Predictive Modeling is set to disrupt the finance sector – Guest Editorial
“Ultimately, a broad understanding of predictive analytics can help individuals and companies to parse future scenarios. AI has become an integral part of prediction, and rightfully so” says Revel Partners senior executive Thomas Falk
By Thomas Falk, a German serial entrepreneur and investor. In 1998, at a time when few believed in the potential of the internet, he founded Falk eSolutions, one of the first internet companies in Germany. Eight years later, the company merged with its American competitor DoubleClick and was acquired by Google in 2007. What followed was a serial founding of successful businesses, including United MailSolutions AG and smartclip. In addition, Thomas Falk supported companies such as EyeWonder and The Trade Desk as a board member from formation to maturity. Today, Thomas Falk is a General Partner at the US venture capital fund Revel Partners and primarily invests in European and American companies in their growth phase.
When using predictive modeling, we seek to predict what is most likely to happen in the future based on Big Data, AI and statistics. The data models arising out of this can prognosticate almost anything – from weather to sports game outcomes to TV ratings.
Predictive modeling, also called predictive analytics, has furthermore found its way into healthcare, marketing and retail, in the latter of which it is often used to improve the costumer experience by predicting what the shopper is most likely going to buy. Due to its broad spectrum of applications and its ability to substantially improve business results, predictive analytics is among the top three technologies industry professionals are investing in in 2020, a recent Statista survey found – a finding that corresponds with my own experience and observations.
Naturally, one would also like to foresee the future in the financial sector, especially when speculating in stock markets. Predictive models can be built for different asset classes such as stocks, currencies or commodities.
To do so, mathematically advanced software is applied to evaluate indicators on historical data like price, volume and open interest, and to discover repeatable patterns. The technology is extensively used by trading firms to develop strategies for trade, as a successful prediction of a stock’s future price maximizes investors’ gains. 80 percent of the daily moves in U.S. stocks are machine-led, fund manager Guy de Blonay told CNBC in 2018.
New start-ups have also been shifting their focus to use large amounts of data (Big Data) or artificial intelligence with hedge funds. San Francisco based blockchain start-up Numerai is one of them.
The AI-run, crowd-sourced hedge fund is calling data experts to take part in a weekly tournament for the best forecast model. The data scientists calculate forecast models using encrypted data sets representing stock market information and create an algorithm to discover corresponding data patterns. In the case of Numerai, the different calculated trading recommendations are evaluated during the weekly tournament, where the actual situation and the proposed models are being compared.
The underlying mathematics and algorithms may sound complicated, but they translate into measurable financial gains: Finance data platform Preqin found in 2019 that AI funds outperform the hedge fund benchmark. In order to investigate this, the platform tracked performance information for 152 artificial intelligence hedge funds. Based on three-year cumulative returns, between August 2016 and June 2019, AI hedge funds outperformed the Preqin All-Strategies Hedge Fund benchmark by a margin of three percentage points. AI funds returned +26.96% over the past three years and all hedge funds returned +23.87%.
Given examples like Numerai and the increasing interest in the technology by professionals in various sectors, predictive modeling seems to have proven itself as highly beneficial. It does provide managers and executives with decision-making tools to influence, forecast, and optimize processes, but it’s not the ultimate panacea.
Data is everything in predictive modeling, but it isn’t infallible. No data model saw the corona crisis coming at this exact point in time, as our modern society has not been hit by a pandemic of this magnitude before.
We were able to adapt, make short-term forecasts and initiate countermeasures, which is why we are now seeing the stock market recovering from the first corona clash: The Dax and the Dow Jones are up again. In the United States, job market data are providing a boost, as the situation on the employment market eased again in May. Lack of historical data to take into consideration makes it hard for data scientist to develop accurate models, but even if sufficient data is available, algorithms might fail to consider variables — especially when it comes to predicting human behavior.
Hence, investing and trying to predict the behaviors of our world’s investors is considerably trickier than predicting the next Amazon purchase. It takes the best scientists to continuously calculate forecast models. Continuous innovation is paramount. No matter how promising any current winning formula is, in a fast-moving world driven by innovation and sudden fluctuations, it is short-lived.
Ultimately, a broad understanding of predictive analytics can help individuals and companies to parse future scenarios. AI has become an integral part of prediction, and rightfully so. But it will always require a mix of different technologies, large data sets and human experience to be the most accurate.
The subject matter and the content of this article are solely the views of the author. FinanceFeeds does not bear any legal responsibility for the content of this article and they do not reflect the viewpoint of FinanceFeeds or its editorial staff.
