Loading the Numbers
Reading the published files. Nothing is shown until they arrive.
TidelineSame pages and numbers in both modes. Easy uses everyday words; Pro has the full detail.
TidelineWhat this page answers: where every number comes from, and how you can tell it was not bent to fit.
Where each number comes from, what was decided before looking, and where public figures disagree.In short: Tideline uses public data anyone can download again, wrote its rules down before looking at the results, and seals each with a stored on NEAR’s . Below is the detail.
Where the Numbers Come From
Volume, fees and revenue for the last 30 days, and hourly token prices. No key.
The public NEAR dataset: a year of intents.near logs for fee changes, and 90 days of transactions and contract use, re-run each week. Stored here, so a re-run needs no account.
FastNEAR
intents.near settlement logs: every transaction that names a known solver or the Confidential Intents treasury, read in full for each day, not sampled. Swaps, frontend, fees and solver come from these. No key.
Intents Explorer and 1Click
The rolling panel only: totals re-pulled once a day, kept 24 hours and stored nowhere. Their terms bar keeping this data, so it sets no reading and none of it is in this repo. Intents Technology is the technology provider of these APIs.
NEAR RPC
Token supply, token decimals and names, and the testnet contract that anchors each frozen day. No key.
Every answer comes from public data that anyone can download again. The copies we keep need no key or account to re-run.
DefiLlama
Swap totals, fees, NEAR’s fee income and hourly prices.
BigQuery
Google’s copy of the public NEAR record: a year of fee changes and recent weekly activity.
FastNEAR
Reads every NEAR Intents transaction of each day, in full.
Confidential Intents
The private system. We only see money going in and out of its doorway, the .
Intents Explorer and 1Click
The outside count: kept for one day and used for no answer.
NEAR RPC
NEAR’s coin supply, and the testnet record that stores each day’s fingerprint.
Volume, fees and revenue for the last 30 days, and hourly token prices. No key.
The public NEAR dataset: a year of intents.near logs for fee changes, and 90 days of transactions and contract use, re-run each week. Stored here, so a re-run needs no account.
FastNEAR
intents.near settlement logs: every transaction that names a known solver or the Confidential Intents treasury, read in full for each day, not sampled. Swaps, frontend, fees and solver come from these. No key.
Intents Explorer and 1Click
The rolling panel only: totals re-pulled once a day, kept 24 hours and stored nowhere. Their terms bar keeping this data, so it sets no reading and none of it is in this repo. Intents Technology is the technology provider of these APIs.
NEAR RPC
Token supply, token decimals and names, and the testnet contract that anchors each frozen day. No key.
Decided Before Looking
The list of fee changes and the level for each reading are written down in the pre-registration before volume is looked at. If the list changes later, the change is shown on the page.
It is not blind: an earlier private estimate by the same author was seen first, and the pre-registration says so on its first line.
Amendment A2 was added on 6 Nov 2026, after the shift to the treasury was seen and with the published readings known, and it says so. It stops a loss of data from moving the verdict: without it, Solver Competition going Not Measured would have lifted the verdict to Bullish.
Amendment A3 sets out the rolling panel and the visibility check, and A4 the year of chain history for fee changes. Each was added on 7 Nov 2026 before the data it governs was read, and each says what was already known.
We wrote down the rules, and the line for each answer, before looking at the results. It is not perfectly blind: an earlier rough estimate had been seen, and the rules say so. A few rules were added later, after money started moving into the private system; each one says what was already known.
The list of fee changes and the level for each reading are written down in the pre-registration before volume is looked at. If the list changes later, the change is shown on the page.
It is not blind: an earlier private estimate by the same author was seen first, and the pre-registration says so on its first line.
Amendment A2 was added on 6 Nov 2026, after the shift to the treasury was seen and with the published readings known, and it says so. It stops a loss of data from moving the verdict: without it, Solver Competition going Not Measured would have lifted the verdict to Bullish.
Amendment A3 sets out the rolling panel and the visibility check, and A4 the year of chain history for fee changes. Each was added on 7 Nov 2026 before the data it governs was read, and each says what was already known.
How Much the Public Chain Still SeesHow Much Is Still in Public View
3.88: DefiLlama volume divided by chain-counted USD over 7 days, on 2026-11-11 ($907M against $234M). Before 27 Oct it ran from 3.65 to 4.38. A higher ratio means the chain sees less, which is lost sight, not lost demand. It carries no line and no state: chain days reach only 23 of the 90 days before 27 Oct, too few to set one (amendment A3).
DefiLlama counts about 3.9 times the swapping we can see on the public record, in the range it had before 27 Oct (3.6 to 4.4 times).
Every number worked out from our saved data re-runs with npm run verify. Source data such as price and supply is used as given.
Where Public Figures Disagree
Figure
Ours
Other Source
Why They Differ
Checked Against the Chain
40 frozen days, 2026-10-04 to 2026-11-12, no gaps. Every day read every listed transaction. On 4 days a few more were read than listed (2026-10-10: 1, 2026-10-27: 1, 2026-10-31: 6, 2026-11-03: 1). The newest 7 days are shown. Swaps through the Confidential Intents treasury count as paired confidential flows, which are not a full count of private swaps.
Day
Transactions Listed
Read
Swaps
Unpriced Swaps
What Differs
Which Swaps the Costs Cover
Over 30 days (2026-10-14 to 2026-11-12), the frozen days hold 539,720 swaps on 4,907 routes. 61% of the swaps, 50% of the USD, have a price on both sides; only these are costed on Follow the Fee. 37% have a price on one side, so they count in volume and size but not in cost. 2.6% have no price at all and are counted, never guessed. Swaps through the Confidential Intents treasury are paired confidential flows, not a full count of private swaps, and Follow the Fee reports them apart.
Largest Routes
Volume
Share
Priced
Public Activity on intents.near
Each week of transactions, signing accounts, receipts and NEAR burnt as gas, from the BigQuery public dataset. Treasury traffic is in these counts, so more activity is not more swaps, and signers are mostly relayers and solvers, not users.
Weekly counts of activity on NEAR Intents. They include the private system’s traffic, so more activity does not mean more swaps.
Tideline is research, not financial advice.Follow Us on Telegram (opens in a new tab)
We compared 4 of our figures with other public sources. Where they differ, the detail says why.
Figure
Ours
Other Source
Why They Differ
Each frozen day lists every intents.near transaction that names a known solver or the Confidential Intents treasury, then reads each one. A day freezes only when at least 99% of the list was read; otherwise it stays a gap. Its file hash is anchored on NEAR testnet (Reproduce). A re-run from scratch reads the same transactions and swaps, but not always the same hash: DefiLlama can later give slightly different hourly prices for small tokens, which moved USD by at most 0.03% when four days were frozen twice.
We saved 40 days of the public record, with no day missing. Every transaction on each day’s list was read. Each saved day’s is stored on NEAR’s , so nobody can change it quietly.
40 frozen days, 2026-10-04 to 2026-11-12, no gaps. Every day read every listed transaction. On 4 days a few more were read than listed (2026-10-10: 1, 2026-10-27: 1, 2026-10-31: 6, 2026-11-03: 1). The newest 7 days are shown. Swaps through the Confidential Intents treasury count as paired confidential flows, which are not a full count of private swaps.
Day
Transactions Listed
Read
Swaps
Unpriced Swaps
What Differs
Each frozen day lists every intents.near transaction that names a known solver or the Confidential Intents treasury, then reads each one. A day freezes only when at least 99% of the list was read; otherwise it stays a gap. Its file hash is anchored on NEAR testnet (Reproduce). A re-run from scratch reads the same transactions and swaps, but not always the same hash: DefiLlama can later give slightly different hourly prices for small tokens, which moved USD by at most 0.03% when four days were frozen twice.
The biggest routes priced on one side only: USDT (Tron) to USDT (Ethereum) (9.1%), USDT (Ethereum) to USDT (Tron) (7.0%), USDT (Tron) to NBTC (NEAR) (4.5%), USDC (Ethereum) to USDT (Tron) (4.2%), NBTC (NEAR) to USDT (Tron) (3.9%). The fee figures on Follow the Fee hold for routes priced on both sides; other routes can cost more or less. Volume is the priced side, so routes with no price at all show none.
Over the last 30 saved days, 61% of swaps had a dollar price for both coins, so only those are used for the costs on Follow the Fee. The others still count in the totals, and no price is guessed.
Over 30 days (2026-10-14 to 2026-11-12), the frozen days hold 539,720 swaps on 4,907 routes. 61% of the swaps, 50% of the USD, have a price on both sides; only these are costed on Follow the Fee. 37% have a price on one side, so they count in volume and size but not in cost. 2.6% have no price at all and are counted, never guessed. Swaps through the Confidential Intents treasury are paired confidential flows, not a full count of private swaps, and Follow the Fee reports them apart.
Largest Routes
Volume
Share
Priced
The biggest routes priced on one side only: USDT (Tron) to USDT (Ethereum) (9.1%), USDT (Ethereum) to USDT (Tron) (7.0%), USDT (Tron) to NBTC (NEAR) (4.5%), USDC (Ethereum) to USDT (Tron) (4.2%), NBTC (NEAR) to USDT (Tron) (3.9%). The fee figures on Follow the Fee hold for routes priced on both sides; other routes can cost more or less. Volume is the priced side, so routes with no price at all show none.
Week To
Transactions
Signers a Day
Receipts
NEAR Burnt