To make Wealthtender free for readers, we earn money from advertisers, including financial professionals and firms that pay to be featured. This creates a conflict of interest when we favor their promotion over others. Read our editorial policy and terms of service to learn more. Wealthtender is not a client of these financial services providers.
➡️ Find a Local Advisor | 🎯 Find a Specialist Advisor
[The “active” versus “passive” investment debate has raged for decades fueled by performance comparison reports like SPIVA, Alpha Architects, and many others chronicling how a large cross-section of equity investment managers fail to beat their benchmarks. The funny thing is that there are a good number of reports like Enterprising Investor, the seminal Cohen, Polk and Silli Best Ideas research study, and a recent Georgia Institute of Technology academic paper that indicate active managers can add a great deal of value and have proven stock picking skills. The latter report goes on to show that active managers are challenged by traditional portfolio management and investment structures that dilute their potential returns.
The answer to addressing this active manager paradox may well lie in strategically redesigning the structure and active management process around active manager’s strengths. This article illustrates how innovation eventually finds a way to address this kind of conundrum. In this case, in the form of a new asset management firm that uses data science and machine learning technology to harness the human strengths inherent in active managers abilities to add alpha and the due diligence research process used in manager selection and portfolio construction.
We reached out to co-Founders Paul Ahern and Jeff Seiple, along with Stephen Beinhacker, Chief Investment Officer of NextFolio – an asset management firm pioneering the application of Ensemble Methods to active portfolio management. Ensemble Methods are a branch of machine learning used to improve forecasting accuracy in complex challenges and represents cutting edge data science and advanced machine learning. While this technology has had a revolutionary impact on many industries with over 250,000 uses today, including facial recognition, self-driving cars, MRI tumor detection, and hurricane tracking, they have only recently been applied to investment management.
As developed by NextFolio, the key innovations of Ensemble Active Management (EAM) are the ability to access daily fund holdings data to identify the real-time insights and high-conviction investment ideas of 10-12 carefully chosen top fund managers through replication technology, and the use of Ensemble Methods to combine these insights in a way that enhances the predictive power over individual managers. They assert that this “ensemble modeling” approach results in reducing the over-diversification of portfolios, concentrating Active Share, delivering excess alpha over benchmarks, and tackling the key challenges of traditional active management—its inconsistent performance, high fees, and especially, its dependence on a single manager’s insights.
NextFolio offers a series of turnkey SMA strategies across all nine domestic equity style boxes for seamless integration into any firm’s asset allocation, as well as offering the ensemble investment technology and methodology on a white label or sub advisory basis for financial intermediaries and wealth advisors’ new product development.
I asked them questions to further explore and understand the nature of this new investment approach powered by machine learning whose endpoint is not artificial intelligence but maximizing human intelligence and the implications of applying this ensemble methodology as a new paradigm for active equity management.]
Hortz: Why did you embark on this endeavor of reimagining active equity management? What were you trying to address and why did you feel Ensemble Methods was the solution?
Ahern: There has been a steady stream of assets under management (AUM) going from active to passive management with $3.4 trillion of outflows from active funds since 2016 compared with $3 trillion of passive fund inflows. The reason for that is that active management has really struggled to consistently perform relative to its underlying benchmarks. Particularly, in the mutual fund industry which has been in net redemptions.
On the one hand, mutual fund structure and portfolio construction efforts can potentially dilute fund manager’s performance. Particularly in some of the larger asset classes, only about half of the mutual funds are invested in the active, high conviction, best stock picks of the equity manager. The other half of mutual fund portfolios may be invested in their benchmark or their beta position as ballast to stabilize the portfolio and as a tool for risk management. But if you think about this conceptually, if you take a large portion of your portfolio and you move it off the table from its ability to perform to beat the benchmark, and then if you add in fees and expenses, you can see that whatever returns you would produce from your best investment ideas would be diluted.
On the other hand, another factor is the way investors want managers to diversify risk for them at the individual fund level and for managers to build portfolios that reduce career risk and raise the ability to have significant assets under management. These all add to a structural disadvantage that dilutes the talent and skillsets of the underlying managers. That is why NextFolio was created: to recalibrate active management and amplify the stock picking talent and alpha creation of mutual fund managers using data science. In essence, NextFolio EAM puts the inherent talent of fund managers to work in a more efficient and accurate portfolio structure.
Seiple: In pursuit of a solution, Paul and I learned about and intensively studied Ensemble Methods and its successful application to other industries. We saw that it could be an application of a proven branch of mathematics into finance in a way that was not really available until very recently. Most importantly, we realized that Ensemble Methods are not just some artificial intelligence “black box” that you do not understand how it got to its results or logic. It is just applying advanced mathematics and processing power in a way that kind of marries active fundamental investing with systematic quantitative investing to produce a way to capture the alpha that active managers actually have skill at attaining.
We worked with our technology partner, Turing Technology, for four years by taking a deep dive and gaining a thorough understanding of their investment technology based on Ensemble Methods – particularly on the application of advanced machine learning techniques and mutual fund replication. That first-hand immersion helped us build our strategies to focus on alpha generation and we were happy with the results we were achieving. So, in the beginning of 2024, Paul and I decided to launch an asset management company and then spent the year building the firm and our professional advisory board.
Hortz: Can you walk us through how you are applying Ensemble Methods to asset management? What are the components of the differentiated returns that EAM brings to the table?
Seiple: There are two components to performance. One is from fund replication which is the ability to understand the daily holdings of mutual funds on a real-time basis and their relative weights to their benchmarks. In other words, we know where the managers are overweighting positions or placing their best bets. This is a unique technology to determine on a real-time basis what managers are doing – where they are overweighting versus benchmark.
That fact would be dramatic enough, but then the technology can also look across at a cross-section of top managers in an asset class or sector to look for high conviction overweights that they themselves individually do not even know exists. The tech does not extract just the wisdom of a single manager. It is the collective wisdom of a team of 10-15 top managers all competing against the same benchmark, all trying to solve the same exact problem at the same time, each from a unique perspective.
That is what the core of the technology represents a unique capability. It is an information and analytic advantage that it is not readily available elsewhere – to understand where top institutional managers, the best minds of the business, are placing their high conviction overweights relative to their benchmarks.
Hortz: What is the other main component of differentiated returns from Ensemble Asset Management?
Beinhacker: Another key component of performance is to pick the right cross-section of top managers per asset class. It is not sufficient to just have 10 random managers or the top 10 managers by numerical performance in each investment area. That is the other side of the coin of how ensemble methods work.
If you stop and think about it, a mutual fund is a predictive model. That is because each fund has a benchmark and the manager tries to build a portfolio to outperform that benchmark, and in so doing, is making a forecast, in essence, to beat that benchmark. Once you realize that mutual fund portfolios represent a forecast, that is what opens the door to the use of Ensemble Methods.
But to maximize the effectiveness of this technology, each manager needs to represent a successful but different predictive model. You want your predictive models to be independent of one another, using different research teams, trying to beat their shared benchmark from their own unique perspectives and formulations. That is one of the tenets of ensemble methods that you need to feed into the system 10 – 15 different successful models all trying to solve the same problem but in different ways.
Applying Ensemble Methods to the collective insights of the funds, betters the predictive accuracy of any single fund. We are using data science and replication technology to be able to glean where a cross-section of managers are placing their emphasis and enhancing their human insights. We have just peeled away the mutual fund wrapper to get to the core of what the managers are really good at and we leave everything else out that dilutes performance.
As early pioneers in bringing this new technology and methodology to investment management, we can deliver this differentiated equity management strategy to clients.
Hortz: How do you go about selecting the right mix of successful mutual fund and ETF managers to feed into this EM technology?
Beinhacker: As mentioned previously, for ensemble technology to really work, you need to carefully select a cross-section of successful managers in the same investment space that use different methods and approaches from each other. You cannot choose managers trying to fish in the same corner of the investment pond. You need different methodologies, ten to fifteen different variations of investing in that asset class.
We certainly want to avoid inferior managers, but it is not a requirement that the ten to fifteen funds all be top quartile. We do not need that because the ensemble methods are going to enhance the underlying returns. We just need a good cross-section of differentiated managers dedicated to beating the same benchmark. The most critical factor is knowing that the funds you are going to use are going to consistently adhere to their style and philosophy because when they drift, that undermines the strength of their overall signals.
Number two, stability of the team is always important because that validates the track record and the historical application of the philosophy. Monitoring that is critical as investment management is still a people-led process.
The third factor is really knowing how the managers and the funds that you are selecting are going to fit together. Because the EAM process looks for those high conviction ideas across fund models, how you comprise the portfolio is really important. So really paying attention to how that construction and mix is going to work is the additional step that really makes for successful fund selection and capture of ensemble alpha.
My professional experience as the manager of a 20-person investment manager research team at SEI, where we built multi-manager strategies or fit them into an asset allocation program, has been very helpful for this aspect of EAM. Here at NextFolio, the evaluation framework is similar.
The lens we evaluate them through is to look through and think about how this basket, this composition of investment management decision-making philosophies and processes is going to work. You need to move from just fund selection by itself to portfolio construction and ensemble management. If you are just doing fund selection, you are going to miss the bigger picture and not capture the full potential the way you would through a single manager selection process. I think that is a key differentiating factor for the EAM process that we bring to the table.
Hortz: How would you suggest that advisors and asset allocators explain and deploy Ensemble Active Management with their clients? How do work you with financial professionals?
Ahern: In a nutshell, EAM has a very simple elevator pitch in appeal. It is access to the real-time, daily, high-conviction best stock picks of elite institutional mutual fund managers. That is very appealing and easy to understand. There are lots of layers that go below this but the beauty is this has a quantitative backbone with machine learning capabilities. This is not just a good story. Ensemble methods are the perfect marriage of data science leveraging human intelligence, proving that investment managers do make a difference.
Seiple: Our current target market is for investors through their advisors – financial advisors, wealth managers, RIAs, bank trust divisions, family offices, brokerage firms – that have clients that want to use active investment strategies to achieve alpha over passive strategies. Currently our EAM products are available as SMAs or CITs. We want them to know that there is a way to get at the alpha that top managers generate by peeling off the wrapper of the mutual fund and using technology and data science to bring that to their clients.
Beinhacker: We can also work with an asset management firm as an outsource research team. NextFolio strategies are designed for seamless integration and can be white-labeled for use in mutual funds, ETFs, Separately Managed Accounts, and Collective Investment Trusts, enhancing their investment offerings effortlessly. We want to be an active and strategic partner to the industry to expand Ensemble Asset Management as a new enhanced option for active equity asset management.
This article was originally published here and is republished on Wealthtender with permission.
About the Author
Bill Hortz
Founder Institute for Innovation Development
To make Wealthtender free for readers, we earn money from advertisers, including financial professionals and firms that pay to be featured. This creates a conflict of interest when we favor their promotion over others. Read our editorial policy and terms of service to learn more. Wealthtender is not a client of these financial services providers.
➡️ Find a Local Advisor | 🎯 Find a Specialist Advisor