The Wash Sale Blind Spot in Multi-Custodian Wealth Accounts

Every RIA I've talked to says the same thing about wash sales: "our custodian handles that." It's the assumption baked into how most firms think about tax-loss harvesting — sell the loser, book the loss, don't buy it back too soon, trust the 1099-B to catch you if you slip. That assumption is wrong in a specific, provable way, and it's quietly costing clients real tax benefits without anyone noticing.

We spent a chunk of this year digging into the actual operational mechanics of independent wealth management firms — not survey data, primary sources: IRS text, FINRA filings, SEC exam findings. One finding kept surfacing that I haven't seen written up anywhere else in this kind of detail, and it's worth an ops team actually sitting with it.

What a wash sale actually is, and who's responsible for catching it

The rule itself — Internal Revenue Code Section 1091 — is simple to state and easy to get wrong in practice. If you sell a security at a loss and buy something "substantially identical" within 30 days before or after that sale, the loss is disallowed. The tax benefit you thought you locked in doesn't exist. It gets added to the cost basis of the replacement position instead of showing up as a deduction this year.

Here's the part that gets glossed over: the law applies to the taxpayer, across everything they own — every account, every custodian, every account type, including IRAs and 401(k)s. It is not scoped to one brokerage relationship. A client's wash sale exposure is the sum of every account they hold, whether or not their advisor can see all of them.

Where the automatic check actually stops

Brokers are required to track and report wash sales to the IRS on Form 1099-B. We read the actual IRS instructions for that form rather than take anyone's summary of them, and the scoping language is explicit: a broker only has to match and disallow a loss "if both the sale and purchase transactions occur in the same account." Same account. Not same client, not same household, not even the same custodian if a client holds two accounts there under different registrations.

So the automatic net every advisor assumes is running underneath their tax-loss harvesting program only checks inside a single account. If a client sells a fund at a loss in their taxable account at one custodian, and their 401(k) — which they don't even think of as "your" account, and which the advisor may not see at all — automatically rebalances into a substantially identical fund at a different provider two weeks later, no 1099-B flags it. Nobody gets a warning. The loss quietly gets disallowed, and nobody notices until, at best, a CPA catches it eighteen months later reconciling a return, or at worst, nobody ever catches it and the client believes they received a tax benefit they didn't actually get.

This isn't a hypothetical edge case. Held-away assets — old 401(k)s, an employer plan, a spouse's account at a different firm — are exactly the kind of thing a wealth manager routinely has partial visibility into but doesn't control. The more custodians and account types a household touches, the more surface area this blind spot has.

A simple version of how it happens

Say a client has $80,000 in a taxable account you manage, and separately holds a 401(k) at their employer with a target-date fund that rebalances automatically every quarter. You harvest a $6,000 loss on an S&P 500 index fund in the taxable account in early November — good, defensible tax planning. Two weeks later, the 401(k)'s quarterly rebalance buys back into a fund holding the same underlying index inside the target-date fund, because that's just what the plan does on autopilot. Nobody coordinated those two events. Nobody was supposed to. But the IRS doesn't care that they happened in two different accounts at two different institutions — the wash sale rule still applies, the $6,000 loss is disallowed, and the 1099-B from the taxable account's custodian will never say so, because it only ever looked inside that one account.

Why this is worse than a rounding error

Tax-loss harvesting is one of the few places an advisor can point to a concrete, defensible number and say "this is what I did for you this year." If that number is quietly wrong — if the loss you told a client about got disallowed by a transaction they didn't even know was relevant — that's not an efficiency problem. It's a credibility problem, and eventually a fiduciary one. You told the client something that wasn't true, even though you had no way of knowing it at the time.

It compounds, too. A disallowed loss doesn't just disappear — it raises the cost basis of the replacement shares, which changes the tax math on every future sale of that position. Get this wrong across a book of a few hundred households running systematic tax-loss harvesting, and it isn't one bad number. It's a slow accumulation of quietly incorrect cost-basis records that nobody is actively tracking, sitting underneath client accounts your firm is contractually and fiduciarily responsible for getting right.

Why the software you already own doesn't close this gap

This isn't a knock on any specific vendor. Portfolio management and rebalancing platforms are built around the data they can actually see, which is the custodial feed they're connected to. A platform monitoring a Schwab account for wash-sale risk is watching Schwab. It has no visibility into what happens in a 401(k) at Fidelity, a spouse's IRA at Vanguard, or a taxable account somewhere else entirely — unless someone manually pulls that data together and someone else manually checks it against every trade before it executes.

That manual aggregation is exactly the kind of task that gets done inconsistently, or skipped entirely when a team is busy — which, per every operations benchmark we looked at for this industry, is most of the time. It also tends to fall on whoever is wearing the compliance hat at a firm this size, which at most independent RIAs under a billion or so in assets is one person doing that job alongside two or three other titles, not a dedicated team with time to manually cross-reference held-away holdings every time someone harvests a loss.

What actually closes it

The fix isn't more diligence from an already-stretched ops team. It's a system that sits above the individual platforms — pulling held-away and cross-custodian holdings data where a client has granted access, watching for overlap across every account tied to a household, and flagging a potential wash sale before a trade executes rather than after a 1099-B shows up the following February. That's an orchestration problem, not a portfolio-management-software problem, which is exactly why buying another point tool doesn't fix it — you need something that sits across the tools you already have, not one more tool added to the pile. It's the same reasoning behind most of the compliance automation we build for RIAs: the value is in connecting systems that were never designed to talk to each other, not replacing any one of them.

Done right, the system never makes the trading decision. It flags a risk and hands it to a person — the advisor or the ops team — to decide what to do, the same way any well-built compliance workflow should. Nobody should want an automated system making tax decisions unsupervised, and honestly, most advisors we've talked to wouldn't trust one that did.

Who's actually supposed to catch this at a firm your size

At firms managing a few hundred million to a couple billion, the honest answer is usually "nobody, formally." A dedicated compliance officer with the bandwidth to build and run a cross-account monitoring process doesn't tend to show up until a firm is well past the size where this kind of thing gets prioritized — most firms in that range have one person wearing the ops hat and the compliance hat at the same time, alongside two or three other responsibilities. Tax-loss harvesting itself is usually owned by whoever runs trading or portfolio management, and cross-custodian visibility is, structurally, nobody's full-time job. That's not a criticism of any individual firm — it's just what happens when a task doesn't map cleanly onto an existing role, so it falls through the gap between roles instead.

That's also exactly why this is a systems problem and not a "hire someone" problem. Adding headcount to manually reconcile held-away holdings against every harvested loss doesn't scale past a handful of households, and it's the kind of repetitive, error-prone checking work that people are genuinely bad at doing consistently, week after week, compared to a system built to do exactly one thing correctly every time.

Frequently Asked Questions

Does my custodian already check for wash sales across all of a client's accounts?

No. Broker wash-sale reporting on Form 1099-B is scoped to transactions within the same account at that broker. It does not see other accounts at the same custodian under different registrations, and it has no visibility at all into accounts at other custodians or held-away assets like an employer 401(k).

Is cross-account wash sale monitoring a regulatory requirement, or just good practice?

The wash sale rule itself, IRC Section 1091, is a tax law that applies to the client rather than a direct RIA compliance obligation in the way SEC recordkeeping rules are. But an advisor who tells a client they delivered a tax benefit that was actually disallowed is making a materially false statement about the value they provided — a fiduciary and reputational problem even without a specific rule requiring the monitoring itself.

What counts as "substantially identical" for wash sale purposes?

The IRS doesn't publish a precise, mechanical definition, which is part of what makes this hard to automate naively. Two share classes of the same fund are generally treated as substantially identical; two different ETFs tracking the same index are a genuine gray area, argued case by case rather than settled by a clean rule. Any system built to flag this needs a defensible ruleset, reviewed by someone who understands the ambiguity, not a rigid ticker-match.

How would I even know if this is happening in my book of business?

Most firms don't, which is the point of this piece. The only real way to find out is to pull held-away and cross-custodian data for a sample of households doing active tax-loss harvesting and manually check for overlapping purchases in the 61-day window around each harvested loss. It's tedious by hand and exactly the kind of check that's trivial once it's automated.

If this is news to you, it's worth 20 minutes

We'd rather show you what this looks like in your own book of business than talk about it in the abstract. If you want a straight answer on whether this is a real exposure for your firm — not a sales pitch, an actual look at the mechanics — book 30 minutes with us.