Alternative Data Category Description
Financial Data is a type of alternative data that contains datasets and categories that fall between traditional data and alternative data. Some years ago Eagle Alpha clients began asking for data that fell into the traditional realm. The thought process was that alternative data would just become data at some stage. We have broken this data category into four key sub-categories. Tick and pricing data falls into the traditional data bucket. However, asset managers are looking for other sources to Bloomberg, Refinitiv, and Factset for backtesting, trading, and portfolio risk assessment. As a result, we have established a pricing or tick sub-category.
The second sub-category is flow data, which might be considered a grey area between traditional and alternative data. Over the years flow data has seen a very strong uptake across all fund strategies from quant to macro/CTA. Flow, or money allocation, is used to track where fund managers are putting money to work. Flow datasets show where money is being allocated at the asset class level, sector allocations, at the ticker, and bond type. The third sub-category is fundamental data, which includes vendors that offer up this type of data in a novel way.
The last sub-category is trading signal datasets. We regard this as alternative data, with the financial market, or pricing, inputs. It has become quite a large sub-category. We had thought of allocating a separate primary category but refrained. Data vendors use inputs of various types along with stock prices to build a trading signal or sentiment indicator across the spectrum of asset classes. In general datasets across the financial markets category sell for in a region of $40 to $80k but might hit $100k.
Subcategory - Flows & Allocations Data
Market participants want to know where other market participants are allocating capital. Capital flows show if investors are allocating capital to bonds or equities. In a typical 60/40 strategy where is the money going, overweight equities or overweight bonds? With an allocation what type of equities and what type of bonds? What parts of the market are seeing capital inflows or outflows? Simplistically, don’t fight the tape. Collectively if investors are selling tech stocks and buying value stocks they can trade with the trend. In recent years with more capital flowing to passive investment vehicles flows into ETFs and ETF rebalancing have become an important flow allocation trend for systematic and HFT funds. In a similar manner to equity flows, credit market investors monitor the corporate debt market junk and investment-grade bonds. In recent years investors track regional flows into/out of the US, Europe, Japan, China, and emerging markets. Within these regional markets where capital is flowing, bonds, equity, and what type of bonds and equities.
Subcategory - Fundamental Data
Fundamental data can be considered traditional data but there are alternative data providers that offer up this type of data in a novel way or interpret and present fundamental data in a different manner. There are many alternative data providers that offer fundamental data, KPIs, or valuation models that are allocated to this alternative data sub-category.
Subcategory - Tick & Price Data
Tick and pricing data largely covers actual pricing data across all market instruments from equities, ETF, bonds, FX, Options, futures, and commodities. One new area of pricing data is for the Crypto markets. On-chain and reference data have seen a lot of interest with the growth in Crypto in recent years. As pricing data is more traditional in nature the history can be longer than alternative data datasets. Pricing data usually comes from some sort of direct feed with exchange and is then repackaged for delivery to an asset manager. The data can be the end of the day or intraday, but the infrastructure needed for intraday feeds needs to be more developed. Equity data can be a basic closing price to deep level three data.
Subcategory - Trading Signals Data
These are datasets that can use several inputs, such as fundamental data, sentiment scores, factor models, and market prices to build a trading signal. These are typically for equities but can also be credit signals for corporate debt markets as well as options market signals. We have put these datasets in the primary category of financial data as the signals typically use financial market data, like stock process and credit spreads, as a core input of the data offering
Data Structure
- This category of alternative data is generally mapped to the ticker and PIT
- History is mostly at least 7-10 years and can be much longer
- Given the data is a trading signal it is at least daily and hourly can also be expected
- Delivery can be via API, S3, and platform, depending on the date vendor and use case.
Compliance Considerations
Trading signals, modeled output, or a factor model of some type do not in general run the risk of Personally Identifiable Information or Material Non Public information. There can be multiple inputs to a model that contains traditional data and alternative data. The alternative data is usually derived from some other source and NLP, AI, or machine learning is then applied to it. The one thing to focus on for this alternative data category is data provenance. The alternative data provider can source data from multiple data sources and compliance, data provenance, and contractual relationships need to be examined closely. For example, if a data vendor is using a sentiment score using an NLP model on news sources the compliance question is are the news sources all public facing. Assurance would need to be made that the third-party supplier is not using news accessed from behind a paywall.