👩‍💻 D.R.E.A.M. - (Alternative) Data Rules Everything Around Me 📊

:notes:
Data rules everything around me
Ping
Transform the data
Retrain another model ya’ll

(adaption from an old song)

Asset managers are becoming more data-driven

Proprietary datasets + infrastructure + talent = the edge for Amazon, Apple, Netflix, Spotify, YouTube, Microsoft, etc. Some of them also hold the most cash and short-term investments /marketable securities on their balance sheets. No, Apple is third in that list.

Hedge funds are after all kinds of data too, as traditional sources, such as company filings, aren’t enough.

Humans are creatures of habits

We humans are very predictable and our stuff can be machine learned.

Facebook makes a living by printing billions every quarter thanks to our data.

Asset managers are looking to profit from “alternative” data too.

The digital trail we leave when logging in, swiping, liking, posting, listening, skipping, clicking, etc creates many many labeled datasets for companies to run predictive machine learning algorithms on.

Data brokers collect data when you visit websites and use apps. The likes of Apple (with Safari) and Mozilla Firefox are very much trying to limit that. But they can’t stop the developer tools - the SDKs etc.

So next time you want to buy a cheap connected toaster (why?) or a “Smart TV” (WIFI-connected computers), think about why they are cheap and whether the firmware is any good and up to date.

Even if it’s made by Roku - they are not just a hardware company.

Your apps are leaking data like crazy too thanks to all the APIs and SDKs.

Labeled datasets - we did the hard work (the labelling bit)

It’s tough to mark Xs with Ys so the algorithms can learn the relationships. Especially when the size of the data is in petabytes.

Fortunately, we already do this by interacting with applications and filling in forms and liking stuff.

Labeled datasets can be fed into supervised and semi-supervised learning algorithms. Sometimes (a lot of the time) running a regression can do the job - no advanced deep learning techniques required. No AutoML etc.

Just beware of the garbage in-garbage out problem.

Finance and data mining hedge funds

Bloomberg:

From Fitbits to Rokus, Hedge Funds Mine Data for Consumer Habits

…That’s why some of the world’s biggest hedge funds, from Steve Cohen’s Point72 Asset Management to Ken Griffin’s Citadel, have been snapping up large swaths of alternative data. Many are paying big money for it.

“There is not one major hedge fund or asset manager that doesn’t have data initiatives underway or that are not using alternative data in some way,” said Michael Marrale, chief executive officer of M Science, a firm that provides data and analytics to hedge funds.

Firms can keep tabs on the number of Roku video-streaming devices or Fitbit fitness trackers being used, the length of time consumers spend on them and their approximate locations. Similarly, if you buy a Tesla Model 3 car and use its Bluetooth-enabled media, a data provider can capture when your new ride is hitting the road.

Source - https://www.bloomberg.com/news/articles/2019-07-24/from-fitbits-to-rokus-hedge-funds-mine-data-for-consumer-habits

1. What are examples of alternative data?

A former chief economic adviser to Indian Prime Minister Narendra Modi crunched data such as vehicle sales and electricity consumption to show that the nation had overstated annual growth by about 2 percentage points on average from 2012 to 2017. Researcher and entrepreneur Apurv Jain found clues to where U.S. employment was headed by analyzing 1.2 billion tweets from 230,000 Twitter users who posted about losing or finding a job. The hedge fund Point72 Asset Management shorted Weight Watchers shares after reviews of social media, online search, and credit-card transaction data suggested that competitors were gaining momentum.

3. Who uses the data?

Some of the world’s biggest hedge funds are leaders in snapping up large swaths of alternative data, many paying big money for it. Investment firms that use so-called quantitative strategiescan pump the raw data directly into algorithmic trading models. Trying to get an edge is as old as investing itself, but the profusion of alternative data sources – for those investors who can afford them – can offer faster and more detailed analysis than the government economic reports released on a monthly or quarterly basis.

4. How big is this business?

The number of providers has tripled in three years to 1,126, according to Eagle Alpha Ltd., an alternative data provider that projects global spending will reach $900 million by 2021, nearly double the 2017 level. (AlternativeData.org, which collects information on the industry, puts the number of providers at 445.) Players include NPD Group Inc., which says it crunches millions of receipts from brick-and-mortar stores and e-commerce sites to analyze consumer trends; Quandl Inc., which used aircraft-tracking data to sniff out a deal in the works between Occidental Petroleum Corp. and Buffett’s Berkshire Hathaway Inc.; Thasos Group, which measured overnight smartphone activity inside Tesla Inc.’s headquarters to anticipate a surge in production of its Model 3; satellite tracking firms Orbital Insight Inc. and Ursa Space Systems Inc.; and Predata Inc., which vacuums up data from online conversations and comments to feed country-specific “geopolitical risk” indices. (Bloomberg LP, the parent of Bloomberg News, providesclients with access to alternative data.)

Source - https://www.bloomberg.com/news/articles/2019-11-09/how-big-investors-cash-in-on-alternative-data-quicktake?srnd=premium

D.R.E.A.M.

1 Like