Case: GameStop, January 2021

The setup: short interest larger than the float

Short interest in this stock was reported above **100% of the float**, with widely cited figures around **140%** — a number that sounds impossible until you remember what a short sale is. A short is not a bet against a share that exists somewhere; it is a **borrow** (M14). A holder lends shares to a short seller, who sells them; the buyer becomes a new holder, whose broker may lend those same shares out again to another short. Each leg creates a claim, so the same economic share can support several borrows, and the reported short interest is a count of claims rather than of shares. Add **days to cover** — total shares short divided by average daily volume — and you have a position that cannot exit quickly at any price, which is the condition for a squeeze rather than merely a rising stock. Three obligations quietly raised the pressure as the price rose. The **borrow fee** is charged on the value shorted, so a doubling price doubles the daily rent — and on a crowded, hard-to-borrow name the rate moves from single digits to tens of percent. The **maintenance margin** on the short is also computed on the position's value, so a rising price against a short creates a margin call while the position is still losing money. And a **recall** can force the borrower to return shares on demand, which can be impossible when everyone is short the same name. Every one of those mechanisms gets worse exactly as the trade gets more painful, which is what makes a crowded short unstable rather than merely risky. One more structural fact belongs here: because the borrow chain runs through brokers, a squeeze is also a **settlement** event. Shares lent and re-lent must be tracked and returned, and each of those movements needs a counterparty that is still standing. That is why this story ends in a clearinghouse rather than in a stock chart. The three pressures on a crowded short — Borrow fee: Charged on the position value — doubles when the price doubles · Maintenance margin: A percentage of the position value, so it rises as the price rises ← · Recall: The lender can demand the shares back, whatever the price · Days to cover: The exit queue, measured in days of average volume

The loop: covering, call buying and dealer hedging

A squeeze needs buyers who have no choice. This episode had three groups of them, and they reinforced each other. The first is **shorts covering**, which is mechanical: a short has to buy back the shares it borrowed, and the more the price rises the more urgent that becomes — especially on margin calls that brokers can satisfy themselves by buying the shares. That is a demand for stock that increases with price, which is the opposite of a normal demand curve and is precisely why the arithmetic does not stop. The second is **new buyers**, whose reasons are their own, but whose activity matters structurally because it is concentrated: retail order flow arrives at a handful of wholesalers who must fill it, and fills require stock (M9). The third is the most interesting and the most often mis-described: **dealers hedging customer options**. When customers buy call options, the market maker who sells them is **short the calls** and must hedge — buying stock as the price rises to stay delta-neutral. The deeper out of the money the calls and the closer to expiry, the more violent that hedging becomes, because gamma is what decides how much stock has to be bought for each dollar the price moves (the mechanism of O10 and O19). So a wave of customer call buying can translate into mechanical stock buying by dealers, which pushes the price up, which obliges more hedging — a **gamma squeeze** layered on top of the short squeeze. The SEC's staff report documents the options activity and the hedging flows; the honest caveat is that the *magnitudes* are estimated, so the mechanism is solid and the precise attribution is not. Read the loop as a whole and the episode stops looking like a battle between two groups of people and starts looking like a system of obligations: shorts that must buy, dealers that must hedge, and a settlement layer that must be collateralised against the resulting volatility. The third of those is what actually ended it. The distinction worth keeping: a short squeeze is driven by borrow and margin, and a gamma squeeze by hedging. They are different mechanisms that happened on top of each other — so an analysis that mentions only one of them is incomplete.

The Friday that broke it: collateral, not conspiracy

The price path is easy to state: the stock traded in the tens of dollars in early January, closed at **$347.51** on January 27, printed an intraday high of **$483** on January 28, and closed that day at **$193.60**. Two weeks later it was in the $40s. The interesting part is what happened inside the plumbing on the morning of the 28th. Every U.S. equity trade is guaranteed by a clearinghouse, which becomes the counterparty to both sides (M13). To do that safely, it collects **collateral** from its members, and it **recalculates the requirement every morning** from the risk of the positions it is guaranteeing: volatility, volume and concentration. All three had exploded, and the requirement rose with them — for one broker, from about **$124 million to about $1.4 billion** in three days. Industry-wide figures were larger still. That cash has to be posted **that morning**; it is not a fee, it is capital. The broker did not have it. Its options were to raise money immediately or reduce the exposure the requirement was computed on, and the only lever that worked within hours was to stop customers **opening** new positions while letting them close — because a closing trade reduces the position the clearinghouse is pricing, and an opening trade increases it. That is exactly the asymmetry customers saw, and it is why the restriction looked arbitrary and was in fact arithmetic. The broker raised about **$1 billion** from its existing investors that day and roughly **$2.4 billion** more the following week, by which time the position had collapsed. The SEC's staff report, published in October 2021, analysed the episode and attributed the restrictions to clearinghouse margin requirements rather than to any decision about the merits of the stock. What it also documented is the mess underneath: payment for order flow concentrated a large share of retail flow with a few wholesalers (M9); the settlement cycle was still **T+2** at the time, which left more unsettled exposure to collateralise; and the options market's hedging flows amplified the move. Several of the changes since then trace directly to this episode — the move to **T+1** in May 2024 was motivated in part by exactly the margin risk on display here — along with a series of market structure proposals you will argue about in M23. The restriction was not a halt, and that distinction matters to a trader. A halt pauses the whole market for everyone; a broker-level restriction removes one side of the trade for one firm's customers, which is why the price kept trading and why a customer of a different broker could still buy.

Sources

• The official account of January 2021, including clearing and optionsSEC — Staff Report on Equity and Options Market Structure Conditions in Early 2021
• Short interest, borrow chains and how a share can be lent more than onceS3 Partners / FINRA — short interest and securities lending data
• Dealer hedging flows from customer option demandSEC Staff Report, January 2021 — options market structure and gamma
• Clearinghouse margin methodology and why it rises with volatilityDTCC / NSCC — clearing fund and margin methodology
• The congressional record of the episodeU.S. House Committee on Financial Services — hearing, February 18, 2021