1. Data Analysis
With the abundance of data available, it would typically require thousands of human hours to collect, compute, and decipher the underlying correlations to come up with an investment thesis. KDI Invest is able to streamline the process through automation, ensuring high data quality and accuracy. Like a robot that never sleeps, our automated data collection system screens and reviews over a thousand data points each day and categorises them into several key areas such as fundamental, technical, and economic data. The data then is converted into unique formats which are then fed into our machine learning algorithms.
2. Machine Learning
The cornerstone of KDI Invest’s A.I. investment framework, machine learning is governed by two major processes. The first is “feature selection”, where undesirable or inappropriate input variables are detected and deprioritised, leaving the important factors that are expected to be most useful to the models. The trained data is then fed into the machine learning processes that make use of the features identified to make predictions about the investment performance variable.
KDI Invest uses multiple machine learning methods, according to the underlying conditions at any given time. Machine learning methods range from the standardised “supervised learning” techniques, which include the most basic regression (y=mx + c) to the increasingly advanced “unsupervised learning” techniques such as k-means clustering or even deep learning. KDI Invest makes the best use of machine learning’s predictive analytics to help determine the best possible outcome for investment decision-making and asset allocation.
3. Asset Allocation
Once the machine learning process is performed to evaluate the ‘investability’ of each exchange-traded fund (ETF), KDI Invest then employs the use of A.I. to run the asset allocation automatically in accordance with the Markowitz Portfolio Theory (MPT). First developed by Nobel laureate Harry Markowitz, MPT is an optimization method whereby portfolios are constructed with the goal of ensuring the risk-adjusted return is at its most efficient level.
KDI Invest understands that not everyone has the same risk profile. After using machine learning to contextualise the characteristics of each ETF to the ongoing economic conditions, KDI Invest then uses a mean-variance optimization process to calculate the efficient portfolio allocation that: maximizes expected returns, while minimizing expected risk, all the while taking into consideration each client’s unique characteristics such as investment horizon, age, and risk tolerance.
4. Trading & Rebalancing
Upon completion of the portfolio allocation process, KDI Invest further leverages on A.I. to derive the required number of shares to be transacted on behalf of all clients on an omnibus level. This includes fractional shares to ensure that the targeted asset allocation is achieved. Doing this via A.I. at an omnibus level allows investors to enjoy economies of scale instead of engaging the broker to perform buy/sell transactions for each individual investor. A.I. technology is also used in daily monitoring, whereby rebalancing will be performed automatically if there is a significant deviation in weightings from the targeted asset allocation.
If you are interested in technology that will improve the quality of your investments, check out how KDI Invest can offer one of the lowest fees in town with A.I.’s efficient and precise portfolio management by clicking here.