Introduction

“Tokens are broken” has become shorthand for one of the industry’s most persistent structural problems. Anyone active in crypto, especially liquid token investors, understands it intuitively. We all know the classic examples: the token-equity split, low-float and high-FDV launches, aggressive unlock schedules, inflationary emissions, incentive-driven demand, and value accruing everywhere except to the token itself. What has been missing is a market-wide measure of how severe the deterioration has become.

This report attempts to do that. We identified every token that closed a month above $50M in circulating market cap from January 2020 onward, producing a universe of 2,114 assets from a 12,014-coin CoinGecko audit that includes inactive listings. A token qualifies at the first month-end when its circulating market cap exceeds $50M. Its qualification price becomes the starting point for measuring subsequent returns, peak upside, and the incidence of 90% losses. Each asset is then tracked through June 2026, remains permanently in the sample, and is not removed if it later becomes inactive or delisted.

For market-breadth analysis, we use the broader listing record of approximately 10.7k crypto-native assets that closed at least one month above $1M. Stablecoins, wrapped assets, LSTs, receipt tokens, tokenized RWAs, and duplicate listings are excluded throughout. For example, ETH counts once; WETH and stETH are excluded.

Market Breadth

The most consequential development in crypto market breadth over the past six years has been the industrialization of token issuance. Venture-backed projects turned token launches into repeatable fundraising and exit liquidity events, while Pump.fun and other permissionless launchpads later accelerated issuance among smaller tokens. Token supply expanded dramatically, but the liquidity and fundamentals needed to sustain it did not.

The number of tokens worth more than $1M reached a new all-time high in December 2024, at 3,648, well above the November 2021 peak. The $10M and $25M tiers peaked alongside it, while higher-value segments moved in the opposite direction. Every tier from $250M upward peaked in November 2021 and never recovered. As of June 2026, only 102 tokens are worth more than $250M, matching the lows of the 2022-23 bear market despite BTC being far above its 2022 levels, while the $1B club has fallen to 43 assets, 64% below its 2021 peak of 118.

The timing of each peak highlights the divergence. Lower market cap tiers continued expanding well beyond 2021, while the upper end of the market never recovered. Indexed to November 2021, only the $1M tier is larger today. 

Concentration data reinforces the same point. BTC’s share of total universe capitalization fell from 71% to 42% in early 2022 before recovering to 66%, while the top 10’s share moved from 91% to 79% and back to 91%. By June 2026, the market had almost entirely reversed its 2021 broadening. 

BTC vs. the Equal-Weighted Altcoin Universe

From January 2020 through June 2026, BTC gained 731% while the equal-weighted alt index lost 53%. The only clear exception was the February 2020 to November 2021 bull market, when alts gained 904% against BTC’s 508%. During this period of exuberance, 1 in 3 tokens outperformed BTC. Even then, returns were heavily skewed: the median token lost 28% in 2021, while the top decile gained a median 770%. “Alt-season” was real, but concentrated in a small group of extreme winners.

Of note, the equal-weighted index assigns the same weight to every token and rebalances monthly. It is a statistical measure rather than a tradeable portfolio; see Methodology for more details.

The widening gap after 2022 may look like decoupling, but the data shows the opposite. The correlation of monthly returns between the equal-weighted alt index and BTC rose from 0.72 in 2020-22 to 0.78 in 2023-26, while beta remained broadly unchanged at 1.04 and 0.95. In fact, correlation was lowest during the 2020-21 bull market, when the two cumulative lines appeared closest together on the chart above.

What changed was drift. On a beta-adjusted basis, the alt index returned 1.3% per month through 2022 and lost 5.5% per month thereafter (t = −3.6). Before 2023, alts captured 125% of BTC’s average upside and 117% of its downside, roughly the leveraged exposure many investors expected. Since 2023, they have captured only 56% of the upside and 165% of the downside. 

The reversal was even more severe among the 187 tokens that outperformed BTC during the 2020-21 bull market. Of those winners, 86.1% subsequently lost at least 90% from their November 2021 price, with a median return of −97.6% through June 2026. Only one, the exchange token OKB, outperformed BTC over the same post-peak period.

Two clarifications so this statistic is read correctly.

1. This measures persistence rather than lifetime performance. Several 2021 winners still outperformed BTC over their full histories, but only OKB continued to outperform from the November 2021 peak onward.
2. BTC gained just 4% over the post-peak window, so the 2021 winners underperformed a benchmark that was itself broadly flat.

To test whether “alts outperform in bulls” generalizes as a regime rule, we applied a mechanical classification. Classify every month by BTC’s distance from its running all-time high at month-end: bull when BTC is within 25% of its high (43 months), bear when it is more than 50% below (16 months), chop in between (19 months).

Across all bull months, the EW alt index shows no reliable edge over BTC at all (−0.6 pp/month, t = −0.14), because the two bulls were opposites. In the 2020-21 bull months, alts beat BTC by +9.6 pp per month (t = 1.25, wide dispersion); in the 2023-26 bull months they lost 8.3 pp per month, and reliably so (t = −5.69). The lesson is uncomfortable for the "alt season is coming" trade: alt outperformance was never a bull-market property. It was a 2020-21 property, and it has not reappeared in a single regime since: alts underperformed in chop (−5.3 pp/month, t = −2.06), in the bear (−3.4, t = −1.27), and in this cycle’s bull.

Vintage Analysis: Each Class Reaches −90% Faster

Credit analysts evaluate loan books using vintage curves, which track cumulative defaults by time on book for each origination cohort. We apply the same framework to token markets. Default is defined as the first month-end at or below 10% of qualification price, equivalent to a 90% loss. We then use Kaplan-Meier survival analysis to estimate the cumulative default rate for each class as it ages, accounting for newer tokens that have not yet accumulated a full observation window.

The curves resemble a deteriorating underwriting book. At the same point in their lifecycle, each successive token class has performed worse than the one before it. By month 24, 86% of the 2024 class had lost at least 90% of its entry value, compared with 70% of the 2021 and 18% of the 2020 cohorts. Overall, 73% of qualifying tokens eventually crossed that threshold, with a median time to default of just 13 months. 

The 2026 class is too young to plot (its oldest members have 5 months on book) and the year’s intake is only half formed at 92 qualifiers through June, but the early incidence points the same direction: of the 2026 entrants old enough to be observed at month three, 13.5% had already hit −90%, versus 10.3% for the 2024 class and 8.3% for the 2025 class at the same age. 

Token proliferation likely contributed to the deterioration. Each successive class entered a larger investable universe, while capital increasingly concentrated in BTC and the top 10, leaving more assets competing for attention and liquidity.

Upside Left the Liquid Market

A token’s first month above $50M says nothing about the valuation at which liquid investors could first touch it. Among 2023-26 entrants, 12.4% qualified at valuations above $250M and 2.4% entered above $1B. The share that crossed the $50M threshold within a month of listing also rose from 32% in 2020-21 to 47%. In other words, private investors increasingly capture the move from zero to a large valuation before a given token reaches the liquid market.

The consequence is a convexity collapse. The 2020 class’s median token went on to trade at 5.1x its qualification price, with 51% offering a 5x exit at some point and 33% a 10x. For the 2023-26 entrants, the median peak multiple is 0.93x. In other words, the median token never closed a month above the price at which it qualified; only 3.9% ever offered a 5x, and only 1.4% a 10x.

The downside profile is unchanged (median terminal outcome near −95% in both eras), so the asymmetry that once justified token risk has simply left the market. Larger launches did not offer meaningfully better upside. The 26 tokens that debuted above $1B since 2023 were less likely to fall 90% than $50M-$100M entrants, but their median peak multiple was still just 1.0x.

The data shows that buyers of high-profile launches absorbed substantial downside without comparable upside convexity. Of the 955 tokens that qualified since 2023, only 37 ever reached 5x from qualification, and roughly 13 reached 10x. The 10x group was dominated by memes, including PEPE at 44x, WIF, and BONK, with VIRTUAL and HYPE as notable non-meme exceptions. The broader 5x-10x cohort also included infrastructure and DeFi assets such as Rollbit, Pendle, Celestia, Ondo, and Sei.

The Rising Bar for Relevance

The valuation required to become a meaningful part of the market rose dramatically over the period. In June 2020, the 100th-largest token was worth $47M, below this report’s $50M qualification threshold. By December 2024, the top-100 cutoff had risen to roughly $900M, while entering the top 20 required a valuation above $8B. Later entrants therefore began almost twenty times further from the top of the market.

Very few later entrants closed that gap and became durable market leaders. SHIB spent 49 months in the top 20, supported by retail-scale meme distribution. TON has remained there for 36 months through its Telegram ties, while HYPE entered the top 20 after qualification (November 2024) and has remained there, supported by its position as the leading perps DEX, substantial revenue generation, and tokenholder alignment through buybacks.

Factor Analysis: What Predicted Token Returns

Traditional finance has a large body of research examining whether characteristics such as size, momentum, volatility, and turnover systematically predict equity returns. We applied the same framework to tokens, sorting the universe by each characteristic every month and comparing the subsequent returns of the highest- and lowest-ranked groups. 

Start with size. Across the complete sample, there is no statistically significant difference between large- and small-cap token returns in either 2020-22 or 2023-26. A significant small-cap effect does appear when the sample is restricted to assets captured by historical public top-200 rankings. However, those sources omit tokens that never became large enough to enter the rankings, introducing survivorship bias.

Momentum produced the clearest result. After 2022, tokens with the strongest returns over the previous three months underperformed the weakest by 3.8 percentage points per month (t = −3.52). The highest quintile lost 87% cumulatively, compared with 50% for the lowest quintile. The inversion also holds using one- and six-month lookback periods.

That said, the effect is concentrated in smaller assets. Restricting the sample to tokens above $100M in market cap, the spread narrows to 1.1 percentage points per month and is no longer statistically significant (t = −0.72). Momentum's inversion is therefore best read as a small- and mid-cap phenomenon rather than a market-wide one, though it survives liquidity screens at every turnover threshold we tested.

The joint Fama-MacBeth regression confirms the pattern. From 2023-26, an increase of one standard deviation in momentum predicted 1.2 percentage points of monthly underperformance, while higher volatility predicted 0.6 percentage points. Size and turnover had no significant effect. 

Momentum was largely irrelevant during the 2020-21 bull market, inverted during the 2022 bear market, and remained negative in the years that followed. Since 2023, strong recent performance and high volatility, traits that frequently attract investor attention, consistently predicted worse subsequent returns.

Who Outperformed Bitcoin

Only 81 tokens, or 4.1% of those with at least six months of history, beat BTC over their own measurement window. Each token is compared with BTC on month-end prices, and counts as an outperformer only if the final month-end in its return series carried at least $10,000 of trading volume. That figure is still inflated by newer entrants with short histories. Among the 1,305 tokens observed for at least 24 months, only 22 (1.7%) outperformed.

Exchange tokens account for 32% of those long-window winners despite representing just 2.1% of the long-window set, making them 15 times over-represented. BNB, OKB, GT, LEO, BGB, WBT, and MX all beat BTC (7 of just 27 CEX tokens, a 26% hit rate against roughly 1.2% for the rest of the universe). PancakeSwap's CAKE was the only DEX token to clear the bar and shares their defining feature: fee revenue routed to recurring buybacks and burns.

The remaining winners include established assets such as ETH, SOL, and TRON, alongside a small number of memes and cycle-specific outperformers. These exceptions should not be mistaken for broader sector strength. Memes, for example, produced a median return of −97% across 275 qualifiers and had a slightly below-average BTC beat rate.

Of note, sector labels here come from CoinGecko's category tags. We manually reviewed every token classified as a centralized exchange (CEX) token, but did not review the category assignment of all 2,114 assets, so a full review would likely move some tokens between categories. The report's other statistics are computed on the full universe and do not depend on sector labels at all.

The closest comparison in equities comes from Hendrik Bessembinder’s influential study, “Do Stocks Outperform Treasury Bills?” He found that roughly 4% of listed US companies accounted for the stock market’s net wealth creation since 1926. Crypto displays similarly concentrated winners, but with far worse outcomes for the median asset. The median token lost 97%, rather than merely underperforming its benchmark. Crypto is Bessembinder on hard mode.

The exchange-token result is unlikely to be a coincidence. In a market where nearly every token underperformed, the over-represented group of outperformers carried the closest thing crypto offers to an equity-like claim: recurring fee revenue paired with programmatic buybacks. 

What This Means for Allocators

These findings apply to the liquid portion of the token lifecycle, measured from the first month-end above $50M in circulating market cap. They do not capture venture entry prices, vesting schedules, or distributions realized before or around listing.

• BTC should be the default hurdle. Only 4.1% of tokens beat BTC over their measurement window, falling to 1.7% among assets with at least 24 months of history. Broad token exposure has historically been a poor substitute for simply holding BTC.
• Position sizing should reflect the base rate. The median token lost 97% from qualification, and 73% eventually fell at least 90%. Near-total loss should be treated as a normal outcome, not a tail risk.
• Be cautious with new listings. Each successive token class has deteriorated faster than the one before it. The median token reached a 90% loss within 13 months, while the upside available after listing has also declined sharply.
• Do not chase recent winners. Since 2022, high momentum and high volatility have predicted worse subsequent returns. The high-momentum, high-volatility tokens that tend to attract the most attention consistently underperformed recent losers.
• A larger token basket offers limited diversification. It may reduce idiosyncratic risk, but it provides little protection from systematic crypto market risk. For allocators with the resources to underwrite individual assets, a concentrated portfolio of high-conviction positions may therefore be more sensible than broad token exposure. Meaningful diversification still has to come from outside crypto.
• Cash flows and tokenholder alignment deserve greater weight. Exchange tokens were heavily over-represented among long-term BTC outperformers, with recurring fee revenue and buybacks as their defining shared characteristic. Further research is still needed to determine how consistently buybacks translate into stronger returns across DeFi protocols.

The constructive interpretation is that consolidation increases the value of fundamental research. The opportunity is no longer broad exposure to an expanding token market, but identifying the small number of assets capable of sustaining relevance, generating cash flows, and returning value to tokenholders.

Final Thoughts

I began this report intending to quantify the effect of token buybacks. Once I started working through the return data, however, a broader and more important story emerged. The results put numbers behind the disillusionment liquid token investors have felt over the past several years.

For someone approaching crypto from the outside, the asset class has become fundamentally unattractive to underwrite. The baseline long-term outcome for a newly qualified token is a drawdown of more than 90%, while only a small minority outperform BTC. Under those conditions, it is difficult to justify owning a broad basket of tokens rather than simply buying BTC and moving on.

Recognizing the problem is the first step toward fixing it, and there are signs that the market is beginning to respond. Blockworks’ Token Transparency Framework and MetaDAO’s ownership coins are attempts to improve disclosure and strengthen tokenholder rights. 

These efforts are still early, but they give me reason to remain optimistic about the future of tokens. The purpose of this report was not to argue that the asset class is beyond repair, but to quantify the severity of the problem. I still believe this is one of the best periods to be a fundamental investor in crypto precisely because the investable universe is so narrow. Finding the exceptions will require more discipline than in prior cycles, and the industry will need time to work through the structural problems left by the previous era, but in the end that effort will be worth it.

Appendix: Methodology and Robustness

Universe Construction

• The primary universe includes 2,114 tokens that closed at least one month above $50M in circulating market cap between January 2020 and June 2026. The assets were identified through an audit of 12,014 CoinGecko listings, including inactive and delisted tokens.
• Each token enters the sample at its first month-end above $50M and remains in the universe thereafter. Assets are not removed if they later become inactive or are delisted.
• The market-breadth analysis uses a broader dataset of approximately 10.7k crypto-native assets that closed at least one month above $1M. This was constructed from roughly 54k listings, including approximately 36k inactive or delisted listings.
• Stablecoins, wrapped and bridged assets, liquid-staking and receipt tokens, tokenized real-world assets, commodities, and duplicate listings are excluded throughout.

Return Construction

• Returns are measured from month-end to month-end. Observations more than 10 days stale are excluded.
• Each token is compared with BTC over its own measurement window. A token is counted as outperforming only if the final month-end in its return series carried at least $10,000 of trading volume. The condition applies to gains only: an untraded mark can overstate a token's value but never understate it, so losses are counted regardless of data quality. This screen removes three long-window outperformers whose gains were marked on delisted or dormant listings and lowers the six-month beat rate from 5.2% to 4.1%. Peak-multiple statistics apply the same rule, measuring the highest month-end close that carried real trading volume.
• The equal-weighted index gives each eligible token the same weight and is rebalanced monthly. Monthly returns are winsorized at the 1st and 99th percentiles to limit the impact of residual micro-cap data errors. Median returns and all count-based statistics are unaffected by this treatment.
• Reported t-statistics use Newey-West standard errors with three lags.

Data Quality

• A full audit removed 106 wrapper, receipt, pegged, duplicate, and wash-traded assets, along with 602 corrupted return-months associated with deprecated tickers. A further 91 assets with likely redenomination errors were excluded from BTC outperformance and alt season round-trip calculations. These exclusions strengthened rather than weakened the headline findings.
• Vintage curves are conservative in the same direction: 90 tokens that never reached the −90% threshold carry stale terminal marks, so marking them to realizable prices would raise default incidence rather than lower it.
• Sector classification. Sector labels derive from CoinGecko category tags applied through a fixed precedence rule, with manual review of the centralized exchange bucket. We did not individually review all 2,114 assignments, and a full review would reclassify some. Sector labels affect only the composition analysis (every headline statistic is computed on the full universe) and the exchange result is robust to the judgment calls, ranging from roughly 15x to 30x over-representation across alternative reasonable definitions.

Robustness Checks

• Qualification threshold: Raising the entry threshold does not materially change the results. At $100M, the median return remains −97% and 4.3% beat BTC. At $250M, the median remains −97% and 3.8% beat BTC.
• Entry timing: Tokens often cross $50M after appreciating, which could place qualification near a local peak. Delaying the start of measurement by six months produces a median return of −94% and a 3.5% BTC beat rate. Delaying it by 12 months produces a median return of −91% and a 3.6% beat rate.
• End date: The results are not driven by ending the analysis during a market drawdown. When the sample is cut off in December 2024, the strongest month of the current cycle, the median token had already lost 81% and 82% of the universe was below its qualification price.
• Market-regime thresholds: Alternative bull and bear thresholds of 20% and 40%, or 30% and 60%, produce the same broad conclusion. Alts outperformed BTC by approximately 8–10 percentage points per month during the 2020-21 bull market and underperformed by 8.0–8.3 percentage points during 2023-26 bull months. The latter estimates have t-statistics between −5.2 and −5.8.
• Winsorization: The precise equal-weighted index return varies with the clipping threshold. The full-period return is −30% using the 0.5th and 99.5th percentiles, −53% using the 1st and 99th percentiles, and −71% using the 2nd and 98th percentiles. Each remains far below BTC's 731% return. Fully unwinsorized equal weighting is dominated by residual errors in micro-cap price data. Count-based findings do not change under any of these specifications.
• Momentum: The post-2022 momentum inversion holds using one-, three-, and six-month formation periods, with t-statistics ranging from −3.1 to −3.5, and it survives liquidity screens, with t-statistics between −2.1 and −2.5 at prior-month turnover floors from 0.1% to 1% of market cap. It is, however, concentrated in the smaller half of the universe: restricted to tokens above $100M, the quintile spread is not statistically significant (t = −0.72). The effect is therefore best read as a small- and mid-cap phenomenon rather than a market-wide one.

The information contained in this report and by Blockworks Inc. and related affiliates is for general informational purposes only and is not intended to provide legal, financial, or investment advice. The report should not be construed as an offer or solicitation to buy or sell any security, token, or financial instrument and does not represent any recommendation or endorsement of any investment or financial product or service. Blockworks Inc. and related affiliates are not registered as a securities broker-dealer or an investment advisor in any jurisdiction or country.