GameStop’s buy button was switched off in the clearinghouse

The week

In the last week of January 2021, a struggling video game retailer became the most talked-about stock in the world. Trading in GameStop and a handful of other heavily shorted stocks, meaning stocks that many professional investors had bet would fall, reached extraordinary levels. Much of the buying came from retail investors, ordinary people trading on their phones, who coordinated on social media. Prices and trading volumes moved far outside anything their histories would have suggested.

The numbers, as SEC staff later set them out, were extreme. Through most of 2020, short interest in GameStop had hovered around 100% of its public float, peaking just above that at the end of December. On January 22 the price rose from $43 to $72 in about three hours. On January 27 it closed at $347.51, more than 16 times its close on January 11. The next day it touched $483 during trading. From January 13 to 29, about 100 million shares changed hands on an average day.

Then, on January 28, the buy button stopped working. Robinhood and several other retail brokers stopped customers from opening new positions in some of those stocks. Customers could still sell what they held, but they could not buy more. To millions of people watching the price swing, it looked as though someone had changed the rules in the middle of the game.

The restrictions came in steps. According to the SEC staff report, one broker had begun raising margin requirements on GameStop on January 26, from 80% to 100% the following day. It cut the number of GameStop option contracts a customer could hold, several times. On January 28 it put eight securities into “position closing only” status: AMC, BlackBerry, Bed Bath & Beyond, Express, GameStop, Koss, Naked Brand and Nokia. The next day it replaced that with limits on how many shares a customer could hold, and at one point the list grew to over 50 names. Fractional share purchases in those names were stopped. Robinhood’s later IPO filing said its purchase restrictions ran from January 28 to February 5.

Robinhood’s explanation pointed somewhere most of its customers had never heard of: its clearinghouse. A clearinghouse is an industry utility that stands in the middle of stock trades between brokers. It guarantees that each trade will be completed, even if one side fails before the shares and cash change hands. To protect itself, the clearinghouse requires each member broker to post a deposit, a kind of security bond, sized to the risk the broker’s unsettled trades create. Robinhood said those deposit requirements had risen sharply, driven by the wild price swings and by how concentrated its customers’ positions had become in the same few names.

The broker was not alone in facing a large bill. On the morning of January 28, the National Securities Clearing Corporation issued margin calls to 36 members totaling $6.9 billion, according to the SEC staff report. That brought total required deposits across all members to $25.5 billion. Margin at the Options Clearing Corporation, which clears listed options, rose in a similar pattern, though it took no unusual steps of its own. The clearinghouse has no power to tell a member to stop trading a stock. The decision to restrict was each broker’s own.

The restriction, then, came from a broker running short of the cash needed to keep trading safely and choosing to stop the exposure from growing while it caught up. Robinhood raised $1 billion from existing investors on January 29 and announced a further $2.4 billion on February 1.

Robinhood was not the only broker to restrict. Interactive Brokers also said that in some securities customers could only close positions, not open new ones. Some smaller apps that relied on another firm for clearing said the limits came from that firm. On the other side of the trade, some funds had closed out their short positions by the end of January, taking heavy losses.

The explanation soothed nobody. The restrictions provoked public anger, congressional attention and lawsuits. Customers felt that the little guy had been shut out to protect the big players, and many suspected the worst. On January 29, the SEC’s acting chair and commissioners issued a statement saying the agency was monitoring the volatility and would look closely at actions by regulated firms that might disadvantage investors. A House committee held a hearing on February 18, where executives of the broker, a market maker and a hedge fund testified that hedge funds had played no part in the decision. In October 2021, SEC staff published a report on the episode that walked through the market mechanics in detail.

For financial services firms, the week was a vivid demonstration of how the plumbing works. The back-office systems that move money and securities after a trade can suddenly decide what customers see on their screens.

Most coverage of the meme-stock week focused on hedge funds, social media and retail trading. Those were the colorful parts of the story. For technology and risk chiefs at broker-dealers, the lasting story was about plumbing.

The decision that angered customers followed from how trades clear and settle, how collateral is calculated, and how fast a firm can raise cash when it needs it. It then reached customers through a mobile app that had no good way to explain it.

The episode separates neatly into three layers: the settlement cycle, the firm’s liquidity operating model and the customer channel. Each is owned by a different part of the organization. Each looked reasonable on its own. Together, under stress, they produced a public relations disaster and a regulatory inquiry.

What follows starts with a short account of what happens after a customer presses “buy,” since the three layers make little sense without it. It then takes the layers in order. Two sections follow on how the same chain appears at a bank and how to rehearse the handoffs between teams. A set of questions closes.

After you press “buy”

When you buy a stock on an app, the purchase feels instant. The price is locked in and the shares appear in your account within seconds. Behind the scenes, though, the trade is only a promise. The actual exchange of cash for shares happens later, in a step called settlement.

Buying a house works the same way. You agree a price and sign a contract on one day, but the keys and the money change hands at closing, some time later. Between signing and closing, things can go wrong: a buyer’s financing can collapse, or a seller can disappear. Everyone involved carries some risk until closing day.

Stock markets handle that gap with a few pieces of shared infrastructure:

• The broker is the firm that takes your order, such as a retail trading app or a traditional brokerage.
• Clearing is the process of confirming who owes what to whom once trades are agreed, and netting it all down so that each firm knows its final obligations.
• The clearinghouse is the central utility that stands between brokers. It becomes the buyer to every seller and the seller to every buyer, so that if one broker fails, the others still get what they were promised.
• Settlement is the final step, when cash and securities actually move and the trade is complete.

In early 2021, U.S. equities settled two business days after the trade, a cycle known in the industry as T+2. For those two days, the clearinghouse is on the hook. If a broker went bust on day one, the clearinghouse would have to complete that broker’s trades itself.

No sensible guarantor takes on that kind of risk for free. So the clearinghouse asks each member to post deposits, cash set aside as a cushion, sized to the risk of its unsettled positions. The bigger and more volatile the risk, the bigger the deposit.

How the deposit is built

For U.S. equities, the clearinghouse is the National Securities Clearing Corporation, a subsidiary of the Depository Trust & Clearing Corporation. Its members contribute to a Clearing Fund, in cash and eligible securities. The fund exists to cover losses if a member defaults, until the clearinghouse can close out that member’s positions.

The SEC staff report describes the main pieces. The largest part of a typical requirement is a volatility charge, an estimate of how much a member’s unsettled portfolio could lose while it is being closed out. Mark-to-market charges cover price moves that have already happened. The clearinghouse can add special charges when it sees unusual volatility in particular securities. It can also impose an excess capital premium when a member’s margin exposure exceeds its excess net capital, meaning the capital it holds above the regulatory minimum.

Each piece is formula-driven, but not all of it is predictable from the member’s side. The volatility charge moves with market data. The special charge depends on the clearinghouse’s judgment about specific securities. The capital premium depends on the ratio between the other charges and the member’s own capital, so it can jump when both move against the firm at once.

Self-clearing and introducing brokers

Not every broker deals with the clearinghouse directly. A self-clearing broker is a clearinghouse member, posts its own deposits and makes its own decisions about restrictions. Robinhood cleared through its own affiliated clearing broker, which was why the deposit call landed on the group directly.

An introducing broker passes its customers’ trades to a separate clearing firm under a contract. The clearing firm posts the deposit and carries the risk, so it decides what to restrict. Several smaller apps said in January 2021 that their restrictions came from their clearing firm. Their customers met a constraint set two companies away.

The difference matters for anyone choosing an operating model. A self-clearing firm controls its levers but must fund them. An introducing firm pays its clearing firm to carry that burden, and in a crisis it learns the outcome when the clearing firm decides. Banks and credit unions that offer brokerage to their customers or members, often through a third-party broker-dealer, are further removed still.

The settlement cycle: time is risk, and risk needs collateral

That simple idea, that unsettled time creates risk and risk requires collateral, lies at the heart of what happened on January 28.

The clearinghouse protects itself by requiring members to post deposits sized to the risk of their unsettled positions. Those requirements rise with volatility, meaning how wildly prices are swinging, and with concentration, meaning how much of a firm’s exposure is piled into a small number of securities.

A broker whose customers all bought the same few volatile stocks at the same moment therefore faced a large, sudden deposit call. SEC staff later reported that on January 28 one retail broker faced a $1.4 billion requirement, plus an excess capital charge of more than double that, which the clearinghouse waived. That excess charge is an add-on that applies when a firm’s requirement is large relative to its own capital. It signals that the guarantor is worried about whether the member can absorb losses.

The waiver was not targeted at one firm. The clearinghouse waived the capital premium for all members that day, because the ratios were driven by market-wide volatility rather than by any member’s own actions. It removed the waiver on February 2. A broker could not plan on a waiver; it was a discretionary act in an extreme week.

The way the number was produced matters. The amount was driven by models the broker did not control, using inputs that moved by the hour. The broker could not negotiate the formula in the moment. It could only meet the call or reduce the exposure that generated it. Restricting purchases stopped the exposure growing while the firm met the call.

From a risk management point of view, that is a coherent response. From a customer’s point of view, it looked like the broker was choosing sides. Both views can be true at once, which is part of why the episode was so bruising.

Why buying was stopped and selling was not

Critics asked why restrictions fell only on purchases. The mechanics give the answer. A new purchase adds unsettled exposure in the most volatile names, which raises the volatility charge. A sale of an existing position reduces the customer’s holding, and closing out positions is what a clearinghouse wants to see under stress. A “position closing only” status is the standard tool for capping exposure without trapping customers in what they already hold.

The counter-argument is that one-sided restrictions have one-sided price effects. Removing a large group of buyers from the market while sellers remain free will tend to push prices down. Both things are true. A broker facing a capital problem can only act on its own customers’ flow, and it cannot make that action neutral for the market. For that reason the decision invites suspicion, and it needs a clear record of the reasons.

Margin as the first lever

The restrictions were the end of a sequence that began with margin. Raising the house margin requirement on a stock to 100% means customers can no longer borrow to buy it. That reduces the broker’s own credit exposure to its customers, separate from the clearinghouse deposit. Limits on option contracts work the same way for derivatives.

The SEC staff account shows these levers being used first, over two days, before purchases were stopped. That sequence is sensible. Its weakness was speed: each step was a reaction to that day’s numbers, not the next morning’s.

The length of the settlement cycle is a direct input to the collateral a broker must hold. Shorter settlement means less unsettled exposure and, other things equal, lower deposits, because there is less time for anything to go wrong. That reasoning drove the SEC’s later move to one-day settlement, discussed in the hindsight section below.

Concentration as a product property

The same logic explains why retail brokers are more exposed to sudden spikes than firms with diversified institutional flow. A broker serving pension funds and asset managers sees buying and selling across thousands of securities, which tends to balance out. Retail positions can concentrate around the same few names within hours, especially when an app’s design and social media point customers the same way.

That last point deserves emphasis for anyone on a product committee. Concentration is a property of the customer base, and product design shapes the customer base. Features that highlight trending stocks, push notifications about big movers, and the general feel of an app all influence what customers buy and when. Those choices end up in the treasury department’s collateral bill.

Firms differ in how much this applies. A wealth manager whose advisers build diversified portfolios rarely sees retail clients pile into one name in a day. A full-service brokerage with a mix of institutional and retail flow has natural offsets. A self-directed, mobile-first broker whose customers are young and active has the least diversification and the most exposure to a single viral idea. The SEC report noted that Robinhood said its average customer was 31 with a median account balance of $240.

The liquidity operating model: forecasting a number you do not control

“Liquidity” in this context means having enough cash, or the ability to get cash quickly, to meet obligations as they fall due. A firm can be profitable and well capitalized on paper and still run into trouble if it cannot produce cash on a particular morning.

Brokers knew their clearinghouse requirements were formula-driven. Fewer could forecast them intraday, meaning during the trading day, under stress. Fewer still could model the effect of concentrated customer flows, and decide in advance which levers they would pull when the number came in high.

Those levers include:

• Drawing on committed credit lines, which are pre-arranged loans a bank has agreed to provide on demand.
• Raising new capital from investors.
• Raising margin requirements on specific securities, so customers borrowing to buy them must put up more of their own money.
• Limiting position sizes for individual customers.
• Restricting purchases altogether.

That capability is part technology and part governance. On the technology side it needs:

• A near real-time view of customer positions by security, with concentration measures that update through the day. Many firms reconcile positions overnight in batches. That is fine on a calm day and dangerously slow on a wild one.
• A replica of the clearinghouse’s risk calculation good enough to estimate the next morning’s call before it arrives, including the add-on charges that apply when requirements are large relative to a firm’s capital. The aim is that the bill is never a surprise.
• Stress scenarios that ask what happens if the top five names double in volume and volatility. These are “what if” simulations that show the size of the call under conditions the firm has not yet seen.

How fast each source of cash arrives

The levers differ most in speed, and speed is what matters at six in the morning. Cash already held at the firm is immediate. A transfer from an affiliate can be fast if it is pre-approved. The SEC staff report noted that one member avoided a special charge by offsetting its exposure with a transfer from an affiliate. A committed credit line can usually be drawn the same day, if the paperwork is in order and the firm meets the conditions.

New equity is slow. Even a firm with willing investors needs days to agree terms, as the January raises showed. Customer-facing levers, margin changes and restrictions, are the fastest of all, because the firm can switch them on itself. So they get used when the slower options were not prepared.

Credit lines have conditions that only appear under stress. Lenders may limit what collateral they accept, or include terms that let them refuse to fund if the borrower’s condition has changed. A firm should test, in calm times, how fast each line can be drawn and against what collateral. A line that cannot be drawn by mid-morning does not help with a call due that morning.

Where the cash has to be

Cash in the wrong legal entity does not help. The clearinghouse call falls on the broker-dealer, which must also meet the SEC’s net capital rule. Money raised at a parent company must be moved into the broker-dealer, and capital contributions can take time to document. Firms with a bank in the group face a further limit. Section 23A of the Federal Reserve Act restricts how much a bank can lend to its affiliates, so a bank parent’s deposits are not a free source of broker-dealer funding.

What a replica can and cannot do

The clearinghouse publishes calculators so members can estimate their deposits for actual or hypothetical portfolios. A firm can build these into its intraday risk systems and refresh them as positions change. That covers the formula-driven parts of the requirement.

It cannot cover the discretionary parts. Special charges for unusual volatility, and the decision whether to waive a capital premium, depend on the clearinghouse’s judgment. A sound liquidity model treats those as a range, not a point. It asks what the firm would do if the call came in at the top of the range, and holds a buffer or a pre-arranged lever for that case.

The playbook

On the governance side it needs a playbook. The playbook names who can impose each lever, on what triggers, in what order, and how quickly funding can be drawn. Like most contingency plans, it is boring to write and invaluable when needed.

A good playbook sets triggers in terms the systems can measure. Examples include the estimated next-day call as a multiple of the normal level, concentration in the top five names, and the call as a share of excess net capital. Each trigger maps to an action and an owner. The early actions are internal: alert the chief financial officer, pre-position collateral, notify lenders. The later ones reach customers, in a fixed order, with restricting purchases last.

What the board should see

Boards rarely see clearinghouse exposure at all. A short monthly view is enough. It shows the normal range of the deposit requirement, the stressed estimate under the firm’s own scenarios, and the committed funding available against it. It also shows how long each source of funding takes to arrive.

Alongside those numbers, the board should see the playbook’s ranked levers and the date the firm last rehearsed them. With that, directors can judge whether the buffer is a deliberate choice or an accident of history. They can also ask the right question when a new product feature promises growth in active trading.

The playbook also names who must be told before a customer-facing lever is pulled: compliance, the customer support leads, communications and, depending on the firm, its regulators. A broker-dealer’s primary regulators are the SEC and FINRA. Firms should check with counsel on notice obligations that may apply in their circumstances.

Firms that ran this as a daily treasury routine had more options when volumes spiked. Firms that learned the size of the call in the early morning had fewer. With hours or minutes to act, the options narrow to the ones that can be switched on immediately. The most visible option, restricting purchases, was also the most damaging to customer trust.

The trade-off is cost. Committed credit lines and spare capital are expensive to hold for a rare event. Banks charge fees for credit they promise but may never lend, and capital lying idle earns little. A firm that models its exposure well can size that buffer deliberately, and can explain the choice to its board and regulators. A firm that does not is either overpaying for insurance it cannot justify or underinsured without knowing it.

The customer channel: restrictions as product behavior

To the customer, the restriction appeared as a change in what the app allowed. Buttons disappeared or orders were rejected, often with little explanation. Explanations came after the fact and were widely read as evasive, whatever the underlying reasons.

Firms had underinvested in this layer. A broker’s app is also its disclosure channel. In an older world, a broker might phone clients or send a letter. In a mobile-first business, the screen is where the firm speaks to its customers, and in a crisis the screen is all most customers will see.

Rights in the agreement, surprise on the screen

Brokerage account agreements commonly reserve the right to decline orders or cancel trades without prior notice, as the SEC staff report noted. The legal right existed. Few customers had read it, and none expected it to apply to the most popular stocks of the moment. A contractual right that customers do not know about protects the firm in court and does nothing for it in public.

What must be ready

When back-office constraints change what customers can do, the firm needs three things ready in advance:

• Prepared language for each kind of restriction, written so a customer can follow it and approved before it is needed.
• A fast approval path through compliance, so that wording for an unexpected situation can be cleared in minutes rather than days.
• The ability to show the message at the point of action, at the moment a customer tries to trade, instead of in a blog post or a press release they may never read.

The approval path is a regulatory question as well as an operational one. Under FINRA’s communications rule, a message sent to many retail customers is generally a retail communication. Retail communications generally need approval by a qualified principal before use, and must be fair and balanced. Templates approved in advance therefore matter. Each template covers a restriction type: position closing only, a position limit, a margin change, no fractional purchases. The firm then fills in only the security names and times.

Restrictions as states in the product

The January restrictions came in several forms within a week. There was a closing-only status, then share limits per customer, then a block on fractional purchases, each eased in turn. Each form is a state the app must handle. The order screen needs to show why a button is disabled. The rejection message needs to say which rule applied. The portfolio screen needs to show what the customer can still do.

Each state should be built and tested before it is needed, on every surface: the mobile app, the website, any public trading interface, and the screens used by support agents. If the app says one thing and a support agent another, the customer will believe the more suspicious of the two.

Without that, a defensible risk decision becomes a reputational failure. Customers fill a silence with the most suspicious explanation available, and social media amplifies it within the hour.

Capacity under a herd

Platform capacity mattered too. Some retail platforms reported slowdowns or access problems on the peak days, which compounded the frustration. Capacity planning is the discipline of making sure systems can handle the heaviest load they are likely to face. Capacity plans for retail platforms have to assume that interest in a single security can arrive all at once, from customers who all opened the app because of the same viral post.

That kind of load is unusual in shape as well as size. Everyone requests the same quote, the same chart and the same order ticket. Caches that work well for spread-out traffic can fail when one key is hot. Order systems that queue fairly across securities can stall on one. Load tests should include a single-security surge as well as a uniform rise in traffic.

The same chain at a bank

A bank’s customers can meet a payment cap at the checkout because of a funding formula they have never heard of. The pattern appears in any business where a back-office obligation set by someone else can suddenly constrain what customers are allowed to do.

Suppose a regional bank offers a popular digital wallet. Its payments are processed through a network that requires the bank to pre-fund settlement accounts. A viral promotion triples payment volume over a weekend. On Monday morning the network asks for far more pre-funding than treasury expected. If the bank cannot meet it immediately, it may have to cap transfers, and its customers will discover that cap when a payment fails at a checkout. The same three layers appear: an external formula, a treasury response and a customer screen.

Where the pattern shows up

Instant payment systems make the pattern concrete. On The Clearing House’s RTP network, participants fund a joint account held at the Federal Reserve Bank of New York, and payments draw on those balances. A bank whose customers send far more than they receive can see its position drain over a weekend. Card settlement, wire cut-offs and correspondent banking arrangements each have their own version.

The pattern differs by institution. A large bank has a treasury desk watching intraday liquidity and direct access to Federal Reserve accounts. A community bank or credit union may rely on a correspondent or a corporate credit union for settlement, and sees its position later and less precisely. A fintech that offers accounts through a partner bank often has the least visibility. The partner bank holds the settlement relationship, and the fintech learns of a cap when its customers do.

The operational fixes carry over directly. Treasury needs an intraday view of settlement positions by payment rail, not an end-of-day balance. The playbook needs ranked levers, from moving funds between accounts to temporary limits on individual customers, with blanket caps last. Product needs prepared messages that say why a payment is held and when it will go. A fintech and its partner bank need to agree all three in their contract, because neither can act alone in the moment.

The financial stakes are real on every side:

• Money. Emergency capital raised in a hurry is expensive, and committed credit that is never used is a standing cost.
• Risk. A firm that cannot meet a clearinghouse call faces consequences far worse than an angry customer base.
• Customers. Trust lost in a single day can take years to rebuild, and the customers most affected are often the most vocal.
• Regulators and lawmakers. The episode produced a congressional hearing, a detailed SEC staff report and lawsuits. Leaders should assume their decision records will be read by outsiders.
• Reputation. The public remembers the button that stopped working and forgets the collateral formula behind it.

Rehearsing the handoffs

Because the three layers have different owners, they are usually exercised separately. Market infrastructure and regulators set the settlement cycle. Treasury and risk run liquidity. Product and compliance own the customer channel. Each team may be excellent at its own job and still fail at the handoffs.

A single scenario exercise that starts with a clearinghouse call and ends with the message on a customer’s screen tests those handoffs. It can be run as a tabletop drill, a structured role-play in a conference room. Treasury announces that tomorrow’s estimated call has tripled. Risk explains which customer positions drive it. Product shows what the app would display if purchases were restricted in three named securities. Compliance approves, or rewrites, the message. Leadership decides which lever to pull first.

Designing the exercise

The exercise works best on a realistic clock. It opens the evening before, when concentration data first shows a problem. The next inject arrives before dawn, when the actual call comes in higher than the estimate. A third lands at the market open, when the app starts to slow and social media notices. Each inject forces a handoff: risk to treasury, treasury to leadership, leadership to product and compliance, product to support.

The useful measures are times and artifacts. How long from the first signal to an estimate leadership trusts. How long from a decision to a message live on customers’ screens. The artifacts are the decision log, the approved message and the support script. A regulator reviewing the episode months later would ask for all three.

The test is whether each team sees what the others see, in time to act, and whether the decision record would satisfy a regulator reading it months later.

Objections, and why the argument holds

Three objections deserve a hearing. The first is that the discretionary parts of a clearinghouse call cannot be forecast, so the effort is wasted. They cannot be forecast precisely, but they can be bounded, and the response to the upper bound can be prepared. A firm that has done this has slower, cheaper levers ready, and does not need the fastest, most visible one.

The second is that shorter settlement fixes the problem. It shrinks the exposure, which helps. It does not change the fact that a concentrated surge can still produce a large call at short notice, or that the customer will still see the result on a screen.

The third is that the real failure was the decision to restrict at all, and a well-run broker would never do it. A firm that cannot meet a clearinghouse call has no good options. The case here is for making restriction the last lever rather than the first, and for preparing to explain it if it comes.

Firms that are ready tend to share a few features:

• Treasury can estimate the next day’s clearinghouse call during the trading day rather than after the fact.
• The playbook ranks levers so that the most customer-visible one, restricting purchases, comes last.
• The board has seen and approved the size of the liquidity buffer and understands why it is that size.
• Customer messages for each restriction type already exist and have been tested on real screens.
• Peak-load testing uses scenarios where everyone wants the same thing at once.

The unready make recognizable errors:

• Treating the clearinghouse requirement as a fixed cost of doing business instead of a variable that can spike.
• Relying on overnight data to manage a risk that moves by the hour.
• Keeping the treasury playbook in treasury, so that product and compliance learn about a restriction at the same time as customers.
• Assuming that a correct risk decision will be understood as correct without a clear explanation.
• Designing app features that encourage concentrated trading without telling treasury what they may cost.

From the deposit call to the customer’s screen

Nobody needs to understand every detail of clearinghouse models to ask the questions that reveal whether the organization is ready. Five are a good place to start:

1. If our clearinghouse or payment network called for three times the usual deposit tomorrow morning, how many hours of warning would we have?
2. Who is allowed to restrict customer activity, on what trigger, and who must they inform first?
3. How much committed funding do we hold for a stress event, and when did the board last review that number?
4. What exact words would a customer see on their screen if we restricted a product today, and who approved them?
5. When did treasury, risk, product and compliance last run one exercise together from start to finish?

The answers belong to different people. The first and third belong to the treasurer and chief financial officer. The second belongs to whoever owns the playbook, usually the chief risk officer. The fourth belongs to compliance and product jointly. The fifth belongs to the chief executive, because only the top of the firm can make four functions rehearse together.

For an introducing broker, a bank or a credit union whose brokerage runs through another firm, the questions point outward. The useful versions ask what the clearing firm or partner would do, how quickly the institution would hear, and what its own customers would see. Those answers belong in the contract and the due diligence file.

The questions are not technical, but all of them reach straight into the plumbing. The GameStop week showed that systems customers never see decide what customers are allowed to do. The organization that connects those systems in advance gets to choose its response instead of having one forced upon it.

With hindsight

The most concrete outcome was regulatory. In February 2023 the SEC adopted rules shortening the standard settlement cycle for most U.S. securities transactions to one business day, effective May 28, 2024. Less time between trade and settlement means less unsettled exposure for the clearinghouse to guard against.

The package went beyond the settlement date. Broker-dealers handling institutional trades had to complete allocations, confirmations and affirmations by the end of trade date, either under written agreements or under their own written policies. Registered investment advisers had to keep records of those steps. Central matching service providers had to support straight-through processing and report to the SEC each year on progress. Canada and Mexico moved to one-day settlement at the same time.

The change was not trivial to deliver. The transition required significant work on trade affirmation, allocation and funding, especially for firms with international clients. Affirmation is the step where both sides confirm a trade’s details. Allocation is the process of splitting a large institutional order across the individual accounts it was placed for. Funding means having the cash in the right place in time. Firms with clients in other time zones had much less room to complete those steps overnight.

Even so, early settlement data pointed to a smooth transition. On the first T+1 settlement day, DTCC reported that about 94.6% of eligible trades were affirmed by its 9:00 p.m. trade-date cutoff, up from 73% at the end of January. Fail rates were in line with or slightly below the May average under T+2. NSCC’s Clearing Fund requirement fell to $9.1 billion, about 25% below the previous month’s T+2 average. That drop is the mechanism this article describes, seen from the other side.

The October 2021 SEC staff report linked the restrictions to clearinghouse deposit requirements and to brokers’ own risk management. It named four areas for further study. They were the forces that may cause a brokerage to restrict trading, digital engagement practices and payment for order flow, off-exchange trading through wholesalers, and short selling. On the first, it said directly that shortening the settlement cycle was one way to reduce the risk posed by thinly capitalized brokers.

The other areas had mixed fates. Payment for order flow, the practice of brokers being paid by trading firms to route customer orders to them, was debated for years and not banned. The SEC requested public comment on digital engagement practices in August 2021, meaning design features borrowed from games that can encourage more frequent trading. In 2023 it proposed a rule on conflicts of interest in firms’ use of predictive data analytics, and withdrew that proposal in 2025. In October 2023 it adopted new short-sale reporting rules, though compliance dates were later pushed back.

State and self-regulatory action came faster. In June 2021, FINRA fined Robinhood and ordered restitution totaling about $70 million over outages, misleading information and options approvals. Massachusetts securities regulators, who had sued Robinhood over its app’s game-like features in December 2020, settled with the firm in January 2024. Robinhood went public in July 2021. The consolidated private lawsuits made little headway; courts dismissed the central antitrust claims.

What shorter settlement did not fix matters as much. It reduced one input to the problem. The rest still depends on each firm’s design: how fast its treasury can see a call coming, and how clearly its app can tell a customer why a button has changed.

The speed problem also spread beyond brokers. In March 2023, Silicon Valley Bank failed after a deposit run that the Federal Reserve’s own review linked in part to social media and the speed of digital banking. A bank’s funding can now move as fast as a brokerage customer base can pile into a stock. The three layers apply to deposit outflows as much as to clearinghouse calls.

A concentrated rush of customer activity can still produce a large deposit call. A firm that learns about it at dawn will still have few options. A message written in haste will still be read as evasive. The settlement clock got shorter; the need for treasury, risk, product and compliance to work as one team did not go away.

The back office deserves front-office attention from boards. The obligations a firm owes to market infrastructure, the cash it keeps on hand to meet them, and the words it shows customers when those obligations bite are one chain. A weakness anywhere in that chain shows up, sooner or later, on a customer’s screen.