Bitcoin Lightning users can face a wait of several days to recover funds after closing a payment channel without their peer's agreement. A new study found that such closures remained relatively cheap in 2026, even though operators could not immediately access their Bitcoin.
In a November 12 research preprint, Bitcoin Data Labs researcher Saurabh Kumar examined 973,581 Lightning channel closures recorded between January 2018 and October 31, 2026. Of those, 523,815 were forced closures, in which one party settled the channel without a joint closing transaction.
The study examined 14,435 public forced closures from 2026 through October 4. Their median total transaction fee across both parties reached 925 satoshis, compared with 197 satoshis for channels closed by agreement. That made the median forced closure 4.7 times more expensive.
The fee calculation covered the closing transaction and the additional transactions needed to recover funds. It also included fees for resolving pending payments and certain attempts to speed up confirmation. Kumar divided shared transaction fees across the affected inputs to avoid charging the same expense to multiple channels.
Fees have changed considerably across the years. The median forced-close cost reached 14,318 satoshis in 2020 and 7,346 in 2023. Kumar wrote, "The fee market sets the level, not the ratio."
Lightning allows two parties to make payments through a shared channel without recording every transfer on Bitcoin's main blockchain. When they cannot agree to close it either party can publish an already signed transaction to settle the balance.
However, the channel's rules require the party initiating the closure to wait before spending its share. The safeguard gives the other party time to challenge an attempt to close using an outdated channel balance. The Lightning specification calls this waiting rule to_self_delay.
The research found a median waiting period of 2.5 days for the 2026 public-channel sample. Across the full study period, the median was 1.7 days. These figures measure when funds become eligible to move, not when a later withdrawal transaction confirms.
Nearly half of the observed delay settings were 144 blocks, or about one day. One in ten were set to roughly two weeks.
Kumar also calculated the annual return at which the cost of waiting would match the fees paid to close a channel. That break-even figure stood at 1.2% in 2026, compared with 16.7% across the full period.
Lightning adopted a zero-fee commitment format in May 2026. Despite its name, the format does not remove mining fees from a channel closure. Instead, another transaction pays to get the closing transaction confirmed, while separate transactions can handle pending payments.
The waiting rule for the closing party's balance remains in place. Kumar estimated that the new format would reduce block space used by the previous anchor mechanism by 12%. His research tracked 31 zero-fee commitment closures by the October cutoff, including 13 public channels.
The study does not cover private cooperative closures, and it excludes certain newer channel formats from its detailed fee analysis. Blockchain records also cannot capture fee increases that never confirmed or transactions that other versions replaced. The paper's cost comparison therefore applies to the classified channel closures, rather than every Lightning exit.
ALSO READ: SpaceX & Tesla Merger Could Unite USD 2.1 Billion in Bitcoin