above, NYC subway ridership statistics tell the urban unemployment story as well as any of them. Here are the current numbers and their comparisons to 2019. Weekday volumes are still less than half pre-pandemic levels and have shown no improvement from the last several months.

Takeaway: our approach to data is to use freely available resources to evaluate market narratives. Because of the economic dislocations over the last 18 months, headline numbers – and stories they seem to tell – are often misleading or just wrong. Because of the wealth of no-cost information available now, it is possible to find the real story. You just have to know where to look.

Sources:

Panel participants: https://info.wallstreethorizon.com/dataminds