1 min read
With Alternative Data, More is Better.
More is not always better, but when it comes to integrating Alternative Data sources into predictive models, that seems to be the case. Data scientists at Maiden Century ran a study based on thousands of predictions of publicly reported company KPIs and tested various forecasting models based on the number of datasets they used for inputs. They found that the average prediction errors tend to decrease as you add more and more data sources. This is especially true with the first five datasets. The rate of improvement slows down significantly after adding the 10th dataset.
Even as the spending on Alternative Data grows exponentially, most Hedge Fund managers and other investment management firms, still rely on fewer than three data sources, implying that they are leaving some Alpha on the table. Better predictions yield better results, especially if you are one of the few firms in the market that can make superior-quality forecasts.
The reason that managers are reluctant to add a lot more data to their mosaic has to do with the challenges of integrating and monetizing new datasets. There are a series of steps, both time-consuming and resource-intensive, that hold managers back from seeing value on a shiny new dataset.
Just normalizing and cleansing the data can take hundreds of hours – hours that are precious to the money manager.
Luckily, this is one of the many problems that technology can help overcome for managers of all sizes.
Download our free ebook and learn how capital managers just like you overcome the challenges of building a robust AltData operation.
Ready to talk? Get a no-obligation demo of our time-tested industry-trusted platform.