October 2026 was the worst month for crypto hacks on record. The DefiLlama hacks database, as summarised by Quantstamp, counted $6.50 billion lost across 35 incidents, nearly nine times September's $742.0 million. Security firm PeckShield counted 56 major incidents and $6.48 billion, up about 745% on September.
Those headline numbers are accurate, but on their own they are easy to misread. One incident, the 29 October breach at Binance, accounts for almost all of the month's total. About $3 billion of the stolen assets was frozen soon afterwards. Once you separate the single largest event and subtract frozen funds, October looks like a very different month. This article explains how the main trackers count losses, why their totals differ, and which version of the number answers which question.
Methodology Note
This article compares monthly loss figures from three public trackers: the DefiLlama hacks database, PeckShield's monthly summary, and the CertiK security dashboard. Each one keeps its own list of incidents and its own rules for what counts as a hack, an exploit or a phishing loss. None of them publishes a full rulebook on the pages cited here, so we compare their published totals rather than trying to rebuild them.
Two terms matter throughout. A gross loss is the value of assets taken at the time of the incident. An adjusted or net loss subtracts assets that were later frozen by token issuers or returned by the attacker. Dollar values are measured at the time of each incident, so the same theft can be valued differently by different trackers depending on the prices they use. All figures are in US dollars. Data is as published between 1 and 16 November 2026, and trackers continue to revise their numbers as investigations progress.
Three Trackers, Three October Totals
Here is how the three trackers reported October 2026:
| Tracker | Incidents | October total | Adjustment shown |
|---|---|---|---|
| DefiLlama, as of 1 November | 35 | $6.50 billion | None in the monthly total |
| PeckShield, 1 November | 56 | $6.48 billion | Notes $3 billion frozen for Binance |
| CertiK dashboard, 16 November | 96 | $6.61 billion | $3.72 billion adjusted |
The gross totals sit within about 2% of each other, which is reassuring. The incident counts do not. CertiK lists 96 incidents, nearly three times DefiLlama's 35. The gap comes mainly from scope. CertiK's figures cover exploits and phishing, so they include many small thefts from individual users. DefiLlama's register focuses on protocols, exchanges and other platforms. Neither approach is wrong, but an incident count from one tracker should never be compared with another tracker's count.
The CertiK figures also show how quickly these numbers move. On 31 October CertiK reported about $6.5 billion for October, with adjusted losses of $3.47 billion after $3.03 billion was frozen or returned, as reported by The Defiant. By 16 November the same dashboard showed $6.61 billion in total losses and $3.72 billion adjusted. The difference between those two later figures implies about $2.89 billion frozen or returned, slightly less than the first estimate. A monthly loss figure is a working estimate for several weeks after the month ends, not a final count.
One Incident, Almost the Whole Month
According to Quantstamp's summary, attackers drained an estimated $6.3 billion from Binance's hot wallets on 29 October in about 20 minutes. Hot wallets are the online wallets an exchange uses to process everyday withdrawals. The FBI attributed the theft to North Korea's TraderTraitor group on 31 October. Binance said the attacker got in through a compromised laptop belonging to one of its engineers.
DefiLlama values the Binance incident at $6.36 billion. Set against DefiLlama's October total of $6.50 billion, that is about 98% of the month's losses from a single event. According to CryptoTicker's reading of the DefiLlama data, the other 34 October incidents added up to $136 million. That is less than a fifth of September's total.
The category breakdown points the same way. In the Quantstamp table, which uses DefiLlama data, frontend and infrastructure compromises such as leaked keys and supply-chain attacks account for about $6.46 billion across seven incidents. Smart-contract categories such as token accounting, governance, bridge logic and oracle manipulation add up to less than $17 million combined. In October, almost all of the money was lost through operational security rather than code flaws.
A Worked Example: From Gross to Net at Binance
The Binance incident shows how far gross and net figures can drift apart for one event.
Start with the gross figure of about $6.3 billion. Binance said about $3 billion was frozen, according to Quantstamp, including more than $2.4 billion in USDT and USDC frozen by Tether and Circle. Fiat-backed stablecoins can be frozen by their issuers, while bitcoin and ether cannot. So the share of a theft that can be frozen depends heavily on which assets were taken. Subtracting the frozen amount leaves roughly $3.3 billion that is effectively out of reach.
There is a second question: who bears the loss? Binance said it set aside more than $6.34 billion to make customers whole, using its corporate treasury, futures insurance and emergency reserves, as reported by the BBC. For an exchange customer, the loss shows up as a market and counterparty event rather than a direct loss of funds. For the exchange, it is a balance-sheet loss.
The BBC report also gives a measure of the market reaction. When Binance reopened on 1 November, it processed $4.649 billion in withdrawals and $1.01 billion in deposits between 06:00 and 15:30 GMT, a net outflow of about $3.64 billion. Binance said that was just over 2% of the $171 billion in assets it manages. That figure measures confidence in the exchange, not the size of the theft, but it shows that counterparty risk carries a cost even when customers are reimbursed.
The Rest of the Year Looks Different Too
The same concentration appears in the year-to-date totals. PeckShield said the ten largest hacks of 2026 to 31 October accounted for 92.3% of the year's losses, $8.6 billion in total, before frozen and returned funds were taken into account. Its top ten list includes Bitget at about $387 million and Liquid Network at about $320 million, of which $285 million was returned.
As of 15 November, DefiLlama puts 2026 losses so far at $8.786 billion across 347 incidents. That means one incident accounts for about 72% of a year that has already seen hundreds of separate attacks. CertiK described 2026, at about $9.2 billion, as the highest year for losses on record. That is correct on a gross basis, but the record depends heavily on one event.
November so far looks closer to a typical month. As of 16 November, the CertiK dashboard shows $190.5 million in total losses across 32 incidents, with $13.0 million frozen or returned and $177.5 million adjusted.
What the Data Does and Does Not Tell Us
The data does support a few clear conclusions. October's gross losses were the largest on record for a single month across all three trackers. The losses came overwhelmingly through compromised infrastructure and keys rather than smart-contract bugs. Issuer freezes of stablecoins have become a meaningful factor in how large a theft ends up being.
There are limits to what it tells us. A monthly total that is almost entirely one incident says little about the underlying rate of attacks. Excluding Binance, October's losses were lower than September's, but one quieter month is not a trend either way. Frozen funds are not the same as recovered funds: frozen assets may take a long time to return to their owners, and some may never be returned. Figures also change after publication. CertiK's own October estimate moved by more than $100 million in gross terms and about $250 million in adjusted terms within two weeks.
Comparisons across trackers also need care. Incident counts depend on scope, valuations depend on timing and pricing, and adjusted figures depend on what each tracker classes as frozen or returned. A single headline number cannot capture all of this.
Why This Matters for Allocators and Analysts
For anyone assessing operational and counterparty risk, the way a loss figure is built matters as much as its size. Four habits help.
First, always check whether a figure is gross or adjusted. The difference for October was close to $3 billion.
Second, look at concentration. When one incident makes up most of a month or a year, report the total both with and without it. Otherwise one event can hide, or exaggerate, the trend across the rest of the market.
Third, separate how the attack happened from the headline value. October's losses point to staff devices, signing systems and key management as the main weak points. That is useful for custody and exchange due diligence, and smart-contract audits on their own do not address it.
Fourth, treat recent monthly figures as provisional. The numbers for any month can move for weeks as investigations, freezes and returns are confirmed.
Track Risk Data With Presolt
Presolt brings market, on-chain and risk data together with plain-language research, so analysts and allocators can see what sits behind the headline numbers. Explore the data at Presolt.com.
This article is for informational and educational purposes only and does not constitute financial, investment, legal or tax advice. Figures are drawn from the third-party sources linked above, are subject to revision and may not be complete. Presolt does not guarantee the accuracy of third-party data. Always do your own research and consult a qualified professional before making investment decisions.