Reading large transactions

Inputs and outputs (UTXO recap)

Each input spends a previous output in full. Each output creates a new UTXO locked to an address (technically, a script). Sum(inputs) − sum(outputs) = fee to miners. There is no partial spend of an output — if you spend a 10 BTC UTXO to pay 3 BTC, you get ~7 BTC back as change (minus fee) in a new output you control.

Explorers often show "total moved" as the sum of input values. That can dwarf the economic transfer when someone consolidates many UTXOs. Always ask: how much left the sender's control net of change?

Consolidation

Many inputs → one (or few) outputs. Custodians sweep dust into a single UTXO to reduce future fee surface or prepare for a later payout. Large consolidations can touch tens of thousands of BTC without any external counterparty — see the Bitfinex example on Examples (147 inputs, 1 output; label on input, not destination).

Brief labels this pattern with relatively higher structural certainty when input count ≫ output count. That does not tell you why they consolidated.

Distribution / batch send

One (or few) inputs → many outputs. Exchanges batch withdrawals; services pay many recipients in one tx. Output count can reach hundreds while total value stays small — the Mt. Gox dust example on Examples (402 outputs, 0.64 BTC) shows why output count alone is not "whale size."

Peel chains (introduction)

A peel chain is a sequence of transactions where each step spends one UTXO and creates one continuing output plus a smaller peeled output. Visually: value trickles along a path. Hot wallets and privacy workflows use this shape; so do automated payout systems. Detecting peels requires graph context beyond a single tx — Brief does not claim peel-chain detection across hops; it classifies the local tx shape only.

Change and ambiguity

Change outputs return value to the spender. Heuristics (e.g., "smaller output is change") fail often — wallets randomize, use multiple change outputs, or send exact amounts. Treat change identification as uncertain unless you have wallet-specific knowledge.

Fees and priority

Fee rate (sat/vB) affects mempool priority, not moral weight. A 500 BTC tx with 1 sat/vB during congestion may sit pending while a 0.01 BTC tx at 100 sat/vB confirms first. Large size does not buy priority — fee rate does.

Explorers show vsize (virtual size) after SegWit. A transaction that looks "huge" in bytes may pay a lower fee rate if witness data discounts apply. When you compare two large pending transactions on Radar, the one with higher sat/vB usually confirms first regardless of BTC moved.

RBF (replace-by-fee) lets a sender rebroadcast the same logical spend with a higher fee before confirmation. You may see a large tx vanish from the mempool and reappear with a new txid. That is not a reversal — it is the same UTXOs being spent with updated fee metadata. CPFP (child-pays-for-parent) is the opposite pattern: a small child transaction pays enough fee to drag an underpaid parent into a block. Neither pattern tells you who moved coins or why; they only explain mempool mechanics.

When "total moved" misleads headlines

News often reports "X BTC moved" using the sum of all inputs. That number can include:

• Internal shuffles — custodian moves from hot to cold wallet; economic counterparty is the same entity.
• Change returning to self — most value never left the spender's control.
• Batch payouts — one input funds hundreds of small withdrawals; headline BTC is gross, not per recipient.
• Consolidation — many UTXOs merged; no external payment occurred.

Before inferring market impact, separate structural pattern (how many inputs/outputs, who is labeled) from narrative (exchange dump, whale sale). Our tools report structure; they do not predict price.

Coinjoin and deliberate ambiguity

Privacy tools can produce transactions where many inputs and many outputs share similar amounts. To a naive reader, that can look like a random large shuffle. Heuristics that assume "one clear sender, one clear receiver" fail here. Brief may classify shape as high uncertainty when input and output counts are both large and values cluster — that is honest, not a bug.

Do not treat ambiguous shape as hiding malice. Legitimate users, exchanges, and software all create confusing graphs. The correct response is lower certainty and more cross-checking, not a confident story.

Worked read (checklist)

Use this order every time before you trust a summary:

1. Confirmed or mempool? Pending txs can be dropped, RBF-replaced, or never confirm.
2. Input and output counts — consolidation vs distribution vs balanced many-to-many.
3. Labeled touches — which side of the tx hit a known address? One labeled output among 200 is not "the whole tx is Binance."
4. Net outflow guess — subtract obvious change when heuristics allow; mark as uncertain otherwise.
5. Fee rate — urgency signal only, not sentiment.
6. Cross-explorer check — mempool.space, block explorer, or your own node if you run one.

If step 2–4 disagree with a headline, trust the graph. See also mempool vs confirmed and exchange flows for narrative traps.

Using labels without over-reading

Our label directory matches exact addresses only. A label on one output among hundreds does not make the whole tx "an exchange withdrawal." Read which side of the tx touched the labeled address — input, output, or both — and check confidence and source.

Confirmations and depth

After inclusion in a block, a transaction gains confirmations as later blocks stack on top. Exchanges and custodians often wait for several confirmations before crediting deposits. A large mempool sighting is not the same as a settled transfer — see mempool vs confirmed.

Deep reorgs (chain tips changing) are rare on Bitcoin mainnet but are why high-value workflows wait for depth. For education, note the confirmation count when you compare a Radar blip to a block explorer hours later. If the txid disappeared entirely, it may have been replaced (RBF) or never confirmed.

Multisig and complex scripts can make explorer "from" and "to" rows look noisy. Large institutional moves sometimes spend multisig UTXOs; the underlying pattern is still inputs → outputs, but address strings may be less familiar. Brief focuses on value flow and counts, not wallet product names.

Practice loop

1. Pick a txid from Radar or Examples.
2. Count inputs and outputs before reading the summary.
3. Run Brief and compare its steps to your count.
4. Cross-check on mempool.space; note confirmed vs pending.

Educational only — not trading signals.