How a Korean Built AI is Revolutionizing the Investment Process

The pursuit of above-average returns is one that many have undertaken over the years, with most failing to do it consistently. With the rise of quant models in finance, there seemed to be a glimmer of hope in finding strategies that consistently worked. While these worked for some time, their effectiveness began to deteriorate as more people implemented them and it became harder to find alpha.

Marcus Kim, the founder of Qraft Technologies, was one such trader that saw this trend and decided AI was the best answer to this. Qraft Technologies, founded in 2016, is the product of Marcus’s dream. Qraft’s AI is the solution that powers four ETFs, QRFT, AMOM, HDIV, and NVQ. They have even garnered headlines with timely trades of Tesla, the addition of meme stocks, and positioning the weights of the portfolio to reduce drawdowns[1].

Inside The Supposed Black Box

While it would have been easier to use existing AI platforms and just apply them to the trading strategy, Qraft built the entire AI from scratch. By doing this, Qraft created an AI specifically tailored for investments. Therefore, the team is intimately aware of the capabilities and the improvements that the model needs.

Data is the fuel of AI. Just like a car engine, you put bad fuel in, and you can get bad results. AI is no different and, like a high-performance engine, it needs high-performance fuel. Financial data can be quite messy and cause problems with statistical models. Kirin API[2] is Qraft’s in-house built solution created to make sure that the data is pre-processed in attempting to back test accurately without biases. Once the data is ready, it then goes into the AI model. This can seem mysterious and can lead to some skepticism. However, Qraft’s AI is not doing much different than a human would. The difference is that it can process trillions of possibilities within a few hours. It can take in both traditional structured data (macro data, price, factors, etc.) as well as non-structured data such as patents and sector labels.

Unstructured data is illustrated by Qraft’s experimentation of sector labels. Current sector labels do not account for modern corporations that do business in multiple sectors. For example, is Amazon in the logistics, software, data, or retail industry? Determining the correct sector, or a combination of sectors could lead to more accurate multiples and valuations. To do this, Qraft AI can use NLP (natural language processing) in an effort to accurately label which sector or industry a firm may belong to. In addition, it allows for blended sector labels, which can allow for weighted combinations of sector labels for a firm. We believe this approach would be beneficial as the importance of certain assets, such as intangible assets, may be more important in the context of valuing a tech firm than a retail company. While this labeling is possible with humans, the sheer volume of data that needs to be sourced and processed can make it difficult to do it in a timely and cost-efficient manner.

Structured data can be filtered by the Kirin API, which attempts to accurately determine the quality of the data, excluding data that is deemed irrelevant. This “high-quality data” is then added into the Alpha Factory[3] platform, which is composed of the Factor Factory[4] and Strategy Factory[5]. The Factor Factory uses data from Kirin API to determine the factors that are relevant to the custom parameters set by the engineers. These factors include traditional linear factors such as P/B[6] ratio, market capitalization[7], and other frequently used financial metrics, as well as nonlinear factors such as intangible assets. We believe nonlinear factors offer an advantage over traditional linear factors because of its ability to price in factors such as intangible value, macro trends, and various additional factors. These nonlinear strategies can be combined with traditional linear strategies to seek to find a more accurate value for securities, creating a comprehensive strategy that balances both nonlinear and linear factors.

Identified alpha factors in the Factor Factory are then assembled into a portfolio in the Strategy Factory, which can run through all the possible combination of the factors that have been identified in the Factor Factory. Then it seeks to create the best weighting for each factor and to identify stocks with these properties. The valuation of individual stocks considers the sector labeling through NLP identified above. Factor Factory produces a final list of stocks, given the identified factors are expected to yield alpha[8] value.

After these processes, the AI puts together what it believes is the best potential portfolio. This includes the portfolio of stocks it feels best along with the weights of each stock. That portfolio gets updated into the ETFs during the monthly rebalancing. While the model, constraints, and choice of when to rebalance are all made by humans, the AI is the final decision maker for portfolio.

While it is anticipated the Adviser, Exchange Traded Concepts LLC, will purchase and sell securities based on recommendations of QRAFT AI, the Adviser has full discretion over investment decisions for the Fund.

More than an ETF Provider

While the ETFs provide visibility and transparency to what the proprietary AI can do, it is not the only tool in the toolbox for Qraft. Qraft offers institutional investors an AI order execution product called AXE, which has been in use since March 2020.  The API[9] and a B2B Robo advisory[10] are also products currently available. Overall, Qraft has over 1.5B in assets under AI. These products and services have allowed Qraft to show what AI can do and have attracted the gaze of both retail investors and larger financial institutions. From ETFs to customized solutions, Qraft sees a future of democratizing opportunities in the market and more efficient asset management.

[1] Drawdowns - A drawdown is a peak-to-trough decline during a specific period for an investment, trading account, or fund. A drawdown is usually quoted as the percentage between the peak and the subsequent trough.

[2] Kirin API - Developed by Qraft’s data scientists, integrates multiple vendors to provide both macroeconomic and company fundamentals with the correct point-in-time data.

[3] Alpha Factory – Qraft’s proprietary strategy extraction system composed of two models, Factor Factory and Strategy Factory.

[4] Factor Factory - Factor Factory – Qraft’s core AI technology that automatically finds factors that could bring excess returns. Factor Factory can produce at least 10 factors per day without any human intervention.

[5] Strategy Factory - Strategy factory extracts the investment strategy with a nonlinear asset price model by nonlinearly combining factors extracted automatically from the factor factory.

[6] P/B - Price to book ratio, compares a company’s current market value to its book value

[7] Market Capitalization - Market capitalization refers to the total dollar market value of a company's outstanding shares of stock.

[8] Alpha – Alpha is a measure of the active return on an investment, the performance of that investment compared with a suitable market index.

[9] API - An application programming interface (API) is a set of programming code that queries data, parses responses, and sends instructions between one software platform and another. APIs are used extensively in providing data services across a range of fields and contexts.

[10] Robo Advisor - Robo-advisors (also spelled robo-adviser or rob advisor) are digital platforms that provide automated, algorithm-driven financial planning services with little to no human supervision.