A broad token-market study by Blockworks researcher Carlos found that only a small fraction of crypto tokens beat Bitcoin after reaching scale in public markets. Among tokens whose circulating market cap first closed above $50 million at a month-end between January 2020 and December 2025, just 4.1% outperformed Bitcoin as of June 2026. For tokens with at least 24 months of history, the share drops to 1.7%.
From the point when each token first met that threshold, median returns across the full sample were down 97%, while 73% eventually recorded a drawdown of at least 90%. The report was translated and published by Jinse Finance.
Sample design and scope
The report set out to quantify a market-wide problem often described in the industry as “tokens are broken.” Carlos screened 12,014 CoinGecko listings and built a main sample of 2,114 assets. A token entered the sample the first time its circulating market cap was above $50 million at a month-end, and that month-end closing price became the baseline for later return calculations, peak multiple analysis and 90% drawdown tracking.
The observation period ran through June 2026. Tokens were kept in the sample even if they later stopped trading or were delisted.
A separate market-breadth analysis used a larger dataset of roughly 10,700 crypto-native assets that had at least one month-end market cap above $1 million. Stablecoins, wrapped assets, liquid staking tokens, receipt-like tokens, tokenized real-world assets and duplicate listings were excluded. ETH, for example, was counted once, while WETH and stETH were left out.
Market breadth widened at the bottom and thinned out at the top
Carlos argues that the biggest shift in crypto market breadth over the past six years has been the industrialization of token issuance. Venture-backed projects turned token launches into a repeatable fundraising and liquidity-exit process. Later, Pumpfun and other permissionless launch platforms accelerated issuance further at the small-cap end of the market.
Supply expanded sharply, but liquidity and fundamentals did not keep up. The number of tokens with a market cap above $1 million hit a record 36,488 in December 2024, far above the peak seen in November 2021. The $10 million and $25 million tiers also topped out later. Higher-cap segments moved the other way.
Every market-cap tier above $250 million peaked in November 2021 and never recovered. By June 2026, only 102 tokens remained above $250 million, roughly one-third of the November 2021 peak of 279. Tokens above $1 billion fell to 43, down 64% from the 2021 peak of 118.
The report says all market-cap tiers above $250 million were down more than 60% from their highs, while tiers below $50 million did not peak until December 2024. Using November 2021 as a base index, only the $1 million tier now has more assets than it did then.
Concentration data points in the same direction. Bitcoin’s share of total crypto market capitalization fell from 71% to 42% by early 2022, then rose back to 66%. The top 10 tokens as a group went from 91% to 79% and then back to 91%. By June 2026, the market had largely reversed the diversification seen in 2021.
Bitcoin versus an equal-weighted altcoin market
Across the full period from January 2020 to June 2026, Bitcoin gained 731%, while an equal-weighted altcoin index fell 53%. The clear exception came during the bull market from February 2020 to November 2021, when altcoins rose 904% against Bitcoin’s 508%.
Even then, the gains were concentrated. One-third of tokens beat Bitcoin during that stretch, but the median token still posted a 28% loss in 2021. The median return for the top 10% of winners reached 770%. In the report’s framing, the altcoin boom was real, but it was carried by a small number of outsized winners.
The equal-weighted index assigned the same weight to every qualifying token and rebalanced monthly. Carlos notes that this is a statistical construct, not a portfolio that could necessarily be traded in practice.
After 2022, the gap widened again. The surface-level picture could look like a decoupling, yet the monthly return correlation with Bitcoin actually increased. The correlation coefficient for monthly returns rose from 0.72 in 2020-2022 to 0.78 in 2023-2026, while beta was broadly stable at 1.04 and 0.95. The report says the lowest correlation in fact appeared during the 2020-2021 bull market, the period when the cumulative return lines looked closest on the chart.
What changed was return drift. After beta adjustment, the altcoin index delivered 1.3% per month before 2022, then lost 5.5% per month after 2022, with t = -3.6. Before 2023, altcoins captured 125% of Bitcoin’s upside and 117% of its downside. From 2023 onward, they captured only 56% of Bitcoin’s upside while taking on 165% of the downside.
Most 2020-2021 winners did not hold their lead
The reversal was sharper still for the 187 tokens that beat Bitcoin in the 2020-2021 bull run. Of that group, 86.1% later fell at least 90% from their November 2021 relative highs. By June 2026, their median return stood at -97.6%.
Only one token, OKB, continued to outperform Bitcoin in the cycle after the November 2021 peak. Carlos adds two clarifications. The first is that this measures persistence of outperformance, not lifetime return over a token’s entire history. Some winners from the 2021 cycle still outperformed Bitcoin over their full lifespan, but only OKB kept doing so after the peak. The second is that Bitcoin itself rose just 4% during the post-peak interval used in that comparison.
To test the idea that “altcoins beat Bitcoin in bull markets” as a general rule, the report split market regimes mechanically by Bitcoin’s drawdown from its all-time high at each month-end. A drawdown of less than 25% counted as a bull market, covering 43 months. More than 50% counted as a bear market, covering 16 months. The middle range was treated as a sideways market, covering 19 months.
Across all bull-market months, the equal-weighted altcoin index did not produce statistically significant excess returns over Bitcoin, posting -0.6 percentage points per month with t = -0.14. The two bull cycles looked very different. In 2020-2021, altcoins delivered +9.6 percentage points a month, with t = 1.25. In 2023-2026, they lagged Bitcoin by 8.3 percentage points a month, and that gap was statistically significant, with t = -5.69. Sideways markets showed underperformance of 5.3 percentage points per month, with t = -2.06, while bear markets showed underperformance of 3.4 percentage points, with t = -1.27.
Cohort analysis shows newer token vintages deteriorating faster
The report borrows a framework from credit analysis, treating a drop to 10% or less of the qualification price as a “default,” or a 90% drawdown. Carlos used Kaplan-Meier survival analysis to estimate cumulative default rates by token cohort while dealing with shorter observation windows for recently listed assets.
The pattern worsened with each new vintage. After 24 months on the market, 86% of the 2024 cohort had fallen 90% from the entry price. The corresponding figure was 70% for the 2021 cohort and just 18% for the 2020 cohort.
Across the full sample, 73% of qualifying tokens eventually hit a 90% drawdown, and the median time from qualification to a 90% collapse was only 13 months. Carlos compares the shape of these curves to a steadily deteriorating pool of credit assets.
The 2026 cohort has only a short trading history so far. The earliest assets in that group have been live for five months, and there were 92 qualifying tokens by June, leaving the sample incomplete. Even so, the early trend matched prior cohorts. After three months, 13.5% of the 2026 cohort had already fallen 90%, versus 10.3% for the 2024 cohort and 8.3% for the 2025 cohort at the same point in their lifecycles.
The report also says median lifetime returns for each annual cohort from 2020 to 2024 were -93% or worse. Since 2020, $385 billion worth of new token assets qualified into the market, while capital kept concentrating toward Bitcoin.
Upside compressed while downside stayed in place
Reaching a $50 million month-end market cap does not mean secondary-market investors can easily enter at an early valuation, the report says. Among tokens entering the sample from 2023 to 2026, 12.4% already had a market cap above $250 million at qualification, and 2.4% were above $1 billion. The share of tokens crossing the $50 million threshold within one month of launch rose from 32% in 2020-2021 to 47% later on.
The report’s conclusion is that much of the zero-to-high-valuation gain had already been captured in private or early-stage markets before the public secondary market had a chance to participate.
The payoff profile changed with it. For the 2020 cohort, the median token eventually reached a peak price 5.1x above its entry baseline. A total of 51% at some point offered a 5x exit window, and 33% reached 10x. For tokens entering from 2023 to 2026, the median peak multiple was only 0.93x, meaning the median month-end closing price never rose above the qualification price. Just 3.9% ever posted a 5x gain, and 1.4% reached 10x.
Final outcomes on the downside barely changed. Median final returns in both periods were close to -95%. What disappeared, according to Carlos, was the asymmetric payoff structure that had once supported token speculation.
Even tokens that launched at very high initial valuations did not show meaningful extra upside. After 2023, 26 tokens with launch valuations above $1 billion had a lower probability of falling 90% than assets in the $50 million to $100 million range, but their median peak multiple was still only 1.0x.
Out of 955 qualifying tokens since 2023, only 37 ever delivered a 5x move from baseline, and roughly 13 reached 10x. Most 10x winners were meme coins, including PEPE with a peak 44x move, along with WIF and BONK. VIRTUAL and HYPE were among the few non-meme exceptions. The 5x to 10x group also included infrastructure and DeFi names such as Rollbit, Pendle, Celestia, Ondo and Sei.
The bar for reaching market leadership rose sharply
Carlos says the market-cap threshold for becoming a meaningful player increased substantially during the sample period. In June 2020, the 100th-largest token had a market cap of $47 million, below the report’s $50 million qualification line. By December 2024, entry into the top 100 required roughly $900 million, and entry into the top 20 required more than $8 billion.
That left later entrants starting much farther from the market leaders. The report describes the gap as almost 20 times wider than before. Few newer tokens were able to close it and become lasting leaders.
It highlights a small set of exceptions. SHIB stayed in the top 20 for 49 months through meme-driven retail distribution. TON remained there for 36 months on the back of the Telegram ecosystem. HYPE entered the top 20 after qualifying in November 2024 and stayed there, supported by its status as a leading perpetual DEX, material business revenue and buybacks.
Older assets launched before 2020 held at least 10 of the top-20 slots throughout the period. Tokens minted after 2020 never occupied more than five slots at once.
Factor analysis: momentum and volatility turned into warning signs
The report applied a traditional asset-pricing framework to tokens, testing whether market cap, momentum, volatility and turnover could predict future returns. Each month, tokens were sorted by those characteristics and later returns were compared across the highest and lowest buckets.
On market cap, the full sample showed no statistically significant return gap between large-cap and small-cap tokens in either 2020-2022 or 2023-2026. A small-cap premium appeared only when the sample was restricted to coins that had made historical public top-200 lists, which the report says introduces survivorship bias by excluding 502 tokens that never reached the top 200.
Momentum produced the clearest result. After 2022, tokens with the strongest returns over the prior three months underperformed the weakest performers by 3.8 percentage points per month, with t = -3.52. The top return quintile lost 87% on a cumulative basis, while the bottom quintile lost 50%. The same reversal held when the lookback window was changed to one month or six months.
The effect was concentrated in mid- and small-cap assets. Once the sample was limited to tokens above $100 million in market cap, the return spread narrowed to 1.1 percentage points a month and lost statistical significance, with t = -0.72. Carlos therefore describes momentum reversal as a smaller-cap phenomenon rather than a rule for the entire market.
Fama-MacBeth multivariate regressions supported that view. In 2023-2026, a one-standard-deviation rise in momentum predicted a 1.2 percentage point drop in monthly return, while a one-standard-deviation rise in volatility predicted a 0.6 percentage point drop. Market cap and turnover were not significant. In the report’s summary, only momentum and volatility had predictive power after 2023, and both pointed to weaker forward performance.
The small set of tokens that beat Bitcoin
Only 81 tokens, or 4.1% of the sample with at least six months of history, outperformed Bitcoin over their own observation windows. The comparison used month-end prices for both assets, and a token counted as an outperformer only if its trading volume at the last month-end in the window was at least $10,000.
The report says the 4.1% figure is still inflated by recently listed tokens with short histories. Among 1,305 tokens with at least 24 months of observation, just 22, or 1.7%, beat Bitcoin.
Exchange tokens stood out. They made up 32% of the long-run outperformer sample despite accounting for only 2.1% of the full long-history universe, a 15-fold overrepresentation. BNB, OKB, GT, LEO, BGB, WBT and MX all beat Bitcoin. Seven of the 27 centralized exchange tokens in the sample outperformed, a 26% hit rate, versus about 1.2% for all other tokens combined.
CAKE from PancakeSwap was the only qualifying DEX token. Carlos says it shared the same core characteristic as exchange tokens: fee revenue feeding ongoing buybacks and burns.
The rest of the outperformers included older assets such as ETH, SOL and TRON, plus a small number of meme coins and cycle-specific leaders. The report cautions against reading those exceptions as strength for an entire sector. Within meme coins, for example, the median return across 275 qualifying tokens was -97%, and the odds of beating Bitcoin were slightly below the market average.
Methodology and robustness checks
The appendix lays out the mechanics in more detail. The main sample included 2,114 tokens that had at least one month-end circulating market cap of $50 million or more between January 2020 and June 2026. The larger market-breadth set came from roughly 54,000 listing records, around 36,000 of which were inactive or delisted.
Returns were measured from month-end close to month-end close. Observations with data delays longer than 10 days were excluded. To count as beating Bitcoin, a token needed trading volume of at least $10,000 in the final month of its comparison window. Carlos says this rule removed three dormant delisted tokens whose stale prices had created false outperformance and reduced the six-month outperformance rate from 5.2% to 4.1%. Peak-multiple calculations used the same rule and only counted month-end highs with real trading volume.
The equal-weighted index rebalanced monthly and used 1%-99% winsorization on monthly returns to limit interference from dirty microcap data. Reported t-statistics used Newey-West standard errors with a lag length of 3.
On data quality, the screening removed 106 wrapped, receipt-like, pegged, duplicate or wash-traded assets and excluded 602 erroneous monthly return observations tied to obsolete tickers. Another 91 assets with suspected redenomination errors were removed from the Bitcoin outperformance statistics and altcoin cycle backtests.
Several robustness checks did not materially alter the conclusions. Raising the entry threshold to $100 million still produced a median return of -97% and a Bitcoin outperformance rate of 4.3%. Raising it to $250 million left the median return at -97% and the outperformance rate at 3.8%.
Delaying the starting point by six months yielded a median return of -94% and a Bitcoin outperformance rate of 3.5%. A 12-month delay produced -91% and 3.6%, respectively. Ending the sample in December 2024, which the report describes as the strongest month of the current cycle, still left the median token return at -81%, with 82% of assets trading below their qualification baseline.
Changing the bull-bear cutoffs to 20%/40% or 30%/60% did not shift the headline result. Altcoins outperformed by about 8 to 10 percentage points per month in the 2020-2021 bull market, then lagged by 8.0 to 8.3 percentage points per month in the 2023-2026 bull market, with t-statistics between -5.2 and -5.8.
Under different winsorization bands, total return for the equal-weighted altcoin index ranged from -30% under 0.5%-99.5% winsorization to -53% under 1%-99% and -71% under 2%-98%. Every version still trailed Bitcoin’s 731% rise by a wide margin.
Momentum reversal also held across different lookback windows and liquidity filters after 2022. With one-, three- and six-month lookbacks, t-statistics ranged from -3.1 to -3.5. With prior-month turnover thresholds from 0.1% to 1%, t-statistics remained between -2.1 and -2.5. The report adds that the effect stayed concentrated in the smaller half of the market and disappeared once only tokens above $100 million were kept.
Implications for allocators
Carlos says the conclusions apply only to the period after a token reaches the secondary market and first clears the $50 million month-end circulating market cap threshold. They do not capture private-market entry costs, unlock schedules or gains realized before and around listing.
One implication is that Bitcoin should be the default benchmark. Only 4.1% of tokens beat Bitcoin over their observed windows, and only 1.7% did so when at least 24 months of history were required. Broad token baskets, the report says, have historically compared poorly with simply holding Bitcoin.
A second point concerns position sizing. Median post-qualification return was -97%, and 73% of tokens eventually suffered at least a 90% drawdown. Near-wipeout outcomes, in this framework, are normal rather than tail-risk events.
The report also argues for caution around newly listed projects. Each cohort deteriorated faster than the one before it. The median time from qualification to a 90% drawdown was only 13 months, while the available upside after listing shrank sharply.
Another conclusion is to avoid chasing recent winners. Since 2022, high momentum and high volatility have both pointed to weaker future returns. The tokens attracting the most attention in the market repeatedly went on to lag the names that had performed poorly beforehand.
On diversification, the report says a basket of tokens may reduce single-name risk but offers limited protection against system-wide crypto risk. For investors capable of doing deep project work, a concentrated portfolio built on higher-conviction ideas may make more sense than a broad spread of token exposure. Effective diversification, in Carlos’s view, still has to come from outside crypto.
The final portfolio takeaway is that cash flow and tokenholder alignment should carry more weight. Exchange tokens were heavily overrepresented among long-run Bitcoin outperformers, and the common thread was recurring fee income combined with buybacks. The report says more work is needed to test whether buybacks translate into stronger returns across DeFi protocols more broadly.
Carlos’ closing remarks
Carlos writes that he began the project intending to measure the impact of token buybacks. After going through the return data, a larger result became clear: the figures quantify a broad collapse in expectations among secondary-market token investors over the past few years.
Viewed from the outside, he says, the asset class has become difficult for value investing. The baseline outcome for newly qualifying tokens is a drawdown of more than 90%, with only a tiny share beating Bitcoin. Against that backdrop, he writes, it is hard to justify broad token exposure over simply buying and holding Bitcoin.
Still, the report does not frame the category as beyond repair. Carlos points to efforts such as the Blockworks token transparency framework and MetaDAO ownership tokens as early attempts to improve disclosure and strengthen tokenholder rights. The aim of the report, he says, is to quantify the severity of the problem, not to declare the end of tokens as an asset class.
He closes by saying he remains optimistic about the future of tokens, but believes the investable set has narrowed sharply. Finding the small number of exceptions now requires more discipline than in prior cycles, while the industry still needs time to work through structural problems left behind by the previous era.