According to media headlines, hedge funds spend millions on alternative data. Cost of alternative data sets is one of the most popular discussions at industry events. New vendors are struggling to determine the right price to charge, and data buyers complain about lack of transparency when it comes to pricing.
In search of a meaningful formula, data vendors came up with an AUM-based model. The logic behind it is that larger funds make larger bets and make more money using the data set. Although this approach makes sense and can be used as a ballpark reference, it doesn’t take into account the way funds use data sets to extract alpha.
Other criteria vendors have been relying on are novelty and uniqueness. The assumption here is that new untapped data sets are always more valuable for alpha generation and should be more expensive than widely available data sets that have been on the market for a while. In practice, tolerance to data set overcrowding is predicated on the manager’s investment horizon and the time it takes for a signal to get reflected in market prices. Some data buyers may actually prefer widely used data sets.
Asset managers often determine the price they are willing to pay for the data set by calculating estimated return on investment. The problem is that most managers will not be disclosing sensitive information about return attribution, plus ROI depends on many other variables aside from the data set itself, so this approach is also not suited to transparent standardized pricing.
In the alternative data space, for different types of buyers, demand is driven by different factors, which is the reason why no “one-size approach” works well for price discovery.
Data buyers in the investment industry may be roughly classified into two categories that include:
- Portfolio managers who harvest small amounts of alpha from each of a large number of data sets they use — these are highly diversified strategies aiming to make a large number of small bets (typical quants), and
- Managers who harvest large amounts of alpha from each of a few data sets they use — these funds tend to have concentrated portfolios and make a limited number of large high-conviction bets (typical fundamental).
Quants view data as a commodity — to them, data sets are replaceable. Moreover, they are specifically looking to combine, replace and add new data sets to optimize ROI of the overall diversified portfolio of data sources. For example, if a quant fund trades a few thousand securities on a daily basis, they will consider onboarding any data set that covers the universe, is somewhat orthogonal, meets data quality criteria, and adds a few extra basis points of alpha, regardless of the data category (e.g., consumer transactions may be replaced with shipping data).
Such funds often employs internal data engineers, data scientists and data sourcing teams. They review and evaluate many data sets, and they understand the market well in terms of products available from competing vendors as well as range of market prices. For new data vendors, quant clients will be a great resource of market wisdom to determine an appropriate pricing range to start with.
Since quants view data as a replaceable commodity, a vendor competes not only within specific data categories but also with all other data sets and vendors. Thus, commodity pricing approach will be appropriate in this case, i.e., price is capped by what a competitive market will bear. In most cases it will be hard to pressure quants to pay a sizable premium for a data set.
On the bright side, typical quant tends to stick with the data sets in which they found alpha and will likely renew subscription for a few years. They take a long-term view in terms of performance expectations of data sets and may be willing to keep temporarily underperforming data sets in the portfolio if they see a long-term potential.
Typical fundamental funds are normally interested in very specific data focused on exact themes and companies. For example, if a fundamental manager placed a sizable bet on Tesla and their dedicated internal analyst is researching Model 3, for them, at that exact moment, every bit of extra information about Tesla Model 3 is priceless — the more granular and orthogonal the better. At that moment, they are ready to pay a premium for the data if it gives them a sliver of insight to complement their research findings. They can’t replace such a specific niche data, and often a vendor will have little or no competition.
Fundamental funds may be willing to pay a premium for novelty, timeliness and uniqueness. In this case, vendors can apply value-optimized pricing strategy where price is based on the perceived value of the product and is often specific to the client.
The challenge is that typical fundamental funds’ data needs tend to be short-term; it is often a one-off deal. They need the data set here and now to make a decision on that one large position. Once the position is closed, they no longer need the data.
Ideally, in order to build a long-term sustainable business of selling alternative data, the vendor should strive to have a portfolio of both types of clients: those who perceive data as a commodity and those who are willing to pay a value-based premium.
The next important set of factors determining the value and a fair price of data set has to do with data quality and data structure. These factors’ weights in the data pricing formula will also be different for two client groups.
When selling to a quant, you can add a premium for the data set with five-plus years of point-in-time data, accurate mapping to security identifiers and a consistent, well-documented methodology for data collection. A data set that requires extra QA work and has quality issues, like backfills or methodology change, should be offered to this type of client with a discount.
For fundamental funds, suitability and timeliness are the most important bases for pricing. Point-in-time history may have lower (or even zero) weight, while data uniqueness and orthogonality will add a premium.
By understanding what criteria the fund uses in its value judgment, the vendor gets a better concept of the appropriate sales strategy and pricing structure. Vendors who stick with a one-size-fits-all pricing model can potentially miss many opportunities. In contrast, those who adhere to a client-centric approach will be able to build a stable combination of multi-year contracts paying cost-plus price and shorter-term value-based contracts with premium pricing.