THE FUTURE OF QUANTITATIVE ANALYSIS: DATA SOURCES ByFactSet Insight | December 18, 2018 Financial firms have always relied on an array of data and technological processes to get the intelligence they need to make decisions. However, with the transition to today’s advanced technology environments occurring almost overnight, many organizations are still relying on the same data architectures, software, and workflows that they have for many years. One opportunity that firms now have at their disposal is alternative data (content derived from varied sources of information like images, billing statements, heat maps, social media interactions), which has become an increasingly common consideration over the last five years, and offers a number of interesting new possibilities. Satellite data, for example, can provide real-time insight on car traffic to a business locations and indicate how a company is performing. While the rise in alternative data is an important trend to watch, data sets like fundamentals and benchmarks are still a critical part of the quantitative analyst tool kit. Identifying how these core data sets are being used can shed light on the incorporation of new and novel feeds. To determine which data sets are valued most today, FactSet partnered with Coleman Parkes to survey Heads/Directors/VPs of Quantitative Analysis, Senior Quantitative Analysts, Data Scientists, and Chief Data Officers from 50 hedge funds and institutional asset management firms across the U.S., EMEA, and APAC. The results provide insights into the types of data quantitative analysts value and use today, and what types they are likely to value more in the future. https://insight.factset.com/the-fut...Oe11tMC0WLbVJZMinwA-t3jn6S6_T4&_hsmi=68434749
I have a feeling Semantic Web technologies such as OWL, Linked Data will have some part of this future. assuming Google or amazon does not buy them all then shutter them down the way they did freebase. I might do 1 form of that for my thesis.. "Bad news , good price action " kinda thing... imagine XYZ making new lows past 2 weeks then stock is unchanged for the day in spite of existence of some really bad news from unstructured data using semantic technologies.
I recall distinctly the moment when I started to think that economics was cool...... it was a basic Price Theory/Chicago School set-up: everyone is rational, full information abides, "choice" is the word of the day.... And along come Giffen goods, Bandwagon Effect, Snob Effect, and the enigma-wrapped-inside-a-riddle Veblen Effect. I *love* lecturing on these -- students just eat it up. (Note to self: the next time you teach, *start* with the cool shit.) Anyway... (excepting Giffen Goods' being solely explained by a 'choke price' in one market leading to an increase in real income in another market), these are all concepts which abuse/push around the basic demand curve equation of D = f[income, substitutes/complements, tastes] So, if bad news comes out about XYZ stock, that might diminish your 'taste' for holding the stock, and cause your demand curve to shift left-towards-the-origin. However, what if XYZ had already moved lower on anticipation of the bad news??? How many times haven't we heard of bad news becoming the stimulus for a *surge*, because the news wasn't as bad as thought (or simply that the news had *killed* the uncertainty)??? A LOT, right? (And so, because substitutes' prices were relatively higher, the XYZ demand curve shifts back rightward, perhaps further than the move left.) Here's the cite that got me going, 'back in the day.' https://academic.oup.com/qje/article-abstract/64/2/183/1931945?redirectedFrom=fulltext Sorry it's not the whole enchilada, but maybe somebody can dig up a full url?!?!? A bottom-line thought? So much of those three D-curve twists -- Bandwagon, Snob, and Veblen -- each of them work heavily on available market information. Is the info directed at economic substitutes or complements? It's easy enough to describe, and speed... or 'fullness' of the information? Easy enough to model. But when you get into 'tastes'??? Things start to get *interesting*. And so how this portends to the *market* for all that info??? Whoaaaaaaa. More fun (in Information Theory) than has been had in 20+ years. And YOU are in the middle of it.
The issue is that there are no incentives to open up. It is in fact, quite the opposite... far simpler than using AI to plumb sites. Sadly, you might be right though....the faint heartbeat is LD/Json?