The State of Delisting 2026

Founders, treasury leads, and core team members of live token projects, listed or about to be listed, who need an honest picture of exchange risk monitoring. No prior trading infrastructure knowledge is assumed. It is equally relevant if you have never been flagged, if you were warned this week, or if you are already tagged and need working numbers rather than reassurance.

1.1 The reason

Across recovery engagements, one pattern repeated often enough to be worth investigating properly. Projects arriving in ST status were consistently surprised by it, and the surprise was always the same shape. They believed a risk tag was an unusual event, a signal that something had gone badly wrong with the project itself, and something that happened to weak teams with dead products. They were also consistently wrong about how much time they had, and about how much a recovery would cost.

Anecdotally, none of that matched what we were seeing. The projects entering ST status included ones with real products, funded teams and live communities. The correction windows they assumed they had were often multiples of what the exchange actually gave them. And the amount of capital required to reopen a book to the standard a risk team wanted was routinely several times what the project had budgeted.

Anecdote is not evidence, so we ran the numbers.

1.2 The hypothesis

We set out to test three specific propositions, each of which, if true, would change what a project should do.

• H1. Delisting and risk labelling are not rare events. They are a rising, structural feature of the market, and their frequency is increasing over time rather than holding steady.
• H2. The frequency is driven primarily by exchange-side conditions, regulatory pressure and market environment, rather than by a decline in project quality. If true, the risk of being flagged is partly outside a project's control, and rises precisely when the wider market weakens.
• H3. Requirements, correction windows, and the capital needed to satisfy them vary so widely between venues that any market-wide average is operationally useless. If true, a project needs venue-specific numbers, not general best practice.

What the document is about to prove Open the proof map for H1, H2, and H3 View proof map

H1 Scale Tested in 2.1 and 2.2 through rising quarterly delisting volume and the concentration of removals by venue.

H2 Market pressure Tested in 2.5 where the report links rising delistings to weaker market conditions and tighter exchange-side risk.

H3 Venue-specific reality Signalled in 2.2, then made operational in sections 04 and 05 through exchange windows and required recovery capital.

2.1 2.2 2.5 04 05

• Quarterly delisting volume

• Concentration by venue
• First venue-specific signals

• Weaker market conditions
• Exchange-side risk

• Exchange correction windows

• Required recovery capital

1.3 The method

We compiled and reviewed delisting and risk-labelling activity across the major centralized exchanges, drawing on a dataset of more than ten thousand token listings and roughly four thousand delistings across the ten largest venues. This was cross-referenced against official exchange announcements, monitoring-tag and ST-warning notices, and quarterly listing records covering Q1 2024 through Q2 2026.

Market-wide frequency, concentration, and category data come from that compiled dataset. Correction windows, required funds, and per-exchange operational behaviour come from direct recovery engagements, because most exchanges do not publish thresholds at all. The two sources are kept separate throughout, and every table states which one it draws on.

A note on sourcing Open the source split used throughout the report Open note

Market-wide figures are compiled from published industry data on listings and delistings across major exchanges, and from official exchange announcements. Required funds, correction windows, and per-exchange operational behaviour are our own first-hand field experience, most exchanges do not disclose thresholds publicly. The two are kept clearly separate throughout.

2.1 Delisting volume is rising every year

The headline finding is unambiguous, and it supports H1 directly. Quarterly delistings across major exchanges rose from just 67 in Q1 2024 to a record 786 in Q2 2025, and have since settled into a sustained range of roughly 400 to 500 per quarter through 2026. That is a floor now sitting six to seven times higher than it was two years earlier. Delisting has moved from an occasional cleanup event to a permanent, high-volume feature of the market.

Just as telling, Q2 2026 saw only 351 new listings, the lowest quarterly figure since Q3 2023, and one of only two quarters since 2024 in which delistings actually outpaced new listings. The listing pipeline is narrowing at the same time the delisting machine is running at full speed.

2.2 The activity is extremely concentrated

Delisting is not evenly spread across exchanges. In the first half of 2026 a single exchange, Gate, removed 573 tokens, nearly sixty percent of all delistings across the tracked venues. In Q2 2026 alone Gate delisted 221 tokens, more than every other tracked exchange combined. At the opposite extreme, OKX delisted zero tokens in the same six months.

That is the full spread of the market in two venues: one removing tokens at industrial scale, one removing none at all, over the same six months and the same market conditions. This is the first direct support for H3. The same underlying pressure produces completely different behaviour depending on where a token is listed, which is exactly why the per-exchange requirements in section 04 matter more than any market-wide average.

Transition Market-wide averages stop helping here Open the bridge

01

Same market conditions

The first half of 2026 is one market backdrop shared by every tracked exchange.

02

Different exchange behaviour

Gate removes tokens at industrial scale while OKX removes none at all.

03

This is why section 04 matters

Once venue behaviour diverges this hard, correction windows and recovery rules matter more than market-wide averages.

Compare exchange windows Jump to recovery capital

2.3 Which tokens get removed

Breaking first-half 2026 delistings down by category shows the previous cycle's leaders being unwound first: DeFi led with 207 delistings, followed by GameFi at 141 and Meme at 98. These are the same categories that supplied the bulk of new listings during the 2025 boom, which means exchanges are unwinding exposure roughly in proportion to how aggressively they built it up.

Of all tokens listed during the 2025 boom, about 6.8 percent had been delisted by mid-2026, with NFT, GameFi and Meme showing the highest delisting rates, and tokenized assets showing effectively none. The pattern is consistent: the categories that were listed fastest are being removed fastest.

2.4 ST and monitoring tags are a formal pipeline

Delisting rarely happens without a warning stage, and that stage is increasingly formalised. Binance runs an explicit Monitoring Tag that it publicly describes as one of the final steps before delisting; in June 2026 it added ACT, BLUR, PIVX and QKC to that list.

It also introduced a community Vote to Delist mechanism in 2025 targeting tokens in this monitoring zone. The first round removed 14 tokens, including BADGER, BAL, BETA, CREAM, CTXC, ELF, FIRO, HARD, NULS, PROS, SNT, TROY, UFT and VIDT, and drew 103,942 total votes from 24,141 participants. A second round put a further 17 tokens under review.

MEXC runs the same logic through its Assessment Zone and ST Warning, with delisting possible as little as three days after the ST tag. The label differs by exchange; the pipeline is identical.

2.5 The trend tracks the state of the market

The single most important pattern to understand is why this is rising, and it is the test of H2. It is not that projects have suddenly gotten worse. It is structural: exchanges now operate under tighter regulation, and a bad listing carries real legal, reputational and operational risk for the exchange itself. Delistings used to happen mostly when a project died or was exploited. Today a token can still have an active community and decent daily volume and lose its listing anyway, purely because the exchange has decided the risk is no longer worth the trading fees.

This maps directly onto the state of the broader crypto market. When conditions are strong, exchanges are permissive, listings outpace removals and risk teams move slowly. When conditions weaken or regulatory pressure rises, the same exchanges tighten hard, monitoring tags and ST reviews spike, and marginal projects that would have been ignored in a bull market get flagged. The record Q2 2025 spike and the sustained high floor through the softer 2026 market are exactly this effect in the data.

2.6 What delisting actually costs

The stakes of failing a review are concrete and measurable. Once a token is delisted, only around fifteen percent ever regain a major listing. Roughly sixty-five percent continue only on smaller exchanges with far thinner liquidity. And the price typically falls around twenty-five to thirty percent within hours of the delisting notice alone.

Approximate outcomes for delisted tokens. Source: compiled industry data. Outcomes vary, but the direction is consistent.

Delisting is not a reset a project bounces back from. For the large majority of tokens it is a permanent step down in liquidity, price and credibility.

ST, short for Special Treatment, is a status an exchange applies to a listed token once it decides the pair no longer represents a healthy, tradable market.

It is not unique to one exchange or one name. MEXC calls it ST Warning, following an Assessment Zone period. Bybit and Bitget run the same logic through ongoing internal risk review without a fixed label. Other exchanges use terms like risk tag or delisting watchlist. The mechanism underneath is identical everywhere: the exchange has decided that letting users trade this pair under normal conditions is a liability, to the user experience, to its own risk exposure, or to its reputation, and it moves the token into a separate bucket where either it is fixed within a window, or it is removed.

From the exchange's side this is not an arbitrary punishment mechanism. It is a basic trust-and-safety function. An exchange's core product is the ability for a user to buy and sell an asset at a fair price with reasonable execution. When a pair's order book thins to the point where a modest order moves price by double digits, the exchange is effectively selling its users a bad product under its own brand. Multiply that across enough tokens and it becomes a reputational and even a regulatory problem. Several major exchanges have tightened this exact practice following recent licensing changes in the EU and elsewhere, because regulators now expect exchanges to actively police the quality of the markets they list, not simply collect a listing fee and step away.

Where the causes are covered Open the three-layer diagnostic context Reveal context

A token enters ST status through a combination of three layers, a market layer visible directly in the order book, a structural layer built into tokenomics at launch, and a project layer covering whether the team is still visibly active. The diagnostic framework for identifying which layer is driving your own case is walked through in the live session. This report gives you the thresholds each layer is measured against.

This section is drawn from direct operational experience running projects through each of these exchanges, not from public documentation, most of which stops short of disclosing exact thresholds. It is also where H3 is tested most directly: the correction windows below differ by two orders of magnitude between the most and least forgiving venue.

MEXC

MEXC deserves particular attention. It is the single biggest headache for small and mid-sized projects, both because of how many projects it lists and because of how its monitoring works in practice. The exchange monitors every listed token continuously against a set of hard thresholds, and consistently falling below any of them can trigger an ST warning, after which delisting can follow in as little as three days.

• Daily trading volume, a common working threshold is around 50,000 USDT
• Bid/ask spread kept under roughly 2 percent
• Order book depth maintained across multiple price ranges
• At least regular trading activity, on the order of a trade every hour, with no long dead periods

On top of these visible metrics, the primary and least transparent risk factor is amount of holdings, the volume and number of holders on the exchange itself. The exact threshold is not disclosed publicly, the exchange interprets it at its own discretion, and the request can be difficult to predict in advance. Full disclosure of holder data should be prepared ahead of time rather than scrambled together under deadline.

One practical note that matters for case research: MEXC almost never publishes its delistings the way other exchanges do, so much of what happens there is only visible to the projects directly involved.

Cycle: 30 days in Assessment Zone, then on failure, an ST tag with 3 days to delisting.

BYBIT

A tier-one exchange and one of the most demanding. Following its MiCA licensing, Bybit tightened its approach across all projects, with roughly 370 million potential EU users on the line, policy has become noticeably stricter as a direct result.

• Combined depth of 2 percent on bid and ask sides together, from 2,000 USD
• Liquidity uptime of 98 percent or higher, checked every 10 seconds, maximum downtime of 20 minutes per 24 hours
• Bid/ask spread under 0.5 percent
• Minimum market cap typically from 2 million, used as a working reference point
• General project performance assessed alongside the hard metrics

Exchange communication is weak: messages can be ignored for long periods, and replies are dry or absent entirely. Exiting ST status requires 14 consecutive clean days against every metric, and even then the exchange can hold the exit or add a market cap requirement, minimum 5 million, adjusting the bar as it goes.

BITGET

Considered the strictest exchange overall. Monitoring is approximate rather than exact, but a deviation as small as 1 percent can trigger an immediate response, projects can be delisted for essentially any slip.

• Minimum of 30 orders on each side of the book, spread across levels
• Required balance around 20,000 USD on each side, plus a separate imbalance reserve
• Combined active balance easily reaches 70,000 to 80,000 USD
• Official criteria: trading volume and liquidity, team involvement, project development, smart contract stability, community activeness, project responsiveness, and absence of negligence or unethical conduct

Constant status updates are expected, depth must never lapse, and communication needs to be continuous, the exchange can flag a project simply for appearing inactive in the market. The core practical problem here is the sheer size of the required balances.

KUCOIN AND GATE

A noticeably more lenient regime. Triggers are mostly tied to sharp market events, such as the same token being delisted on another tier-one exchange; otherwise both exchanges are relatively relaxed day to day.

• Annual review: minimum market cap from 4 million
• Bottom 20 percent of the market cap ranking in each cycle is delisted
• Kucoin additionally looks at whether volume is backed by a real holder base

A week of solid work is often enough to correct standing; the cycle is fairly forgiving by comparison. If an exchange decides to actively target a project, however, the tools available are severe, manipulation bots, artificial market pressure, and dumping tokens held in exchange-controlled wallets. Recovery from that specific scenario is considerably harder than the standard case.

BINANCE

Places particular weight on real, demand-driven trading volume and a genuine holder base, and pays close attention to sharp price run-ups immediately before a token loses support. The practical implication is that a real, active community of holders and traders on the exchange matters more here than almost anywhere else.

OKX

Formal requirements exist on order book depth and number of price levels, broadly comparable in spirit to Bybit and Bitget. Exact thresholds require further internal confirmation before this section is finalised for external publication.

BINGX

Sits closer to the Kucoin and Gate end of the spectrum in terms of day-to-day strictness, with comparatively low minimum fund requirements as shown in section 05. Detailed monitoring behaviour is still being compiled and will be added once confirmed internally.

OTHER EXCHANGES

Outside the exchanges named above, oversight is in most cases nominal, active monitoring is minimal, and the priority for these platforms is simply getting a project listed rather than enforcing ongoing metrics.

Recovery costs more than maintaining an already healthy market, and if nobody has said this to you yet, consider this that notice. The cost is not only the monthly market maker fee, it is the actual capital that has to sit in the order book on both sides for the duration of the review.

The table below reflects the internal reference figures used when scoping a recovery engagement: minimum funds needed to realistically exit assessment or ST status on each exchange. These are working minimums based on direct experience, not published exchange requirements, and can move depending on the token's specific situation, price volatility, and how deep the problem already is by the time work begins.

Budget reality Combined minimum is not the full recovery cost Open cost stack

Base

Combined minimum

The working minimum needed to realistically begin a recovery attempt on the selected venue.

Layer 1

Exchange fees

Ongoing exchange fees of roughly 300 to 1,200 USD per month, per pair.

Layer 2

Marketing reserve

A marketing and reward reserve from 5,000 USD, deployed only once the market is stable enough to support attention.

Layer 3

Pressure capital

On stricter exchanges, additional pressure capital: up to 20,000 USD or more on Bybit specifically.

Layer 4

Imbalance reserve

As price moves, the same token quantity costs more or less to defend; this is a normal cost of holding a two-sided book.

5.1 What sits on top of these figures

• Ongoing exchange fees of roughly 300 to 1,200 USD per month, per pair
• A marketing and reward reserve from 5,000 USD, deployed only once the market is stable enough to support attention
• On stricter exchanges, additional pressure capital: up to 20,000 USD or more on Bybit specifically, where manipulation bots or coordinated pressure are involved
• An imbalance reserve. As price moves, the same token quantity costs more or less to defend, this is a normal cost of holding a two-sided book, not a loss

Realistically, a full recovery can run two to three times more expensive than normal operating cost if the process is not tightly coordinated.

Why this cost exists

This is not a penalty paid to the exchange. It is the real cost of turning a tagged token back into a genuinely tradable asset. A healthy market retains community, attracts real volume, and in practice grows and markets noticeably easier once it is out of the zone. What actually prevents the budget from ballooning is coordination, covered in the live session.

The monitoring checklist. Track every one of these continuously, not as a weekly snapshot, exchanges sample your worst hour, not your best.

The instinct in a crisis is rarely the correct action, and a short explicit list is more useful under pressure than a long explanation. Print this page.

8.1 The hypotheses, against the data

8.2 What follows from that

Taken together, the data contradicts the instinct that ST status means a project has failed or was badly built.

Delisting is now a high-volume, rising, market-wide phenomenon that catches otherwise solid projects with real products and real teams, most often through a single point of neglect. An under-resourced market maker. A treasury that did not fund inventory in time. A launch structure nobody stress-tested against an unlock calendar. And increasingly through nothing worse than being a marginal listing at the moment a weak market made an exchange tighten its standards.

The useful way to read this entire report, in light of the numbers, is not as a rescue manual for failing projects but as an operating manual for a stage that a large and growing share of projects, good and bad alike, now pass through at least once.

What separates the projects that come back from the ones that get delisted is rarely the size of the budget or the reputation of the market maker involved. It is coordination: one clear plan, one accountable owner on each side, consistent weekly reporting the exchange can actually act on, and the discipline to avoid the handful of genuinely damaging moves in section 07.

The data says this stage is coming for most projects eventually. This report is the reference layer for being ready for it. The method is the session.

This report gives you the numbers. It deliberately does not give you the method, because a table of thresholds tells you where you stand, not how to get from where you stand to where the exchange needs you to be.

The live session covers the following, in full, with worked examples: