What we do and how we do it.
We’ve been writing this blog for over a year. You probably know by now that Smartleaf provides a system for automating customization and tax management, but we have never actually explained how our system works or how we’re different from other solutions. We thought we’d do so now.
We believe that the basic approach we take to implementing automated rebalancing is pretty universal. Not in the sense that all rebalancing systems are built the way we are (they’re not). But in the sense that any system that automates high levels of customization and tax management pretty much has to do things the same way. In other words, we don’t think that a description of our approach is really about us. It’s about what it takes to automate rebalancing of complex portfolios, regardless of who does it. Call it a guide to architecting an automated rebalancing system for complex portfolios.
And why is automated rebalancing interesting? Because highly customized, tax optimized-wealth portfolio management is becoming “table stakes,” meaning you won’t be paid for it, but you’ll be at a competitive disadvantage if you don’t have it. Implementing it will change your organization and your client experience (in particular, it pushes advisors to focus on financial planning and acting as a “life coach”).
(We’ve previously written about the impact of automated rebalancing on the wealth management industry here: What is Rebalancing Automation?, Rebalancing Automation Will Change Your Job, Can Compliance Be Automated?, Stop. It's Time to Rethink Rebalancing, Automated Rebalancing: Why Investors Care.)
Our approach can be usefully (if inartfully) described as Collaborative, Holistic, Optimized Models-based, Parameterized, Programmatic and Non-sleeved. Or CHOMPPN, for short. (We know. Our marketing department is working on it.) More precisely, the key architectural elements of our system (and, we think, any other system capable of automating the rebalancing of complex portfolios) are:
Let’s walk through these one at a time.
Traditionally, portfolio management has been something of a lone-wolf activity, with each advisor taking responsibility for all aspects of managing their accounts.
In an automated system, portfolio management is collaborative. It’s the joint effort of three groups: the investment policy committee, the advisor and the overlay manager. These terms describe roles, not individual people. The roles can all be assumed by one person or spread among three groups, and each role can itself be subdivided.)
Making portfolio management a collaborative effort in this way enables each person or group to focus on what they do best. Critically, it allows advisors to delegate routine tasks of portfolio rebalancing and spend more time with clients and prospects. The result is simultaneously greater efficiency and the opportunity to provide clients with higher levels of customization and tax management.
Most portfolio management systems are “action forward.” This means that their job is to help managers implement a series of discrete actions, such as a tactical asset allocation, a security swap in a model, a cash withdrawal, etc. With each rebalance, the basic idea for most systems is to implement these actions and do as little else as possible. These systems can also support a “general rebalance,” but usually not in a way that is particularly tax- or expense-efficient unless there is manual intervention.
While the concept of an “action forward” workflow sounds sensible (even exciting), it leads to undesirable outcomes for both advisors and their clients. These workflows tend to produce an environment wherein each day becomes, variously, loss harvesting day, drift compliance day, model change day, etc. That is, the advisor ends up reactively addressing (across as much of their book as they can in one day) a single aspect of a particular policy for investment and trading — while trying to make sure their actions are not in conflict with other aspects of the same policy. There are simply not enough hours in the day or days in the year to make this process work for every investor account.
With an automated system, all of these rebalancing “triggers” (tactical asset allocation change, ticker swap, cash-out request, drift, loss harvesting, etc.) are folded into one. In every analysis, the system is looking to do all these things, the moment they’re needed or appropriate. Every time. Every day.
This approach enables advisors and wealth management firms to implement every aspect of their programs proactively. Instead of “model change day,” “loss harvesting day,” “model 7 rebalancing day,” every day is simply “make-the-portfolio-better day.”
The result of this approach is lower drift (difference of return between the portfolio and its target) and, simultaneously, lower taxes. This is possible because the system takes actions when they’re appropriate — sometimes earlier than conventional approaches (because the system analyses daily), sometimes later (e.g., delaying a sale until short-term gains go long-term).
Could you create an automated system that wasn’t holistic or optimized? We doubt it. At least not with equal levels of tax management or risk control. Rebalancing non-holistically (implementing single actions one-at-a-time) needlessly forgoes opportunities to improve taxes, risk or both. And trying to replicate the logic of an optimizer with a series of rules quickly becomes overwhelming; the rules just get too complicated to handle all the possible combinations of trade-off situations you can run into.
“Models-based intellectual capital” is another way of saying “parameterized inputs for holdings, asset allocation and trade information.” It’s a cousin of parameterized customization, which we describe below, and it’s necessary to support automation. The alternative is to have buy lists and asset class allocation ranges — and manual stock selection and trading.
Each portfolio in our system follows a blend of one or more models (weighted lists of securities). If a portfolio manager or firm wants to swap stocks across multiple portfolios, the primary mechanism for doing so is to adjust the weights of the securities in a model. Tactical asset allocations are expressed by changing asset class weights in what we call a “Target Template.” Target Templates also include firm-set limits on asset allocation modification and firm-set limits on asset-class drift.
For many advisors, “customization” means one-off, trade-by-trade decisions tailored to the needs of each client.
That’s not how an automated system does things. An automated system stores customization requirements and preferences as parameters that describe how each portfolio should be managed. This includes tax management preferences, asset allocation modification, product substitutions, SRI screens, etc.
These preferences and constraints are then automatically incorporated into every analysis.
With this approach, there are no extra steps required to implement tax management or customization. And the incremental cost of tax management and customization is effectively reduced to zero, which leads to an important and somewhat counterintuitive effect. When you make something free, people consume more of it. Customization and tax management are no different. When you reduce its costs to zero, advisors consume more of it. That is, when firms parameterize customization and fold these parameters into every analysis, it leads to a dramatic increase in the level of customization and tax management of client portfolios.
And, not incidentally, it also aids with compliance. There are no client requests on Post-it notes getting lost. Once a request is in the system, it will be included in every analysis. Portfolios need never be out of compliance more than one business day.
An automated system analyzes every portfolio daily and generates trades that follow the joint instructions (i.e. parameters) of investors, clients and the investment policy committee. Firms can then apply a filtering system to select the portfolios they want to trade.
That’s not that unusual in and of itself. What makes this different is that firms can apply the same workflow, the same set of filters, every day.
Having the same workflow every day does not mean trading all accounts every day. It simply means having uniform standards for what is traded. For most of our users, there are two basic reasons to trade an account:
And this basic two-step workflow is applied every day to every account. But individual firms can and will customize this workflow to suit the details of each of their investment programs.
Traditional rebalancing systems are designed to facilitate implementing a series of requests, like a stock swap. What gets done on any given day will depend on the requests the advisor chooses to implement and the order they’re executed.
In contrast, a programmatic approach is designed not to implement specific trades, but to implement specific programs and policies.
Many systems divide holdings into sleeves (also called partitions or sub-accounts), with one sleeve per model. We do not, because sleeves are expensive to operate and interfere with optimal risk and tax management. You can read about our thinking on sleeves in several previous posts: The Ultimate Guide to Sleeves, Part I and Part II, A Guide to Choosing a Rebalancing System, and Don’t Treat Individual Investors like Institutions.
So, there you have it: CHOMPPN: Collaborative, Holistic, Optimized, Models-based, Parameterized, Programmatic and Non-sleeved rebalancing. Easy. (Our marketing department is still working on it.)
For more on this topic, check out What is Rebalancing Automation?