One year after the October 11 crash, has the market recovered? This article breaks down the liquidity repair path from spreads, order book depth, capital flows, and participation behavior, revealing the truth of a recovery with clear divergence among BTC, ETH, and UNI.
Author: danny
From October 10 to 11, 2025, more than $19 billion in leveraged positions were liquidated, affecting approximately 1.6 million trading accounts. The average intraday decline of the sample tokens tracked by CoinDesk was about 47%, and perpetual OI on major platforms fell from $217 billion to $123 billion. The full industry-wide number of affected exchanges, quantitative institutions, and halved tokens is countless.
One year later today, has the market recovered to its pre-shock operating state?
This article will discuss this age-old question from factors such as trading costs, order book capacity, participants, and capital behavior: Has the market come back?
I. What Counts as Recovery: The Basis Given by the Literature and the Criteria Used in This Article
In the IMF working paper "Measuring Liquidity in Financial Markets" (2002), Sarr and Lybek divide market liquidity into trading costs, immediacy, depth, breadth, and resilience, and the paper emphasizes that no single indicator can cover all dimensions. In its 2024 empirical research on the bond market, the BIS further tests resilience by the speed at which spreads return to normal levels and uses 90-day windows before and after the shock for comparison.
The broad notion of "market confidence" also needs to distinguish optimism about gains from trust in the market. "Trusting the Stock Market" (2008) by Guiso, Sapienza, and Zingales, published in the Journal of Finance, links trust to whether people participate and to the scale of allocations after participation. Baker and Wurgler (2006) extract a common sentiment factor through multiple proxy variables and remove some macroeconomic influences. Bekaert and Hoerova (2014) distinguish expected volatility from the variance risk premium, avoiding treating market calm as a period in which risk aversion has faded.
We abstract and summarize the above views into numerical standards that can be tested and identified:
The main pre-disaster baseline is set as September 10 to October 9, 2025; the anniversary window is September 10 to October 9, 2026. In addition, a 90-day baseline, tolerance ranges of 5%/10%/20%, and sample coverage rates of 70%/80%/90% are used to test whether the results are robust.
II. Liquidity: Spreads Recover First, Liquidity Much Later
The research paper Two-Regime Liquidity Recovery After a Perpetual Futures Liquidation Cascade: Evidence from Hyperliquid and the October 10, 2025 Event provides observational data and analytical conclusions beyond the order book. Lim (2026) uses public Hyperliquid trading data to build a 92-day minute-level panel for five perpetual contracts: BTC, ETH, SOL, XRP, and DOGE. In the 23-day window after the event and before another round of selling pressure began on November 3, the quoted spreads of all five contracts returned to within 1.2 times the pre-disaster level, with three returning to within 1.1 times; however, the best bid-offer depth of four of the five continued to shrink from days 1-7 after the event to days 17-23, and in the final segment was only 66%-79% of the pre-disaster level.
Although this independent study cannot directly show whether liquidity has returned to pre-disaster levels, it clearly establishes a judgment logic: quotes appearing to return to normal does not mean that order-placing capital has come back.
If we put the order books of the same contract and the same period together, the differences in recovery become clearer. In addition to the above data, this article also compares Binance's BTCUSDT, ETHUSDT, and UNIUSDT: the pre-event period is taken as 00:00-06:00 UTC on October 10, 2025, and the one-year-later period is taken as the same time period on October 10, 2026. For each snapshot, the notional depth of the same magnitude on both the bid and ask sides is first summed, and then the time median is taken.
UNI exceeds the pre-event depth level at all three tiers, with 5% depth at about twice the level; BTC is below the baseline at all three tiers. ETH lies between the two, with 2% depth slightly above the pre-event level, while 1% depth is about 13% lower. Extending the 2026 observation window to the full day of October 10, the 1% depth recovery ratios for the three are 81.4%, 86.8%, and 120.0%, respectively, and the direction does not change. However, ETH's 5% ratio falls from 92.8% for the same hour to 89.5% for the full day.
ETH's total depth is close to the baseline, but the bid and ask sides have not recovered in sync. The median 1% bid depth fell from 124.2 million to 90.6 million USDT, only 72.9% of the pre-event level; the ask side rose from 95.3 million to 101.2 million, reaching 106.2%. The corresponding ratios for BTC are 94.4% and 76.8%, and for UNI are 109.6% and 138.4%. According to this article's rule that "the ask side cannot make up for the bid side," ETH cannot be considered recovered to the pre-disaster level merely because the two-sided total is close to the pre-event level.
Interestingly, if calculated by coin quantity and by amount, different pictures emerge. The median BTC two-sided 1% order quantity rose from about 4,246 coins to 4,737 coins, an increase of 11.6%; ETH rose from about 50,792 coins to 76,946 coins, an increase of 51.5%; UNI increased by 32.7%. Because prices fell, although the number of BTC and ETH orders increased, USDT notional depth declined; if compared with prices one year earlier, depth has in fact improved markedly.
UNI's data also provide another cross-sectional analysis of the shock. From 20:50 to 22:00 UTC on October 10, 2025, its median two-sided 1% depth was only 158,300 USDT, 93.9% less than the morning baseline; the lowest valid sample was 18,100 USDT, 99.3% less. By October 11, the full-day median had returned to 69.8% of the baseline, on October 12 it was 85.9%, and one year later on October 10 it was 120.0%.
In summary, order book depth seems to support that liquidity is recovering, but the recovery is uneven: UNI exceeds the pre-event short window, ETH has a gap in near-side bids, and BTC's USDT depth lags the recovery in order quantity.
III. Confidence: The Turn Began in Q3, but Not All Players Have Returned
The data from July to September show a sequence. In its early August July review, Coinbase recorded that perpetual, delivery, and options OI rebounded, while spot and perpetual trading volume declined; the relative altcoin OI indicator was about 0.6-0.7, with repair concentrated in major coins. In August, positions and trading volume began to increase together; BTC ETF net flows over the past seven days were negative on only two days, and the corresponding ETH indicator was negative on only one day.
In September, BTC and ETH spot ETFs saw net inflows of about $2.6 billion and $900 million, respectively; the monthly average of the relative altcoin OI indicator rose to 0.86, exceeding 1.0 at month-end, up from an August average of 0.71. Coinbase observed that the options skew shifted markedly in August and maintained what its report called a positive state in September, but it retreated somewhat at month-end. These data jointly support a trend of improving risk appetite.
Therefore, a relatively reliable statement is: in the third quarter of 2026, capital flows and risk appetite showed a mutually corroborating turn; but there is not enough evidence for the conclusion that overall retail and institutional participation has recovered to pre-disaster levels.
IV. Extent of Recovery: Setting Pre-Disaster Scale as 100
The table below provides a reference table of recovery indices based on the above data, so everyone can get a feel for it:
Stablecoins come from two issues of CoinDesk's stablecoin reports; overall trading volume and OI come from its Exchange Review series; CME comes from official quarterly disclosures; both Coinbase trading volume figures are taken from the 2026Q2 earnings presentation.
The reason this article does not combine these data into a single "recovery index," but instead compares stablecoins, trading volume, and positions separately, is to avoid one growing indicator offsetting another shrinking indicator.
V. Equity Perpetuals: Strong Growth in a New Market
In a study commissioned by Bullish, CoinDesk disclosed that equity perpetual trading volume grew from about $16 billion in 2025 to more than $590 billion in 2026. Another October study put the sample OI of equity perpetuals at about $2.7 billion.
At the same time, Coinbase's earnings comparable table shows that non-crypto derivatives trading volume increased from $2 billion in 2025Q3 to $33.6 billion in 2026Q2, about 16.8 times.
VI. Conclusion
Evidence across different time windows presents a sequence of repair: in the weeks after the event, some spreads recovered first while depth lagged; by 2026Q3, capital flows and risk appetite improved together. By early October, the statistical sample shows that BTC and ETH near-price depth rebounded from the crash day, but compared with the pre-event short window, notional amounts have not fully recovered; UNI, however, exceeds the baseline at multiple tiers.In summary, in the year after October 11, signs of confidence recovery emerged; order book depth recovery varied sharply; and stock perpetuals became a new business line.