How automated is automated rebalancing?
The idea of automated rebalancing seems simple enough — it’s rebalancing, only automated! But once you start looking more closely, it turns out it’s more complicated, and more interesting, than just watching a computer trade. There’s no such thing as the optimal approach to “automated rebalancing.” There isn’t one type of investor. There isn’t one type of firm. And therefore it will never be good enough to have one type of automated rebalancing. We’d like to walk through what automated rebalancing means, and some of the choices you’ll face should you adopt automated rebalancing technology.
Automated Rebalancing is Different Than Rebalancing Automation
First things first. By the concept of “automated” rebalancing, we actually do mean rebalancing where the trades are auto-generated and actionable. “Robo rebalancing,” if you will. This is more than just “rebalancing automation,” which includes tools that help you do the things you’ve always done, just more efficiently. The difference between “rebalancing automation” and “automated rebalancing” is roughly the same as the difference between a power screwdriver and a robot. The former helps. The latter is a reinvention of the process and makes possible fundamental changes in efficiency, scalability and value proposition.
How Automated is Automated Rebalancing?
In theory, automated rebalancing is, well, fully automated, all the time. In practice, it doesn’t always work that way. Sometimes, the robot needs help. But not for the reasons most people think. The stuff that humans can find challenging—tax management, substitutions, transitions, social screens, trade-offs, asset class customization, product customization—can be automated relatively easily.
So what’s hard? The most common problem is bad data—unknown securities, late-processing of corporate actions, file transfer failures, etc. If you don’t have good data, the quality of your analytics won’t matter. This is the classic case of “garbage-in / garbage-out.” As a result, any automated rebalancing workflow needs to include a process for dealing with data integrity issues. Usually this means suspending the affected accounts until the data issues are resolved. In extreme circumstances (e.g., the data is bad and the client has requested cash), it will require generating trades manually, the old-fashioned way.
There will also be special cases where a “manual override” is necessary (and, as noted above, tax management, transition or handling complex combinations of constraints are not special cases—automated rebalancing system can handle those just fine). A special case would be something like: “The client called and said they’re transferring in $1mm next week. Hold off trading until then” or “Trading of XYZ stock has been suspended. Let’s not trade any portfolio buying or selling XYZ.” Temporary suspensions of this sort are the most common overrides we see.
The Basic Automated Rebalancing Workflow
So, when all the dust settles, what does rebalancing a book of business look like with automated rebalancing? We see a basic three-step process:
- Suspend Accounts with Bad Data
- Suspend trading for accounts with unknown securities, incorrectly implemented corporate actions or some other data quality issue.
- Manually trade accounts with bad data AND a high-priority mandated trade, such as a cash-out request.
- Trade Accounts with “Mandated” Trades
Export (in batch) system-generated proposed trades for accounts where trading is needed to satisfy client or firm mandates such as “withdraw cash,” “never own Tobacco,” “min/max asset class drift,” etc.
- Trade Accounts with High “Cost/Benefit Scores”
Export (in batch) system-generated proposed trades for accounts where the system assigns the trades a high “cost/benefit score.” Cost/benefit scoring is a different way to think about trading. You take a group of trades and numerically score the benefits (lower drift, tax loss harvesting, higher rated securities) minus the costs (transaction costs, taxes). An automated rebalancing system can, every day, offer up the trades with the highest possible cost/benefit score (that’s what optimization is all about). Once that’s done, you simply trade all accounts with a cost/benefit score above some threshold. The higher the threshold, the less frequently you trade.
There are multiple variations of this basic workflow that differ, not in their efficiency, but in the details of when accounts are traded. For example, some firms will trade only high-priority mandates (such as cash-out requests) in the second step, leaving other mandates to be addressed only when the accounts are traded in the normal course of events under step three. Other firms will have different cost/benefit score thresholds for different types of accounts (e.g., lower thresholds for accounts that haven’t traded in a while).
Here are some of the different preferences we see for how firms rebalance. All are reasonable in their own way:
- Dispersion vs. Tax Efficiency
Wealth managers aim for low dispersion of returns among clients with similar risk objectives. They also aim for tax efficiency. There’s frequently a tradeoff to be made between these two goals. Which is more valuable and by how much? There’s no right answer, and we see firms making different choices.
- Trading Frequency
Firms we work with have different preferences about how often they trade. It’s not really a matter of reducing transaction costs (since few subject their clients to per-ticket commissions). There’s some interest in avoiding perceived churn, but it’s more just aesthetics—they don’t want too many trading days.
Side note: the obvious way to control trading frequency is to just trade at fixed intervals, which is easy to do with an automated rebalancing system, but this ends up being suboptimal for the client—you can end up both trading too little and too much. It’s better for the client to trade when it’s beneficial to the portfolio to do so, which is exactly what you get with the “cost/benefit” threshold approach we described earlier. And it still gives you control over trading frequency—the higher the threshold, the less frequently you trade. We see firms combining calendar and cost/benefit triggers—for each account the cost/benefit trigger starts high if you’ve traded recently but is lowered as the time since the last trade increases. This sort of cost/benefit + calendar workflow is also easy to implement and does not affect rebalancing efficiency.
- Mimicry of Human Rebalancing
Some firms are open with clients that a computer is involved in rebalancing their portfolios; it’s (rightly) touted as a high-end mechanism for serving clients better. Other firms wish to preserve the impression of a manager trading accounts one-by-one, trade-by-trade. This difference shows up in how portfolios are rebalanced even when, in both cases, it’s automated. Firms that want to preserve the illusion of direct human involvement in every trade will set the rebalancing systems to create round lot trades; they’ll put limits on cross-asset class substitutions, etc.
Level of Automation
The above rebalancing workflow variants don’t affect the level of rebalancing efficiency, but other variants do. The biggest driver of efficiency is the level of review—the more review, the less efficient your workflow. There are four basic options:
- Don’t Review Trades (Fully Automated)
This is robo rebalancing. Most firms we work with deploy this fully automated, no review, approach for some of their accounts, e.g., smaller accounts. And for a few firms, it’s basically just the way they rebalance accounts—all accounts.
- Review Every Account
This is the other end of the spectrum. The trades are still generated automatically, and most will be exported untouched, but there’s now a manual step being introduced for every account—reviewing the trades and modifying them as desired. It’s far more efficient than workflows where the trades are created manually for each account, but it’s still short of what is possible.
- Review Outlier Accounts
This is a middle ground. Only “outlier” accounts are reviewed. The rest are traded automatically. The outliers are accounts where the system-generated proposed trades trigger firm-set flags—for example, unusually high turnover or tax impact, or where the recommended portfolio still has unusually high tracking error. These reviews reduce efficiency, though the impact is obviously dependent on the sensitivity of the review triggers.
- Implement a Combination of the Above Review Practices
In practice, we see most firms using a mixture of all three of the above approaches to review. That is, firms will divide their accounts into multiple groups and apply a different workflow for each. Smaller accounts might be traded with no review (full robo), but all trades for ultra high net worth accounts (say, those above, $25mm) get a second look. And with “mid size” accounts, there’s “spot check” reviewing of outliers.
What are best practices? There’s lots of room for each firm to make their own decisions, each right for their own circumstances. But that doesn’t mean that all approaches are equal, either. The most consequential decision a firm can make when deploying an automated rebalancing system is the amount of time they spend reviewing trades.
Our fastest growing clients are the ones that spend the least time on manual review. We’d like to believe that this means that maximum automation inevitably and directly leads to maximum growth and profitability. But the causality is a bit more complicated than that. Consider the possibilities: are these fast-growing, high-automation firms growing fastest because their efficiency permits them to undercut their competition in price? Because they deploy their resources more efficiently (i.e., less time rebalancing and more time in front of clients)? Because they’ve deemphasized trading as a value proposition and topic for client discussion in favor of higher-value deliverables like financial planning? Or is it just that the fastest growing firms were forced to adopt more efficient workflows out of necessity—they were growing too fast to do anything else?
These explanations are not mutually exclusive, but we can exclude price competition as a factor. Our fastest growing clients are not pricing themselves lower than their competition. They compete on service levels. They do tend to put a greater focus on customization, tax management and financial planning. And they deemphasize “product,” trades and performance. We think automation is a big part of their success, but it’s automation in the service of a larger, coherent vision of their value add and their desired client experience. Automation isn’t a cure-all or a solution in itself. It’s a means to an end. And that, I suppose, is the heart of best practices.
For more on this topic, check out What is Rebalancing Automation?