Which DEXes are actually filling trades, read from the blocks themselves rather than from any venue’s own reporting.
Every figure above covers the decoder's last 7d 16h, except pooled TVL, which is the present moment. Volume counts venue trades only — swaps, routes and arbs — over the 94.34K of them carrying a price. Trades the price series does not reach are unpriced, not worthless. Another $5.46M moved through prediction markets, staking, liquidity moves, locks and reward claims, escrows and lending, which are not trades at a venue and are left out of every figure here.
Where trades were actually filled, grouped by platform — open one to see the protocols and pool types behind it. A route touching two venues counts for both, so these sum to more than the total.
| Platform | ||||
|---|---|---|---|---|
 | $5.49M | $6.26K | $302.37K | $1.48M |
 | $1.34M | $2.82K | $97.8K | $580.43K |
 AlgoFi | $112.34K | $598.67 | $3.4K | $25.52K |
 HumbleSwap | $63.79K | $53.36 | $5.9K | $26.63K |
 STAMM | $41.9K | $2.75 | $690.36 | $1.69K |
 | — | $1.89K | $76.48K | $298.88K |
| Biatec CLAMM | — | $0.02 | $238.58 | $398.41K |
| TxnLab | — | $25.65 | $179.01 | $179.01 |
| Liquid staking | — | — | $0.03 | $0.03 |
| loan | — | — | — | — |
Aggregators that picked the venues rather than holding liquidity themselves. Their volume is the same trades seen from the other side, not additional volume, and they hold no pooled liquidity of their own.
| Router | |||
|---|---|---|---|
 VestigeRouter | $429.78 | $18.94K | $121.81K |
| Pact | $145.58 | $14.41K | $72.58K |
 HaystackRouter | $219.72 | $6.9K | $77.51K |
| Tinyman | $207.3 | $5.75K | $20.33K |
 Router | $36.21 | $1.58K | $3.6K |
| Folks Finance | — | $1.55K | $4.74K |
 ASAStatsRouter | — | $0.65 | $9.16 |
By volume traded, not by number of swaps — the busiest address is rarely the one moving the most. Venue trades only, so a staking lock or a direct transfer is not counted.
`window` says which rounds and seconds these figures cover: with no `since` that is the WHOLE store, which is not a day or a week and must not be labelled as one. by_venue traders are DISTINCT per venue across every combination; by_venue volume double counts a multi-venue route, so only volume_usdc above is the deduplicated total. venue counts exclude peer-to-peer exchanges (p2p_* separately). venue counts also exclude stake/unstake/claim and predict_transfer, where nothing was exchanged at a rate; mint and redeem remain, being a real exchange into a different asset, and predict_buy/sell/claim/redeem_set remain, being real prediction trades. by_class splits the SAME rows by what the activity was -- amm, liquidity, staking, prediction, escrow -- and unlike by_venue it sums exactly to volume_algo, because every row belongs to one class and none is dropped. Use it to keep AMM trading apart from the rest; a prediction row carries no protocol, so by_venue never shows it at all. top_addresses is ranked by volume_algo, with n as a secondary figure and unvalued saying how many of that address's rows carry no price. volume_algo values the RECEIVED leg at the asset price effective at that round, or the SENT leg where the received asset has no price -- value_leg on each row says which; unvalued_swaps predate the price series and are counted, never guessed.