Bitcoin mempool, explained

What is the mempool?

"Mempool" is short for memory pool. When you broadcast a Bitcoin transaction, it gets validated by your node, then propagated to peers, who validate and propagate it further. Each node stores the valid pending transactions it sees in its own local mempool — a chunk of memory the node uses as a queue.

A miner trying to assemble the next block looks at its mempool and picks transactions to include — almost always the ones paying the highest fee per virtual byte (sat/vB), because miners get paid in transaction fees plus the block subsidy. When a block is mined and propagated, every node removes those transactions from its mempool. The rest keep waiting.

That cycle — broadcast → wait → get included → drop out — repeats roughly every 10 minutes. When transaction supply exceeds the ~4 MB of effective block space, transactions accumulate. The mempool grows. Fees rise.

How fees are decided

There's no central authority setting Bitcoin fees. The fee market is a continuous live auction. The price is denominated in satoshis per virtual byte — sat/vB — where a "virtual byte" accounts for the SegWit discount on witness data. A normal transaction is roughly 140–200 vB; a complex multi-sig or batched transaction can be 500+ vB.

Wallets typically expose three or four pre-selected fee tiers. They map roughly to expected confirmation times:

The actual sat/vB values that populate those tiers vary minute to minute. mempool.space and other explorers calculate them by looking at what's currently in the queue and projecting which transactions miners will include next.

Reading mempool.space

The most-used live mempool explorer is mempool.space. The dashboard packs a lot into one screen:

• Pending blocks (top row) — a stack of upcoming "projected blocks" miners will likely produce, each showing average fee rate and transaction count. This is the queue.
• Recent blocks (bottom row) — the most recent confirmed blocks, each with the median fee rate, total fees and a miner attribution.
• Mempool size — the total weight of pending transactions in virtual MB (vMB). Anything above 100 vMB means real congestion.
• Recommended fees — the same tier breakdown shown above, but with live sat/vB numbers calibrated to current queue depth.
• Hashrate & difficulty — network security metrics, not directly mempool data but useful context.

Conceptually, mempool.space's projected-blocks view looks something like this:

Each block is colour-coded by fee rate. Green = the next block, paying top tier. Orange and lighter = further-out blocks paying less. When the mempool empties, the queue shortens to one or two projected blocks. When it floods, you can see 30+ projected blocks stacked up.

Why congestion spikes

Bitcoin's block space is fixed: roughly 1 MB of base data plus 3 MB of witness discount per ~10 minutes. Demand for that space is wildly variable. Four kinds of events reliably overwhelm it:

• Halvings. Around each halving (every ~4 years), trading volume and miner shuffling spike. Fees often double or triple for days.
• Bull market mania. When BTC breaks ATHs, exchange flows surge, retail panics in or out, and the mempool fills.
• Ordinals / Inscriptions / Runes activity. The 2023 inscription wave and the 2024 Runes launch pushed fees over 1,000 sat/vB at peak, with hour-target fees alone above 300 sat/vB. The protocol itself didn't change — but a new use case suddenly competed for block space.
• Liquidation cascades. Sharp price drops trigger exchanges to rebalance hot/cold wallets, custodians to move collateral and panicked users to withdraw — all at once.

RBF and CPFP — unsticking a stuck transaction

If you broadcast at 5 sat/vB and the queue then jumps to 50, your transaction can sit unconfirmed for hours or days. After roughly two weeks, most nodes will drop it. You have two ways to escape:

• Replace-By-Fee (RBF). Most modern wallets flag transactions as RBF-enabled. You can broadcast a replacement transaction spending the same inputs at a higher fee. Miners will pick the higher-fee version; the original gets purged.
• Child-Pays-For-Parent (CPFP). When you can't RBF (e.g. the stuck transaction came from an exchange), you spend one of its outputs in a new transaction with a fee high enough to cover both. The miner has to confirm the parent to confirm the child, so both clear together.

What mempool data tells you about the market

The mempool is one of the few real-time on-chain signals that isn't easily manipulated. A few patterns worth knowing:

• Sudden mempool spike + rising fees. Often precedes significant exchange flows. Large holders moving coins before a sell or accumulators consolidating after a buy.
• Sustained empty mempool. Network is quiet — usually correlates with consolidation phases, bear-market apathy or after-mania exhaustion.
• Recent blocks consistently full at high fees. Real organic demand for block space. Bullish for miner revenue, bullish for network security argument, often bullish for sentiment.
• Recent blocks consistently small / partially full. Either low fee demand or, occasionally, a miner intentionally producing empty/small blocks to capture the subsidy without including transactions. Rare but watch for it.

None of these are standalone trade signals. They're context that helps explain why price is moving when it moves — and occasionally hint that movement is coming.

How btclyzer uses mempool data

btclyzer pulls live mempool data from mempool.space — mempool size, recommended fees, hashrate, difficulty, and recent blocks — and surfaces it in the dashboard's on-chain panel. It's also fed into the adaptive multi-factor algorithm as one of the on-chain context inputs on the 1D, 1W and 1M timeframes (network activity is too noisy on 1H/4H to weight heavily).

A mempool reading by itself rarely changes a rating. But a fee spike combined with an EMA 200 break, an RSI divergence and a CBBI distribution reading is exactly the kind of multi-signal alignment that flips the rating to SELL — and lets the algorithm anchor live DCA distribution zones to the resistance levels where outflows would naturally cluster.

FAQ