Alternative data is one of those market buzzwords. The first step is to define precisely what we mean by alternative data. In practice, it is any sort of dataset, which is not traditionally used within financial markets. The idea of alternative data isn’t to use it as a replacement for those datasets we often use in finance, like market data. Instead, we should think of alternative data as a way to augment our existing models. If we have a dataset that fewer market participants use, we might be able to make observations which others are unable to make. However, it is important to note, that just because a dataset is unusual, doesn’t always mean it will help to generate alpha.
Each investor will have a slightly different approach for how they handle alternative data. For quants, typically, they will concentrate on those datasets which can be used for a wide variety of assets. By contrast, discretionary investors will typically invest in a small number of assets, and dig down into each of them more for research. Hence, datasets with a smaller asset universe are likely to be just as valued for them. Some datasets will be amenable to higher capacity strategies, whilst others are likely to be suited more to lower capacity strategies, and potentially might suffer quicker alpha decay.
One big question which comes to alternative datasets it quantifying the value of them. For quants this involves doing backtests and seeing whether the addition of the alternative dataset can help improve returns. Different investors are likely to place a different value on each alternative dataset, given the varying abilities to harvest alpha.
If you are interested in hearing more about using alternative data for trading, I’ll be speaking at the Quant Conference about The Book of Alternative Data: A Guide for Investors, Traders and Risk Managers, which Alexander Denev and I are currently coauthoring. Our book will be published on Wiley in 2020, and is currently available for pre-order on Amazon.