Satisfaction with status quo has never been a defining paradigm for the founders of Turing. Their level of creative expertise and willingness to push into previously uncharted territory has allowed them to devise breakthrough solutions that were, candidly, considered unachievable by established 'experts' at the time.

An example of these breakthrough solutions would be the Hercules Fund Replication System (see below), which is able to replicate the daily holdings and portfolio weights of actively managed funds, on a real-time basis, at a level of accuracy beyond what most 'experts' considered viable.

Turing Technology is thus a repository of breakthrough technologies, powerful derivative applications of the technologies (e.g., the Hercules Database), and a portfolio of Intellectual Property that is poised to reshape active management on multiple fronts.

Turing's core capabilities and technologies enable investment firms, institutional investors, and registered investment advisors to quickly construct and deliver to their clients a number of cutting-edge investment solutions, and include the following:

HERCULES DATABASE

Over the past several years Turing has built out a first-of-its-kind database capturing up to a decade's worth of real-time, daily holdings and portfolio weights of actively managed mutual funds. The Hercules Database currently reflects information from more than $3 trillion in fund assets under management, reflecting data covering 70+ fund families and hundreds of funds. The Database is the repository of the output of the Hercules Fund Replication System (see below).

This daily, holdings-based data allows Turing to look inside mutual funds at the 'DNA' level, and has allowed Turing to gain unprecedented insight into the cause-and-effect of fund managers' decision-making. For example, the first set of findings to be published by Turing based on analysis of the Database identified a 'genetic defect' that has hobbled active managers in their efforts to outperform index funds and ETFs. As explained in the research, this structural flaw has had more of a negative impact on active fund returns than fees.

In addition to supporting breakthrough research, this enhanced data source is 1) critical to creating a 'roadmap' forward for designing superior investment solutions for our clients, 2) the raw materials enabling Turing to create next generation investment solutions such as Ensemble Active Management Portfolios, and 3) the basis of an emerging data research business.

The Database already represents more than half of the actively managed US equity fund assets in the country, and there is an active effort to continue to expand it. Turing's goal is to eventually have within the database the daily holdings and portfolio weights for at least 90% of the actively managed US equity fund assets.

HERCULES FUND REPLICATION SYSTEM

The idea of accurately replicating fund holdings has been an aspirational goal for the industry for decades, but always seemingly falling short. Until now. The early insights for the Hercules System date back nearly two decades, and reflect non-conventional ideas that Vadim Fishman had while working primarily as a consultant into the biotech industry.

As devised today, the Hercules System is a multi‐component technology platform that relies heavily on Machine Learning-based algorithms. And for clarity, it is not regression based (regression modeling is a fundamentally flawed approach for the complexity required to achieve accurate fund replication).

The Hercules System uses multiple data feeds, all of which are publicly available, and then each data feed is subjected to Data Quality modules reflective of the fact that many of the data sources are subject to significant data error rates (such as the periodic and delayed fund data available through EDGAR filings). A rough depiction of the inputs into the System is as follows:

Core Replication Algorithms

To provide a quick snapshot of its accuracy, the replicated funds generated by the Hercules System achieve a correlation to the actual funds of more than 99.7%, and run at a tracking error of less than 1% (and these results are in spite of the fact that the actual funds have to accommodate cash flow into and from the funds, and that cannot be replicated by the Hercules System).

EAM INTEGRATION TECHNOLOGY

Turing's ability to build Ensemble Active Management Portfolios on behalf of our clients relies upon two different technology packages. The first, and most important, is the Hercules Fund Replication System (see above). The second is the EAM Integration Technology, which is used to integrate the High Conviction Overweight positions (i.e., Best Ideas) from a dozen or so underlying funds into a best-of-the-best EAM Portfolio based on the highest consensus agreement of the underlying fund managers.

The mathematical technique that underpins this technology is known as Ensemble Methods, which is a time-tested sub-component of Artificial Intelligence and Machine Learning, dating back decades. Ensemble Methods is used by countless industries to improve the accuracy of predictive algorithms or predictive engines by mathematically identifying areas of agreement across multiple predictive engines.

To learn more about Ensemble Methods, see Turing's White Paper “ENSEMBLE ACTIVE MANAGEMENT: The Next Evolution in Active Investment Management”.

Turing has developed the EAM Integration Technology as a proprietary version of Ensemble Methods used to construct the final set of securities and weights reflective of the EAM Portfolio design of its clients. It reflects a continual improvement process based on on-going insights on subtle drivers of improved returns, and is a key component of Turing's capabilities for the EAM Portfolio market.

REGIME-ID™ AND SECURITY-IQ™ (DOWNSIDE VOLATILITY MANAGEMENT TECHNOLOGIES)

Many of the most important mathematical applications for designing asset allocation portfolios date back to the Nobel winning concepts of Harry Markowitz, who is considered the father of Modern Portfolio Theory. Interestingly, his seminal paper “Portfolio Selection” was published in 1952 – more than 60 years ago and well before the advent of the modern computer.

One of the key simplifying assumptions embedded into his work was the concept that markets can be described by a single set of historical risk and return metrics (or in mathematical terminology “stationary”). This paradigm has been a key driver of the current template for investing characterized by a buy-and-hold model as the optimal investor approach. But this also means that investors are periodically forced to suffer through significant bear markets.

In recent years, many leading academics and investment quants have come to understand that the market, rather than continuous, is instead made up of a collection of many phases, or 'regimes', where each regime is reflective of different risk and return dynamics. The challenge has been, however, to be able to mathematically identify the transition from one regime to the next on a real-time basis. Turing's RegimeID™ technology is believed to be the industry's first means of achieving this identification.

Turing is now able to mathematically identify, on a real-time basis, a number of Risk Regimes for major markets, where each Risk Regime has dramatically different risk-reward dynamics. For example, for equity markets Turing's RegimeID technology can identify:

  1. Standard Risk Regimes: This market type is characterized as 'healthy', where a buy-and-hold strategy rewards investors with annualized returns of approximately 9.4%1, and drawdown events are infrequent and relatively modest.
  2. High Risk Regimes: In this market phase, losses are the expected norm and thus buy-and-hold strategies are penalized. This Regime is also the sole domain of disastrous market declines, with maximum losses two-times as great as seen in Standard Risk Regimes1.
  3. Transitional Risk Regimes: These are generally described as unstable periods, where the market is wobbling between Standard and High Risk Regimes. In general, these market phases will reward buy-and-hold investors, but the potential for a rapid slide into a High Risk Regime means that investment policies need to reflect a more defensive bias.

Visually, a two-phase market with transition zones is analogous to a liquid transitioning to a gas as heat increases. A key concept here is that liquids and gasses behave very differently. For example, water will efficiently conduct electricity, while steam will not.

Increasing Level Of Risk

The power of this technology is that it allows a rationale shift of investment policy based on the different expected market returns of each Risk Regime. For example, in a Standard Risk Regime, where expected annual returns are more than 9% for the US equity markets and maximum losses are modest, it would suggest a capital policy of 'maximum' allocation. Alternatively, in a High Risk Regime, where expected returns are negative and the potential for devasting loss levels are high, a rationale investment policy would be to eliminate, or at least minimize, capital exposure.

This binary capital allocation model, based on the probability of outcomes, is remarkably similar to how card counting in Blackjack uses an understanding of statistics and probabilities to generate significant financial gains.

The impact of this technology on the industry at large can be significant.

In practice, this technology enables a fundamentally differentiated, and enhanced, means of building asset allocation models. This enhanced approach would not be in violation of the core principles outlined by Professor Markowitz more than half a century ago, but would simply function to deliver improved allocation designs based on improved risk and return metrics as inputs to his formulas. Rather than the static model primarily in use today, it opens the door to dynamic asset allocation models with higher overall expected risk-adjusted returns. This approach would not be market timing, but rather a rationale, and fact-based implementation of an investment policy based on the changing expected risk-adjusted returns of the market.

As a complement to RegimeID, Turing's SecurityIQ technology uses the output of the RegimeID technology to translate the information into specific assessment of the attractiveness of an individual security. Turing's delivered downside risk management solutions, reflecting clients' design parameters, incorporate both RegimeID and SecurityIQ as part of the final technology delivery platform.

1 Based on Turing research.