The measured window: 30 complete UTC days
We reconstructed the official dashboard's displayed transaction register from 23 August 2026 at 00:00 UTC, inclusive, to 22 September at 00:00 UTC, exclusive. The current partial day is excluded. After removing duplicate transaction links, the window contains 120 records, four on every day, with six hours between successive displayed timestamps.
The sum is 387,952.42 AVNT and $39,962.07 at the dashboard's displayed precision. These are dashboard-reported purchases, not a receipt-by-receipt audit of all 120 transactions. Our two chain spot checks cover the first and last records of the window. That endpoint selection checks interpretation and period boundaries; it is not a random sample or a statistical assurance of the entire register.
Sources: Avantis Foundation buyback dashboard ↗ · PerpsAtlas: downloadable 120-record transaction CSV ↗ · PerpsAtlas: downloadable 30-day aggregate CSV ↗
The daily distribution matters more than one annualized headline
Using the published daily file, mean reported spending was $1,332.07 per day and median spending was $1,188.94. The largest day was 11 September at $3,437.55; the smallest was 6 September at $347.57. The average therefore describes this particular mix of busy and quiet days, not a contractual daily payment.
Total displayed dollars divided by total displayed tokens gives approximately $0.103008 per AVNT. This is a ratio of aggregate purchase values and units, not today's market price and not a simple average of execution prices. Extrapolating the dollar total at an unchanged daily rate gives about $486,205 over 365 days. We show that arithmetic as a sensitivity input only: the calculation does not establish sustainable earnings or future buyback commitments.
Sources: PerpsAtlas: downloadable 30-day aggregate CSV ↗ · PerpsAtlas: downloadable 120-record transaction CSV ↗
First spot check: follow the actual USDC transfers
The first linked transaction succeeded at 00:00:15 UTC on 23 August. BaseScan shows 285.303364 USDC leaving the sender. Its displayed transfer path allocates 284.875409 USDC to a trading pool and 0.427955 USDC to another recipient; these sum exactly to the sender's transfer. The dashboard displays $285.30. We do not infer the second recipient's contractual role from an address alone.
The same transaction sends 2,843.883706163844430764 AVNT to 0x000000000000000000000000000000000000dEaD. The register displays 2,843.88 AVNT. The final transfer exceeds the pool's displayed AVNT transfer by 0.000000000000000214 AVNT, so we retain the final destination amount rather than pretending every intermediate quantity is identical. This is a transfer inspection, not an audit of the router's full accounting.
Last spot check: two pools, one final destination
The last selected transaction succeeded at 18:00:15 UTC on 21 September. It shows 1,051.458155 USDC leaving the sender: 498.693459 and 551.187509 USDC go to two pools, while 1.577187 USDC goes to another recipient. Their sum again reconciles to the sender's USDC outflow. Gas is a separate ETH cost.
The two displayed AVNT outputs are 4,379.333899598156085135 and 4,839.283859159649439894. Together they equal 9,218.617758757805525029 AVNT, the amount sent to the dead address. The dashboard's 9,218.61 AVNT and $1,051.45 show less precision than the receipt; in this sample the final digits are truncated. We preserve displayed cents and hundredths in our dataset, rather than silently manufacturing full-precision monthly totals.
Sources: BaseScan: 21 September, last selected transaction ↗
Three figures that should never be added together
A stablecoin purchase outflow measures funds used in the transaction. The AVNT transfer measures units delivered to a destination. The explorer's dollar annotation values tokens using its own price display. These are alternative descriptions of the event. Adding the purchase dollars to the displayed dollar value of the AVNT would count the same economic activity twice.
Sending tokens to a dead address also differs from paying a cash distribution to each holder. It does not, by itself, prove that the token contract's totalSupply field decreased: that requires checking the contract's burn implementation and supply state. For a supply model, circulating tokens, reported total supply and economically inaccessible balances must therefore remain separately defined. We have not treated the dead-address transfer as proof of a contractual claim on future fees.
Sources: BaseScan: 23 August, first selected transaction ↗ · BaseScan: 21 September, last selected transaction ↗
What this changes in the AVNT valuation research
The foundation dashboard describes 30% of daily trading fees as the live allocation and 50%+ as a later milestone. A ratio is not enough to reconcile actual spending: fee earning dates, eligible fee definitions, execution timing and routing costs can differ. Our measured purchase window should be matched against the same underlying fee window before claiming that the allocation was fully delivered or missed.
The foundation token page reports a nominal total supply of one billion AVNT. Our displayed monthly units equal about 0.0388% of that denominator. This is a scale comparison, not a measured reduction in circulating supply: unlocks and emissions can move the tradable balance in the opposite direction. Nor does one billion specify the next twelve months of circulating supply.
A defensible valuation must state how future purchase spending is normalized, how buybacks are treated economically, which supply horizon is used and why its risk assumptions are appropriate. This dataset improves the observed-flow evidence. It does not resolve all those choices, so this edition does not assign an editorial fair-value range or a buy/sell rating.
Sources: Avantis Foundation buyback dashboard ↗ · Avantis Foundation token overview ↗ · PerpsAtlas: downloadable 30-day aggregate CSV ↗
Download and reproduce the calculation
The transaction CSV contains the displayed UTC timestamp, AVNT units in integer hundredths, dollars in integer cents and the full transaction hash. Divide the two amount columns by 100. To inspect a receipt, append its hash to https://basescan.org/tx/. The daily CSV groups the same rows by UTC date and includes each day's record count. Both downloads are versioned to this research date.
We inspected 18 pages of the public register, captured boundary records, deduplicated full links, and verified the selected window has no missing six-hour display intervals. This proves internal coverage of the captured register; it cannot prove that the dashboard includes every relevant transaction. The display timestamps omit seconds, while both inspected receipts occurred fifteen seconds after their displayed minute. No unspecified time zone or partial current day was silently included.
This edition is a dated manual reconstruction. It is not an hourly automated feed. Future editions should retain these files, publish corrections explicitly, extend receipt reconciliation and add a permitted reproducible collection source. Readers can check the complete captured register today without relying on our narrative alone.
Sources: PerpsAtlas: downloadable 120-record transaction CSV ↗ · PerpsAtlas: downloadable 30-day aggregate CSV ↗ · Avantis Foundation buyback dashboard ↗
What we monitor next
- Reconcile all 120 receipts, rather than extrapolating the two endpoint checks.
- Match buyback spending to a same-period, same-product eligible-fee series.
- Verify burn accounting, inaccessible balances and the forward circulating-supply schedule.
- Keep model assumptions separate from future automatic daily arithmetic.
Frequently asked questions
How much did Avantis report buying back in this window?
The 120 displayed records sum to $39,962.07 and 387,952.42 AVNT for 23 August through 21 September 2026, using complete UTC days. These are dashboard values at displayed precision.
Was every buyback independently audited?
No. The register was reconstructed and its arithmetic checked. Two linked transactions were inspected on BaseScan; a complete receipt audit remains open.
Does a buyback establish that AVNT is undervalued?
No. Purchase spending is one input. Supply, unlocks, persistence of the flow, token-holder rights and valuation assumptions also matter.
Compare the evidence
Open the comparison desk →Source register
- Avantis Foundation buyback dashboard · checked 2026-09-22
- Avantis Foundation token overview · checked 2026-09-22
- BaseScan: 23 August, first selected transaction · checked 2026-09-22
- BaseScan: 21 September, last selected transaction · checked 2026-09-22
- PerpsAtlas: downloadable 120-record transaction CSV · checked 2026-09-22
- PerpsAtlas: downloadable 30-day aggregate CSV · checked 2026-09-22
Research revisions
- 2026-09-22 — Initial 30-day register with downloadable CSVs and two endpoint transaction checks.
AI-assisted research checked against cited sources. Facts, assumptions and interpretation are distinguished; this is not a financial audit or a recommendation tailored to you. Editorial standards.
Compare protocols →