Warning: this post does not discuss AI

I have wanted to write a post on Alternative Data for a long time. There are a number of reasons for that – across business strategy, content strategy and current market catalysts.

1. The Alt. data industry has never lived up to its enormous promise. On paper an enormous TAM. TAM of course is what you tell the PE investor. The SOM is for your actual sales plan. Today’s SOM remains fundamentally limited.
2. Alt Data has never really been a part of the mainstream market or financial data world. Largely separate vendors, events ecosystem and surprisingly low (fundamental) investor adoption rates. It is central to systematic investors and the spend in that world is significant, as well as for many directional HF strategies.
3. I love any content set with alpha implications and Alt. Data, with all its diversity, offers an enormous array of alpha opportunities.
4. It sits at the intersection of multiple topics that I enjoy – product and content strategy, buyside investment processes, commercial models and technology.

The path to product and industry maturity

A brief history of Alternative Data

Alternative data is no longer new. Depending how you want to define it, early versions were in adoption by the Financial Crisis in 2008. As text sources proliferated online and social media adoption exploded, we started seeing new content types – and the first text analytics at scale. Newer / early(ish) forms of data science and machine learning allowed for the first waves of analytics on text.

At the same time, many new types of sensors began to have their data collected. We had the Internet of Things (IoT), mobile phones, cars and seaborne traffic all giving geolocation data, satellite imagery as well as all manner of internet traffic data.

Today, the cool AI Alt. data graphic gives you some sense of major categories (though isn’t exhaustive and I gave up on further prompting). Yet despite the maturation of Alt. Data it has never hit the broad adoption that previous generations of content achieved (macro data, estimates, fundamentals, ownership, filings, transcripts, ratings, default probabilities, you know the list.)

Why the difference? I suggest there are a number of factors at play

High signal / low breadth | commercials

Estimates data is available on over 20,000 listed equities. That provides a solid starting point for understanding future performance, growth rates, valuation ratios and so on. Estimates also give you a view on margins and operating metrics. They’re incredibly useful and thus easy to value. Relevant to every industry and offered by a number of large vendors.

Now, let’s look at alt data. What if you are offered a really powerful signal that gives you insight into 60 stocks only? What is that worth? Obviously it can help you with your views on that cohort of stocks, but that’s a lot of data integration and analysis to undertake for a small part of your investable universe. At what point is the juice not worth the squeeze? Many alt data sets offer precisely this trade – high signal value, low breadth.

Jordan Hauer, an industry veteran and CEO of Amass Insights, reminded me of a 3rd category – low breadth / low signal. Of course, those firms will meet a Darwinian fate at some point, natural selection still exists in markets, VC firms just like to delay it.

Vendor proliferation

Amass Insights‘ carry over 24,000 companies in their Alt. Data database. That number is mind blowing to me and gives you an immediate sense for the challenges of the industry. Aggregation platforms are clearly required and the traditional industry, perhaps aside from Open:Factset, have made only the most cursory of efforts at data integration and onboarding.

Not even a Citibank or JPM has that number of vendor relationships. At least in a pre-AI world, that number doesn’t work for distribution. We will look at some of the models in use today that try to aggregate for customers and manage this complexity.

Workflow and desktop fragmentation

For sophisticated systematic or quantitative investors, they can bear the pain of integrating, wrangling, cleaning, stitching, munging (the number of verbs in this space is amusing) if the data is valuable / differentiated. But spare a thought for an equity analyst. If the data isn’t in Bloomberg or Factset, or isn’t nicely connected up to Claude, chances of adoption become slim. Their day job just gives limited bandwidth for familiarising themselves with, and adopting, some of these 24,000 alternative data providers.

There are desktop platforms offering Alt Data, such as Maiden Century or Exabel, but adoption here is largely a hedge fund phenomenon. The asset management industry has limited budgets and, like Banking and the Advised wealth space, there’s a limit to desktop real estate and an aversion to the “Toggle Tax” on their busy users.

A plethora of Alternative Data solutions for Quant

Multiple aggregation business models exist…

While all these firms play a valuable role in consolidating disparate sources and bringing them to the Buyside’s attention, their relatively small size is noteworthy. Neudata is large and successful in Alt Data terms but 95 associated members on LinkedIn (a reasonable proxy) pales next to 26,000 staff at Bloomberg or 42,000 at S&P.

A successful industry and business model has been built for the top systematic players certainly, yet the ceiling there must start to feel quite low.

…little sign of headway for fundamental adoption

Alternative data is certainly available in the scale decision support tools – the Terminal, Workstation, Workspace, etc., but in limited amounts. Shipping and satellite imagery, some basics. By and large, one must assume the desktop product teams either haven’t seen the customer demand, or the price uplift for meeting that demand was unappealing.

Thinking through the demand lens, the end user analyst, banker, PM and so on must be comfortable with either not using the data or accessing it via other channels. It certainly speaks to low overall adoption levels. When I think of the new generation of desktop tools – Rogo, Finster or Boosted AI, appetite for Alt Data content appears lukewarm as well.

The 3rd Energy Crisis: perfect timing for a reset

Another oil shock

So what’s going on in the market.

We have a major exogenous shock. In the first instance an inflationary supply shock. Over time a demand shock as the global consumer’s budget is reallocated to pay for their higher energy costs.

Perhaps less obviously, we have a major macro event that challenges investors, even perhaps a paradigm shift. Traditional company estimates and macro forecasts have less predictive power when there is a major regime change (referring to economics here, not Israeli decapitation strikes) such as from low inflation to high inflation, or bull market to bear.

Firms need to model how the consumer is reacting and what that means for their spending patterns. Are they delaying their holiday? Eating out less? Downgrading from Whole Foods (M&S) to Walmart (Tesco)? These changes impact consumer stocks, restaurants, beverages, travel and leisure directly, while energy is an input to almost every manufacturing process, not to mention impacting costs and supply chains for the semiconductor industry.

The traditional data sets don’t answer these questions, with many (fundamentals, management guidance, etc.) all having a backwards tilt as they tend to extrapolate historic data.

A solution whose time has come

Into this data analysis gap, step forward the Alt Data industry. In fact, two of the area where it the industry brings obvious value are Commodities and Consumer. For Commodities, you’ve no doubt seen all the shipping tracking and imagery, even on presidential social media.

On top of this – the industry offers an almost limiteless array of tools to watch the consumer’s behaviour adjust in real time. We have:

• Credit Card | Spending data whether from POS, receipts
• Geolocation data (are they, or at least their phone, at the mall?)
• Satellite imagery (how full is the mall carpark?).
• Web traffic | searches (are Walmart, Dollar Stores, etc. seeing more traffic? Are people searching for credit card hardship?)
• Social media | sentiment (measuring consumer confidence and frustration)

So institutional investors incorporating alternative data can maintain high levels of confidence in both macro assumptions and individual company forecasts. This is, of course, critical in identifying mispricing in securities and incorrect market assumptions.

Another way to frame this is “Is your investment process robust for dealing with macro shocks?”

The time is now

Sadly, I’ve seen very little focus from the industry to capture this opportunity. They should be making hay!

• Marketing campaigns on how their data can feed into a real time consumer behaviour or sentiment assessment
• Research on the predictive power of alt. data. Is geolocation data more powerful than sentiment? Does a blend work here?
• For direct customers – offering trials to traditional firms reminding them of the need to stay abreast of a fast evolving market pivot. I hope your teams are asking thought provoking questions like “How are you tracking this real time change in consumer behaviour?”
• For channel partnerships – engaging with vendors, cloud providers and others to broaden the options for Buyside consumption – meeting these firms where they already perform workflow or acquire data. As I mentioned in my 21st century desktop strategy piece, content differentiation is something you want to pitch to these incumbent and disruptor vendors.

Driving Alternative Data adoption.

A final point on the industry is the preferred distribution model has always been to provide raw data to the systematic investor. The guys with calculators in their pockets. Picture the guys at the computers in Houston in Apollo 13.

These firms have deep pockets, they have data integration expertise. They bear the pain of your MVP because some of the data is incredibly valuable. Yet this “here’s an API, good luck with it” really limits industry adoption.

If I wanted to expand my SOM to anything like the TAM claims so often seen, you need to reduce the structural friction to adoption.

Make. It. Easy! As Jerry Maguire said “Help me to help you!”

There are a few options:

---

The hard thing about hard things.

If it was easy to drive broader adoption from the Buyside, it would have already happened. I am not suggesting some obvious path to success was ignored. I do think, however, many firms have opted for a narrow commercial approach that limits their long term viability (and means they’re all chasing the same ICP).

The Quant community is an easy target and an obvious one, your MVP is easier and faster to achieve because they will take your raw, half baked content and do the heavy lifting. If you get their attention.

Fundamental adoption is harder, however it drives such a large TAM expansion. StarMine, today remembered as a Quant solution with multiple factor models, derived over 2/3 of its pre-acquisition revenue from desktop customers. It targeted Quants and hybrids. Remember that stat is from an era where Quant sales were juiced by the zeitgeist of a pre-Volcker Act bull market. There were prop desks on every corner. That UBS trading floor in Stamford was as big as it looked.

Today’s alt. data industry would do well to look at an example like StarMine, unless you’re happy to remain in a sub-scale $2-5M range for the foreseeable. AI should reduce the costs to this type of expansion, allowing you to provide far improved marketing and messaging.

Other paths exist. Of course M&A is another valid alternative to achieving scale, it just comes with a different set of risks.

This is such a fascinating space and hopefully you’ve enjoyed the article. Please feel free to comment below. What other options does the industry has? How do they reignite explosive growth? What comes next?