Big Data and Technological Innovation in Finance
Big Data and Technological Innovation in Finance The financial markets have changed inestimably since the 1792 Buttonwood Agreement and consequent establishment of New York Stock and Exchange Board.
In 1792 1/8 of a dollar was the smallest amount a stock could change in value, asset pricing was determined based on dated news, and size of crowd present.
Today a fraction of a cent and minute by minute news determines buy or sell decisions.
The financial markets have gone through deep and elaborate changes and are continually evolving.
There are a number of enabling factors that have fueled this transformation with the most important ones being technology and access to data.
Other factors have been the growth of many companies into becoming multinational corporations and massive trade across border lines which has created existentialist dependencies between countries for their day-to-day functioning. This has led to the globalization of the economy.
In today’s market, the health of the domestic economy or a corporation’s profitability is closely tied to the global economy. Throughout all these changes the one thing that has remained constant is the uncompromising need for maximizing ROI while managing risk at an acceptable level.
Today’s financial market offers an increasingly wide range of asset classes and investment opportunities. Massive amounts of raw and processed data and news are readily available to all. Despite all of this, managing assets has not become easier.
The inherent data and information transparency has allowed markets to become more efficient.
Such efficiency has made it nearly impossible for the average investor to systematically produce alpha and has even made it difficult for fund managers to deliver superior risk adjusted returns without using massive amounts of leverage on various arbitrage trading strategies.
Such an environment has led to increased speculation among retail traders. Furthermore, the economy and global trades have been politicized and become political tools in the world arena, replacing military might, offering great potency.
The politicization and activism of the financial markets domestically has also been a new phenomenon.
Considering the aforementioned, accurately forecasting future returns requires more than traditional fundamental and technical analysis. Quantitative analysis and machine learning technologies are employed to fill this gap. Quantitative analysis enables reduction and processing of large amounts of market data that are being continually generated.
Machine learning has added a new dimension by its ability to sip through textual or numerical data and identify signals, sentiments, correlations, and to provide a factual assessment. The large firms that have the foresight and can afford to broaden their analytical infrastructure to employ these technologies have an edge over the others. Financial firms are at the early stages of employing these technologies.
Findings from ML, quantitative, fundamental, and traditional financial analysis must be fused to produce actionable results.
As employing more advanced quantitative analysis and ML techniques become more prevalent in finance, the need and hunger for more data increases.
Market efficiency improves as application of quantitative analysis and ML becomes more ubiquitous.
It can be assumed that the big buy-side funds are the ones competing for access to relevant data to feed, train, and back test their trading tools and platforms. For them advanced quantitative analysis and ML has become the true differentiators in hedging the market and getting ahead of the retail investors. Data has become the fuel. The more relevant and potent the data, the better the engine performs.
Now, some hedge funds are using satellite imagery to scope out potential bets with some success; however, even this seems like an inefficient way at potentially finding a competitive advantage. What are the other potential sources of data?
Many corporations generate and possess high quality data that are used internally as part of their supply chain management or determining marketing strategies or increasing their page view or click rates on their website. These are potentially valuable data for financial platforms that, if harvested properly, will add depth, clarity, and improve performance.
For example, if a firm is using NLP models for textual analysis, it would be useful to acquire data from the publisher on how many times that text was viewed and the length of time it was viewed for. This will help calibrate the weights being assigned to each signal/stem word in the model.
Furthermore, if a firm is analyzing the new housing trends, analyzing current data from Home Depot regarding the rate of sale of products used in new constructions may yield a greater competitive advantage than simply using aggregated data from Home Depot or Lumber Liquidator’s 10-k or 10-q filings.
Another example is monitoring the sales of Apple products from Amazon, Best Buy and Costco and correlating them against past data to produce patterns that may be used for predicting Apple stock performance or their earnings.
The same relationship may be correlated and combined with other sectors/securities in the market to predict consumer sentiment and so on. The power that such corporate data has in facilitating greater foresight into future market dynamics makes it clear that there is still a lot of room for improvement as it relates to fully exploiting quantitative strategies.
Currently, data generated by corporations are used internally for their own specific applications. If these datasets from different firms were warehoused, aggregated, and sold to financial firms for use by their financial engines to improve their performance, markets could potentially become more efficient and less volatile.
Figure here shows how such data could potentially be used by a notional financial platform engine:
Notional Financial Engine Architecture
There are many companies that are generating and using large amounts of data internally. As discussed, the same data can be used and repurposed. Selling this data can become another source of revenue for these companies. These data are usually considered proprietary.
A trusted third-party data warehousing company can enter into agreements to collect and sanitize their data and potentially combine them with one another and sell them to financial firms, economists, etc. This could be a potential business model that can generate a new revenue source for companies while serving the need of data hungry financial firm platforms.
Going forward, the development of financial platforms and engines will be punctuated by their hunger for data.
As these platforms evolve, they add to transparency of publicly traded companies. Such transparency may further hinder the average investor’s ability to compete with firms that are able to leverage such data; however, it should also allow for more stable markets which is good for any investor in the long run.
Big Data and Technological Innovation in Finance Written By Kevin Manouchehri
Big Data and Technological Innovation in Finance
