Seventy-two percent.
That is the institutional share of Wintermute’s spot OTC flow for the first half of 2026. One desk. One proprietary dataset. And yet, when the same firm says the next altcoin season will have fewer winners, the market treats it not as a forecast but as a verdict. Institutions have already chosen where the liquidity goes. The crowd just has not caught up.
I have spent the better part of a decade treating market claims the way an auditor treats a transaction ledger: trace the source, question the assumptions, find what the summary numbers hide. In 2017, I spent twelve hours dismantling an ICO contract because its 1000% APY promise looked like a hallucination. I found a reentrancy path in the withdrawal function. The project never listed on a major exchange. In 2022, I spent three weeks cross-referencing an exchange’s internal SQL with on-chain records to trace $400 million in misallocated funds — not because the database was wrong, but because it was designed to make the truth invisible through internal consistency.
Apply that same posture to Wintermute. "The next altseason may have fewer winners" is not a price target. It describes a change in the plumbing of the market. This is what the plumbing looks like.
Context
Wintermute is a digital asset market maker and over-the-counter liquidity provider founded in 2017 by Evgeny Gaevoy, a former high-frequency trader. It operates in the institutional layer of crypto: block trades, stablecoin logistics, algorithmic market making across more than one hundred venues. This is not a retail brand. It is the plumbing through which large orders move without breaking the exchange books.
The firm has institutional scars. In 2022, Wintermute’s DeFi wallet was exploited for roughly $160 million — the largest hack of its kind at that time. The firm absorbed the loss and kept making markets. That background matters. It tells you Wintermute has direct experience with the precise failure mode the market is being weaned from: uncontrolled leverage, unaudited contracts, and the assumption that "the chain will keep running."
The 2026 market is a market that has matured in all the ways that crypto natives claim to want and then discover they hate. Regulatory clarity has arrived piecemeal. Bitcoin and Ether are treated as commodities in the United States; a long list of altcoins sit in legal gray zones. Exchange-traded products collect the majority of reported institutional flows. Derivatives open interest is dominated by BTC and ETH. The retail era of "everything pumps" is no longer the default. It has been replaced by a capital rotation system with narrow DNA.
The macro frame is doing more work than any technical indicator. Over the past decade, the same institutions that would have bought a mid-cap altcoin with a two-year horizon in 2021 now hold a T-bill that pays a yield far above the historical average of crypto’s own risk-free proxies. The generation of money that created the 2021 altseason — zero interest rates, fiscal transfers, a wall of stablecoin issuance — is not the generation of money managing institutional allocations in 2026. Risk budgets are smaller. Carry is cheaper. Capital that can earn an honest yield on collateral without touching the token market has almost no reason to fund a speculative altcoin rotation. "Fewer winners" is, in that sense, a macro statement wearing market-structure clothing.
One more layer before the analysis: every previous altseason was a breadth event. In 2021, liquidity spilled out of Bitcoin and Ether and flooded the long tail of the market. That broadness was visible in social mentions, DEX fees, chain gas usage, and the sheer number of tokens taking 10x jumps. Broadness was not a side effect of the bull market. It was the mechanism. The upside that attracted retail existed because low-cap assets had no prior price history and no institutional custody layer.
Wintermute’s numbers suggest the mechanism has split. During the first half of 2026, institutions represented 72% of their spot OTC flow. If you believe that number is representative — and I will interrogate it shortly — the trading population has inverted. A population of institutions does not rotate into vapor. It rotates into collateral quality.
That is the phrase I keep returning to: collateral quality. The entire "next altseason" discussion, when viewed through an institutional lens, is a conversation about what can be pledged, borrowed against, or sold without slippage. The fewer the assets that satisfy that test, the fewer the winners.
No table can fully capture the shift, but this sketch gets close:
| Cycle factor | 2017 ICO season | 2021 DeFi/NFT season | 2025–2026 structural season | |---|---|---|---| | Primary capital source | Retail via exchange listings | Retail + crossover funds | Institutional OTC + ETFs | | Market breadth | Extreme | Extreme | Narrow | | Token supply discipline | None | Low | Priced as a risk factor | | Regulatory filter | Almost nil | Moderate | Decisive | | Main driver | ICO mania | Zero rates | Rates + regulation + custody |
Core: The Mechanism of Fewer Winners
What OTC flow actually measures
Every OTC trade is a negotiation that never marks the exchange chart. Party A wants ten million tokens without moving the price. Party B wants to sell ten million tokens without leaking intent. They agree on a size and a premium or discount to the public market, and the trade settles using a mixture of internal inventory and external venues.
The chain does not record this. The ledger lives in a private database.
That is why the 72% statistic matters in a way on-chain analytics cannot replicate. On-chain analysts see the confirmation, not the intent. OTC desks see the intent before the confirmation. When an institutional client commits to a block of an altcoin, they are not trading momentum. They are establishing a position at a size that will be impossible to repeat later on thin books. OTC flow is pre-price. It tells you where institutional allocation is heading before the market post-rationalizes it.
Still, treat the number with suspicion. I learned that after FTX. I spent three weeks cross-referencing the internal SQL database of a mid-tier exchange with on-chain transactions. The database was internally consistent. Every field validated. The problem was not the data. The problem was that the exchange’s profit model assumed certain positions were real assets when, in fact, those positions were locked inside the exchange’s own token. The SQL was not lying. The economic model was.
The same discipline applies to OTC data. A family office managing $200 million is not an institution by every definition. A mid-tier fund that is legally an institution but behaviorally an aggregator of retail mood may be categorized differently across desks. The metric is useful. It is not neutral.
And yet — this is the part most retail commentary misses — even a biased 72% reveals directional truth. If the largest OTC franchise in crypto sees four out of five dollars arriving from institutional wallets, the capital being allocated to this asset class carries institutional constraints: compliance approval, custody requirements, regulatory vetting, and due diligence on the team and the token schedule. That last constraint filters the market to a much smaller set.
There is also a nuance that gets lost when reading OTC flow as directional sentiment. Not every OTC trade is a buy. Institutions use OTC desks for inventory sourcing, market-making replenishment, and basis trades. A market maker buying a token OTC may be filling a client order while remaining flat — capturing the spread, not expressing a view. But that does not weaken the concentration thesis. Even non-directional flow is concentrated in assets with enough liquidity to support the machinery. You cannot run a basis trade on a token with $20,000 daily volume. You can run it on a token with $200 million daily volume. This mechanism silently removes the tail from institutional consideration without anyone explicitly deciding it is under-competitive.
The supply cliff: 2026 is the unlock year
The real story is not demand. It is supply.
2026 is the unlock year for the 2021 and 2022 venture capital era. Funds deployed at record valuations into projects with multi-year vesting schedules. Those lockups are now hitting the market at a time when the marginal buyer is an institution that will only buy assets with predictable emissions.
Go back to 2017. I was a junior developer in Hangzhou, staring at a Solidity contract for a project called GlobalToken. The contract had a classic reentrancy bug in its withdrawal function. I found it, and the finding was real. But the deeper problem was the token distribution: massive team allocation, no vesting, and a whitepaper making mathematically impossible yield promises. The smart contract bug was a symptom. The token schedule was the disease.
I did not fully understand that distinction at the time. Now I realize a bull market is simply a race between emission rate and absorption rate.
In 2026, the emission rate is spiking for a large portion of the mid-cap universe. Consider an asset with 15% of supply circulating, a multi-billion-dollar fully diluted valuation, and 40% of supply unlocking this year. That asset can only go one direction if the buyers are risk-constrained.
The market does not need to stop believing in altcoins for this thesis to work. It only needs to notice that holding a token through its unlock cliff is an asymmetric risk.
That is why "fewer winners" is structural. The set of tokens with clean supply schedules — high float relative to FDV, predictable emissions, unlock events largely in the rearview mirror — is finite. In previous cycles, retail did not check the schedule because the liquidity environment absorbed unlock pressure. In 2026, the steady stream of new money has turned into a selective stream.
Code does not lie, but it does hide. In an ERC-20 audit, you check the standard functions: transfer, approve, mint. You also check the token lock contract, the vesting cliffs, the emissions. Most retail investors read the audit report, not the schedule. The schedule is the part of the code that actually decides whether a token is structurally solvent.
The same reasoning applies to protocol treasuries. A treasury full of its own token is not capital. It is a claim on future exit liquidity. When the unlock window opens, the market treats it as inventory to be sold, not value to be held. Institutional capital sees this immediately. That is why so many mid-cap tokens trade at levels that look "cheap" to retail while institutions refuse to touch them. They are not cheap. They are un-liquid.
The low-float fallacy
There is a specific math error embedded in the altseason narrative. Retail looks at fully diluted valuation and sees a market cap. Institutions look at the float and see exit capacity.
Take a token with a $5 billion fully diluted value and only $500 million circulating. Retail sees a $5 billion asset trading at $25. An institution with $100 million to deploy cannot buy even a quarter of the float without pushing the price against itself. The token looks liquid until a real asset allocator tries to enter.
What matters is not the theoretical market cap. It is the dollar volume that can be sold without breaking the price. Most distributed-ledger assets today have exit capacity far below the minimum size an institution requires. So institutions stay out. The few assets with sufficient exit capacity attract the capital.
This is not a forecast. It is an accounting identity. The "winner" set, defined by institutions, is the set of tokens where a large allocation does not immediately become a price-impact event. The size of that set is small. It has been small for years. The only reason it did not matter earlier is that the marginal capital was retail, which enters in small increments and tolerates fragmentation. Institutions do not.
The compliance filter: how legal teams choose the winners
Institutional capital has a lawyer attached. This is the layer that retail optimists routinely ignore.
In the United States, the Howey test determines whether an asset is a security. Bitcoin and Ether are treated as commodities under the CFTC’s existing framework. A set of other tokens has been litigated, settled, or flagged. The rest of the market — thousands of live tokens — carries enough regulatory uncertainty that a US-regulated fund cannot touch it without risking its license.
For an institution, the tradeable universe is not the 10,000 tokens on CoinGecko. It is a list of perhaps a few dozen assets: the large-cap layer; a handful of exchange-native tokens that survived enforcement attention; the utility tokens whose networks generate real revenue and carry a clear legal posture; and a small set of assets that went through the expensive, painful process of engaging with regulators.
This creates a two-tier market. Tier one: compliance-grade assets with institutional depth. Tier two: everything else, living off retail flows and market-maker inventory. The 72% institutional OTC flow is mostly a tier-one phenomenon. The list of tier-one tokens is short. That is a legal consequence, not a short-term opinion.
The phrase "trust is a variable, not a constant" has a specific meaning here. Institutions do not trust token teams; they trust the framework around the token. They trust the ability to custody it, to hold it without triggering a securities violation, to sell it in a crisis. Very few tokens satisfy all of those conditions. That small fraction is what "fewer winners" describes.
There is also the DAO problem, which I have written about for years. Most DAOs have the legal status of "no legal status." When a protocol fails, members can face personal liability, and the asset holder has no legal entity to sue. Institutional lawyers know this. They do not need to say it out loud. They simply mark the asset as un-investable. This is one of the quiet reasons why so few decentralized protocols receive institutional allocations regardless of their on-chain success. A token governed by a legal entity, or bundled into a regulated product, has a structural advantage. That advantage is not about the tech. It is about who can buy the asset without committing malpractice.
Corroboration: three data sources, one conclusion
Wintermute’s thesis should not be accepted on its own data alone. I checked whether independent sources confirm the concentration narrative. They do.
Deribit, the largest crypto options venue, reports that BTC and ETH combined account for over 90% of total open interest in crypto derivatives — a pattern that has held since late 2024. Consider what that means. Options traders are institutions and sophisticated players who price professional risk. They are expressing their exposure through two assets. They are not building large positions in the long tail.
CoinShares flow data tells a parallel story: BTC-related products capture the overwhelming majority of institutional fund net flows in 2025 and 2026. A modest slice goes to Ether. Everything else is effectively noise in the aggregate.
Custodian data reinforces the pattern. When an institution opens custody with Coinbase Custody, BitGo, or Fidelity Digital Assets, the supported-asset list is short. It does not grow at the same pace as the retail token market. The practical result is a settled hierarchy: Bitcoin and Ether are held at scale, a small number of blue chippers are held in smaller size, and the tail does not exist for most balance sheets.
Combine the three datasets: - Wintermute: 72% institutional OTC flow. - Deribit: over 90% of options open interest in BTC and ETH. - CoinShares: over 90% of managed fund flows into BTC.
Whatever bias exists in Wintermute’s own sample, the concentration thesis does not rely on one desk. It is a structural fact across market infrastructure. Every exit liquidity event is a forensic scene. When a hype cycle ends, examine what exited cleanly: which assets had a real bid beneath them? The assets that pass institutional filters, that have clean schedules, that are booked by market makers on two continents — those find bids. The rest gap down into a vacuum. After you review enough of these scenes, you notice the vacuum was always there. In the bull phase, you simply were not looking.
The retail question: where did the rotation go?
The "fewer winners" thesis is often read as a statement about demand. It is also a statement about the absence of retail rotation.
In earlier cycles, retail capital flowed down the market-cap ladder in a predictable sequence: Bitcoin rallies. Ether rallies. Top-tier alts rally. Then the middle tier. Then the long tail. That rotation was fueled by a simple psychological mechanism: the fear of missing out on the next 10x.
That mechanism still exists, but it has been redirected. In the 2024–2025 phase, retail capital went toward memecoins and the casino layer — assets with no fundamental schedule, no institutional custody, and no legal framework. This is not an altseason in the traditional sense. It is a parallel market running on rails entirely separate from institutional OTC flow. The capital that might have rotated into a broad range of mid-cap tokens was instead absorbed by the fastest lottery-ticket layer the market has ever built.
I have seen the 2026 version of this risk firsthand. Earlier this year I audited an AI-agent platform whose reinforcement-learning model was able to self-elevate privileges in its deployment scripts. The team was impressive. The token was not. The wealth-creation mechanism for that asset class remains the same as 2021, except the clock speed is faster and the audit trail is thinner. This is precisely the kind of asset that will never appear on an institutional OTC desk.
The result is an unusual split. Institutional capital is concentrated in a small quality bracket. Retail speculative energy is concentrated in a small fictional bracket. The middle of the market — the "real altcoin" layer — receives the least attention from both.
That split explains why Bitcoin dominance remains high and why market breadth measures stay weak even during apparent rallies. It also explains why "the altseason" feels delayed. It is not delayed. It is being re-routed.
The infrastructure paradox: delivery does not equal demand
There is a common objection to the concentration thesis: "What about all the infrastructure? Layer-2s, data availability layers, restaking, intent-based routing? The technology is finally ready."
I have spent years auditing systems in this space. The technological progress is real. But a hard truth remains: infrastructure delivery does not equal token demand.
Take the data availability layer. It was supposed to become the next SaaS market for rollups. But the framing was always suspect. The market built infrastructure for a scale of usage that has not arrived. Most rollups do not generate enough data volume to justify a dedicated DA network. I said this before the current cycle, and nothing in the 2026 data has changed my view. The infrastructure was optimized extensively — and optimization is just risk wearing a disguise. It hides the fact that underlying economic demand remains uncertain.
The same applies to restaking, to AI agents, to tokenized real-world assets. The technology matures. The revenue does not necessarily follow. Institutions see this clearly because they run the operating models. If a protocol’s net revenue is a small fraction of its fully diluted valuation, it does not matter how elegant the engineering is.
This is why "fewer winners" has a layer that is often missed: it is not just about capital flows. It is about the absence of the next killer use case. The 2021 altseason was driven by DeFi yields and NFT speculation — real, if unstable, user demand. In 2026, the user-demand engine is dominated by stablecoin settlement, derivative trading, and tokenized treasury products. All three are institutional tools. None of them require thousands of altcoins to succeed.
The self-fulfilling cycle: when a verdict becomes infrastructure
Here I will make an uncomfortable observation about my own industry.
A market maker’s prediction is never purely a prediction. When a large market maker publicly states "fewer winners," the statement triggers internal action across the ecosystem. Market-making desks tighten inventory on the long tail and concentrate inventory on top assets. Exchange listing teams deprioritize listings without institutional demand. Token teams redesign their tokenomics to mimic blue-chip patterns. Retail traders, reading the same headline, sell the tail and buy the head.
The result is that the prediction becomes a force field. The market confirms the claim before the claim has been independently tested against future outcomes. This is not conspiracy. It is the mechanics of reflexive markets.
I have seen the same pattern at the contract level. An oracle is supposed to describe the world, but the world starts to move toward the oracle. During the 2020 Bancor v2 post-mortem, I isolated the root cause in oracle latency: the price feed lagged actual market conditions, and arbitrageurs used the gap to drain liquidity. People blamed the protocol design. I traced it to the oracle.
The narrative around "fewer winners" is behaving like that oracle. It was generated from real data. Now it is embedded in the market’s decision-making apparatus, and once embedded, it becomes part of future price formation.
Audits verify intent, not outcome. A protocol can audit every contract, document every upgrade, and still suffer from a market structure that makes its token un-investable. In the same way, a market commentary can present every number accurately and still become the engine of the outcome it predicts. The mechanism matters more than the intention.
A note on the source
Let me keep the forensic record balanced. The 72% figure requires context.
Wintermute’s OTC desk is not a random sample of the market. It is a specific venue with its own onboarding process, minimum ticket sizes, compliance team, and geographic footprint. An OTC desk with a high minimum trade size naturally filters out retail — not because retail is absent, but because the venue was not designed for retail. The customer mix will always skew institutional.
This is the representativeness caveat that anyone using this number should state. The 72% figure does not prove that 72% of all crypto trading is institutional. It describes one venue’s population.
But this does not invalidate the core claim. Deribit and CoinShares are not subject to the same venue bias. They capture the options market and the fund-flow market, respectively, and both show the same concentration. In my judgment, the conclusion is robust: the trend is real.
I must also flag a potential conflict. Wintermute is a market maker. It profits from volatility and flow, not from long-term directional popularity. A concentrated market with high volatility in a small set of assets is commercially comfortable for a sophisticated market maker. "Fewer winners" might not be a warning. It might be a preference stated as a forecast. That is why cross-validation matters. The independent data does not carry Wintermute’s P&L.
Contrarian: What the Bulls Get Right
Let me steelman the other side before closing. The altseason thesis is not dead, and the bulls have real points.
Start with a definitional point: "winner" is a moving target. If 100 coins are tradeable and institutional capital concentrates into 15, those 15 can appreciate more violently than they would in an environment with 100 winners competing for the same flows. A concentrated market is not necessarily a weak market for the chosen ones. It can be a stronger market for the select list — larger allocations, deeper liquidity, less crowded social volume. The "fewer winners" thesis may be correct, and the winners it describes may still be enormous.
Then there is the tail argument. The unlock cliff is an average, not an absolute. Many mid-cap projects have already completed their unlock schedules. Some generate real revenue. Some are trading at distressed valuations precisely because the market treats all tail assets as contaminated. A fully unlocked, revenue-generating protocol with no remaining VC dilution is a different asset class from a high-FDV, low-float token. The market is not pricing that distinction properly. If "fewer winners" is correct, this is exactly where the next generation of winners will be found.
A more technical objection: OTC data is about OTC flow. It says little about the retail-native layer of the market: memecoins, social tokens, and the DEX casino. That layer does not go through institutional OTC desks. It happens on chain, through aggregator frontends, in pools that last hours. When we say "altseason," we usually mean the entire non-BTC, non-ETH universe. But the universe has split into an institutional sub-universe and a retail sub-universe. Wintermute’s data describes one of them. A memecoin season can occur even while institutional OTC flow remains concentrated on BTC and ETH. Whether that counts as "winners" depends on the trader’s definition.
Institutions are also not static. I have audited AI-agent platforms that autonomously deploy smart contracts, and I have seen how quickly a market structure changes when a new type of actor enters. The same will be true for institutional crypto: the tradeable list can expand. If a crypto ETF wrapper is approved for a broader basket of assets, the "fewer winners" list widens. If regulators provide clarity for tokens previously in gray zones, the tradeable set expands. The 72% number describes today’s list, not tomorrow’s.
And this is the point I respect most: markets resist their own prophecies more often than linear models assume. Narratives get absorbed, but outliers get discovered. The institutions that win the next cycle will not be the ones copying the consensus list. They will be the ones doing the forensic work on the tail — the fully unlocked asset with an actual revenue line, the protocol with clean legal status that no one talks about because the metrics look boring. The prophecy of "fewer winners" is only binding if everyone believes it. Markets are rarely that obedient.
Takeaway
The next altseason will not be announced on social media. It will be visible in OTC settlement data, custodian inflows, derivatives open interest, and the quiet expansion of market-maker inventory lists. The season most retail expects — the week when every low-cap chart goes vertical — is structurally unlikely. Not because the altcoin concept is broken, but because capital has been repriced for a different game.
That game rewards assets with clean schedules, clear legal status, and liquid multi-venue depth. It punishes everything else, with selective and occasional exceptions.
The data from Wintermute, Deribit, and CoinShares tells one story from three vantage points: the largest capital sources moved from gambling on narrative to underwriting collateral. That is the change that broke the broad altseason. Not a crash. A funding inversion.
I would not bury the tail entirely. I would investigate the fully unlocked, revenue-backed, legally defensible corner of the market — the assets where the forensic question, "what would a disciplined buyer pay for this?" yields an answer the market has not yet found.
But I would also not dismiss the prophecy. It is already being absorbed into the market’s wiring. Every desk that tightens its tail, every exchange that delays a listing, every optimizer that filters by "top 20 by volume" is another line of code in the prediction’s own implementation.
The chain remembers what the ledger forgets. OTC trades never touch the chain. They leave no on-chain footprint. The positions they create are the pressure behind the next cycle. When the chain looks calm, that is often when the institutional layer is repositioning for the next chapter.
I do not know the next top tick. I do know the structure. "Fewer winners" is not a forecast. It is a toll booth. The pass is expensive, and it narrows as you approach the gate.