The conversation around the interaction between data, algorithms and humans is evolving.
Once again, A crisis like no other!
The COVID19 induced crisis of course, provides us with more data and experience on the topic of data, algorithms, investing etc.
Once again, A crisis like no other, is actually the title of an AQR article at the end of March. The 20+yrs old quant king has suffered after a very difficult 2018 (the so-called Red October) which led to large outflows and over 1000 layoffs- Investors pull billions from quant king AQR as performance slumps – Financial News.
There has never been a better time to bring up the conversation on whether Data (fundamental, conventional and alternative) made a difference during this crisis. Did algorithms help and in what way? And what about the interaction of humans with Data and algorithms?
Fundamental macro data is essential, and this will never change. What can and should change is that it time lag and that not all data sources are trustworthy. We need to build a real-time information system aggregating trustworthy data, that starts with fundamental data and extends to the currently coined `alternative data`. Eventually, we will get rid of the alternative data term, all together.
Lykke recently launched, the Open Initiative project which is a competition with a CHF50k funding award. One of the four different thematics of the competition is focused on building a Real-time information system. Lykke proposes a Wikipedia like system (in certain aspects) with revenue streams built on Blockchain and accessed via APIs. This vision merits a separate article.
In May, I reached out to experienced Quants in different positions and discussed with them about data, machines and humans through the crisis and in the new normal. The demand for alternative data, as expected, spiked during this crisis because everybody needed to access the situation in real-time and needed real-time measures. Investors, traders, portfolio managers, pensions, who were already using some kind of alternative data, needed more and relied heavily on high-quality real-time data. The reliance on such alternative data and on actionable techniques to access exposures and make intelligent predictions, skyrocketed. New entrants in the alternative data space, flocked as they needed to manage risks and exposures.
Those in the field managing risks, uniformly confirmed the increased need to manage thematic and narrative risks. In plain words, humans needed to understand their exposure to airlines or to China during the crisis and in the new normal. They needed to ask the data and the machines what the impact of COVID19 would be in real estate. And on and on….
Humans continue to be in charge of narrating the topics of interest or at stake. The machines need to be able to offer actionable insights and forecasts.
There is an increasing need for real-time and continuous re-assessment of this kind of complexity through lots of high-quality real-time data of all sorts. Machine learning and adaptive trading algorithms that reflect and retrain gave confidence to humans in certain asset classes (e.g. commodities) during this highly uncertain period.
Building trust between humans and machines has always been essential and will continue to be. The recent crisis was a painful but valuable experience that built more trust in humans as to what machines can currently do and with what inputs.
There has been an improvement in actionable techniques that allow humans to extract signals towards generating alpha by combining high-quality real-time data and adaptive algorithms. There has been a better and larger offering from providers, of data, insights, algorithms.
To write this article I looked into and spoke with people in the following companies.
Ravenpack is a leading data analytics platform headquartered in Spain. I was using their free Coronavirus dashboard during the lockdown which had lots of alternative indicators (panic, media hype, fake news etc). You can check out their research around sentiment impact and sentiment investment strategies – here. Ravenpack highlights that negative sentiment has more predictive power for asset prices than positive sentiment.
S&P Global market intelligence is a leading player in alternative data and actionable techniques. They measure people’s flows and money flows and other data ahead of delayed economic data reporting. During the crisis, their free Quantamental Research offered several insights. For example, reports on how COVID19, the weather temperature, and real estate prices (REITs) were connected. Two years ago, S&P Global acquired Kensho. One of my first articles on DailyFintech in April 2015, profiled Kensho: Warren is like Watson and Siri for analysts, investors and traders.
The Book of Alternative Data, by Wiley coauthored by Alexander Denev and Saeed Amen will be available in July. It is a book covering ways of leveraging alternative information sources in the context of investing and risk management.
Sentifi is a Swiss company that was focused on sentiment data and has now built an enterprise analytics platform using AI and sentiment data.
Macrosynergy is a UK macro hedge that is currently acting as a fiduciary quant house for long term institutional investors. They are combining fundamental data and insights with algorithmic trading.
Predictive Layer is a Swiss company focused on predictive Analytics and forecasting applications. Their algorithms adapted to this crisis and offered insights that only machine learning can produce in a timely fashion. Their value add was clear in the commodities space during this crisis.
New readers can see 3 free articles before getting the Daily Fintech paywall. After that you will need to become a member for just US$143 a year (= $0.39 per day) and get all our fresh content and our archives and participate in our forum.
The post Alternative Data gained more trust and ML algorithms got good training appeared first on Daily Fintech.