AI Flagship Hedge Fund
Building Quant Hedge Funds through AI
Since its inception in 1988, Renaissance Technologies’ Medallion fund produced non-precedential returns of 66 percent CAGR, playing the most influential role in widely promoting its legend.
The Medallion fund is known to employ various types of strategies such as managed futures and statistical arbitrage, on top of high-frequency trading such as market making. Due to numerous quant strategies with scaling limits which are not externally disclosed, it is the fund of dreams where only Renaissance Technologies’ employees can invest in.
Outside investors (unavailable for retail / institutional only) are given two choices to let their investment assets managed by Renaissance Technologies. The first option being the Renaissance Institutional Equities Fund (a.k.a. RIEF) undertaking a 175-75 strategy (175% Long, 75% Short), followed by the Renaissance Institutional Diversified Alpha Fund (a.k.a. RIDA) employing a market-neutral strategy seeking absolute returns. Currently, each fund has an AUM of about USD 40 billion, managed by undertaking a low frequency strategy based on large-cap stocks which makes the fund manageable even with large amounts of money.
Large institutions driven by the Medallion Fund’s reputation are investing large amount of capital in these two funds, which, despite their recent sluggish performance, have been performing considerably well since inception to date.
RIEF and RIDA’s average annual returns are 22.48% and 5.79% respectively, largely outperforming their benchmark, as well as their Sharpe ratio consistently outperforming that of the benchmark.
Renaissance Technologies as well as other hedge funds’ flagships such as Two Sigma’s Compass Fund and D.E. Shaw & Co.’s Composite Fund mainly employ 175-75 or 150-50 equity long-short strategies, consistently maintaining good results.
The 175-75 strategy’s advantage is the ability to leverage on alpha while following market beta. When simultaneously taking positions of 175% long and 75% short, the entire portfolio beta is 100%, identical to the market. Nevertheless, if outperforming or underperforming stocks can be accurately classified through strategic research, alpha can be leveraged while maintaining market-level beta but with a higher Sharpe ratio. As such, entities such as large global pension funds and insurance companies are allocating a large percentage of their capital to increase their risk-return profile.
Such flagship hedge funds are highly compensated that matches to their good performance. Most of them receive a 2% annual fixed fee from managed assets and a 20% performance fee. For D.E.Shaw & Co., fees have been raised in 2019 to 3% from assets and 30% from profits as its flagship Composite Fund continued to gain in popularity. This evidences the growing demand in hedge funds that can steadily create alpha.
Hedge funds using the 150-50 strategy can also be replicated with AI. QRFT, Qraft Technologies’ NYSE-listed AI ETF managed based on U.S. large-cap stocks, is an ETF that aims to improve the S&P500 index through AI. One year after its listing, QRFT outperformed the S&P500 index by a wide margin by boasting post-inception total returns of 31.85% compared to S&P500’s 16.34%. (As of 2020-07-29/total return/after fee)
QRFT’s actual performance since inceptionIn the case of QRFT, the investment universe is largely the same as that of S&P500, U.S. large-cap stocks. Nonetheless, the difference lies in management where it is the AI that judges the weight of each stock and rebalances on a monthly basis. QRFT's trackrecord implies that it is doing a good job in guessing which stocks of the S&P500 index will rise (Long) or fall (Short).
TIP) Refer to AI Asset Management Report for more details about QRFT and AI Asset Management
Currently, QRFT is not taking a short position, but QRFT’s AI engine is a structure that generates results with a long-short basis. However, as of present, stocks with short weightage are treated as 0. Similar to the flagships of large hedge funds, the AI engine can adopt a 150-50 structure instead of treating them as 0. When leveraging the long part of the engine results by 150%, taking the short part as 100% and deducting borrowing costs, the results are very interesting.
RIEF vs. QRAFT150/100Since RIEF pursues beta of 0.4, beta has been set to about 0.5 with 150 long / 100 short for a similar fit. In order to avoid overfitting to the greatest extent as possible, all back-tests have been run on a test dataset instead of a training dataset.
Simple comparison may be difficult, but both CAGR and Sharpe Ratio provide results that Qraft 150/50 long-short model is performing better. Of course, RIEF’s performance is an actual track record compared to Qraft 150/100’s live track record since May 2019, hence the necessity to prove not only the validity of forward testing but also the actual performance. If it does, a huge institutional demand is expected.
Qraft 175/75 and Qraft 150/50, which are versions that leveraged alpha more aggressively, show even more impressive results.
Qraft Technologies will soon list a U.S. large-cap 150-50 model in the form of an AI-driven ETF in NYSE at low fees. (If track records of a couple of years are to be successfully accumulated, it is possible to switch to a private company model with higher fees)
It is expected that there will be a lot of concerns with regards to Qraft Technologies challenging the legendary Renaissance Technologies through an AI system. Nevertheless, the level of alpha produced via AI’s data processing is sufficient enough to challenge not the Medallion Fund which is an aggregation of numerous high frequency alpha, but the long-short based RIEF that has been constructed more slowly to manage capital over USD 100 billion. Actually, even after beta correction on RIEF (pursuing beta of 0.4), QRFT has never been overtaken on a post-inception basis.
In order to list an ETF of a 150-50 structure, securities lending that requires prime brokerage services in the U.S. is necessary, under which a minimum initial seeding capital of USD 25 million will be called for. Qraft has completed its product development phase and is currently on the stage of financing seeding capital. Moving forward, it will try to prove that AI-based management methods can sufficiently create alpha rivalling to the level of large hedge funds’ flagships such as RIEF / Compass Fund / Composite, by listing 150-50 equity long-short hedge fund ETFs on NYSE similar to that of large hedge funds’ flagships.
About Qraft Technologies
Qraft Technologies revolutionizes the asset management industry’s inefficiency using AI technology. Our aim is to provide high alpha at a low cost, by solving inefficiency within each stage of the asset management industry via technology, from data processing to alpha research and portfolio order execution. Pre-processing of complex financial data is automated then accelerated through parallel computing, which allows to automatically search for alpha factors by using AutoML technology in a completed simulation environment. By using alpha factors that have been discovered as such, the Deep Asset Pricing Model based on deep learning is built according to the fund universe conforming to the determined fund concept. The final portfolio drawn via the model aims to execute effective orders through AXE, a reinforcement learning-based order execution engine. The NYSE-listed Qraft Technologies’ ETF lineup produced through the above system is a 100% unmanned AI system. The lineup is showing the best performance among similar ETFs by outperforming the benchmark index (S&P500, S&P500 momentum index) by more than 10% during its first year after listing.
QRAFT webpage: http://www.qraftec.com
Subscribe to QRAFT’s newsletter: http://www.qraftec.com/newsletter
Disclaimer
*Past performance is no guarantee of future results.
*This material was prepared for informational purposes only and cannot be used to encouraging sales of financial investment products such as funds.
*This document contains contents that are patented or pending to be patented by Qraft Technologies.
