Part 11. Fintech in Investment Management Flashcards
Fintech
This refers to developments in technology that can be applied to teh financial services industry.
Primary areas where fintech is developing include:
- Increasing functionality to handle large sets of data that may come from sources and exist in variety of forms.
- Tools and techniques such as artificial intelligence for analysing large data sets.
- Automation of financial functions such as executing trades, and providing investment advice.
- Emerging tech for financial record keeping that may reduce the need for intermediaries.
Big Data
A expression that refers to all potentially useful info that is generated in the economy.
Includes:
- Traditional sources such as financial markets, company financial reports, government economic statistics.
- Alternative data from non- traditional sources.
Non-traditional sources of data:
- Usable data - social media posts, online reviews, email, website visits.
- Corporate exhaust - bank records, and retail scanner data from businesses.
- Internet of things - sensors such as radio frequency identification chips embedded in smart phones, and smart buildings.
Characteristics of Big Data:
- Volume - data continues to grow by magnitude; referred to in units as bytes.
- Velocity - how quickly data are communicated; real time data is low latency, but data communicated periodically or with lag is high latency.
- Variety - varying degrees of structure in which data may exist; such as spreadsheets and databases (structured form), photo and webpage code (semi structured form), video (unstructured form).
Data science
This concerns methods of processing and visualising data, processing methods include:
- Capture - collecting data and transforming it into usable forms.
- Curation - assuring data quality by adjusting for bad or missing data.
- Storage - archiving and accessing data.
- Search - examining stored data to find needed info.
- Transfer - moving data from their source or storage medium to where they are needed.
AI
Aim to eliminate/reduce possibilities of outliers, bad or missing data and sampling biases.
Processes larger volumes of data, programmed to simulate human cognition.
e. g. neutral networks
* machine learning = a computer algorithm is given inputs of source data, with no assumptions about their probability distributions, and may be given outputs of target data.
* Designed to learn without human assistance, via training dataset to look for relationships, validation dataset to refine relationship models, and test dataset to analyse predictive ability.
Supervised learning
The input and output data labelled, the machine learns to model outputs from inputs, then machine is given new data on which model to use.
Unsupervised learning
The input data are not labelled and machine learns to describe the structure of data.
Deep learning
A technique that uses layers of neutral networks to identify patterns, beginning with simple patterns, and advancing to more complex ones.
May employ supervised or unsupervised learning, applications include image and speech recognition.
Overfitting
This occurs when machine learns the input and output data too exactly, treats noise as true parameters, and identifies spurious patterns and relationships.
Creates a model too complex.
Underfitting
This occurs when the machine fails to identify actual patterns and relationships, treating true parameters as noise.
This means model is not complex enough to describe data.
Results can be a black box, producing outcomes based on relationships not readily explainable.
Text analytics
The analysis of unstructured data in text or voice forms.
e.g. frequency words and phrases, used to partially automate specific tasks
Natural language processing
The use of computers and AI to interpret human language.
e. g. speech recognition, language translation.
- Potential use to check for regulatory compliance in examining employee communications.
Algorithmic trading
This refers to computerised securities trading based on a predetermined set of rules.
e. g. algorithms may be designed to enter optimal execution instructions for any given trade based on real time price and volume of data.
- Useful for executing large orders by determining best way to divide orders across exchanges.
- High frequency trading = identifies and takes advantage of intraday securities mispricing’s.