Blue-chip cross-chain quality by order size
The same measurement the
tokenized-stock grids apply to bStocks, applied here to the
crypto majors — and the question is
which chain executes closest to Binance. Pick an asset, and the grid compares
BTC on BSC vs Ethereum vs Base vs Solana (same for ETH); click a chain to break it down by the venues inside it.
BNB and SOL have no liquid cross-chain wrapper, so neither has a cross-chain question of its own — but each is the
base asset of the chain it trades on, so the
Native: BNB vs SOL chip puts them side by side to ask which chain keeps its own base trading closest to Binance. That view adds a
gap row (BSC − Solana) and ranks BSC's BNB venues against Solana's SOL venues in one list below.
Every trade is matched to the Binance price at the
exact second it executed — with Solana's own timestamp corrected by one second first, because it isn't a clock reading (see below). Greener = tighter.
Trades = the full count in that cell; the
spread is measured on the subset that matched a Binance price at its own second (hover any cell for both).
Why USDC-quoted chains aren't credited with free "price improvement"
Binance prices the majors against USDT, but most pools outside BSC price them against USDC — and USDC is not worth exactly one USDT. On 2026-08-03 it traded at 1.00110, i.e. 11 bps rich. Scored naively, every Base trade looked like it beat Binance by ~11 bps, which would have made the whole cross-chain ranking wrong: it was stablecoin basis, not execution quality. So we also record Binance's USDCUSDT price every second and convert the fair price into whatever currency the trade was actually quoted in. A USDC-quoted trade is compared to a USDC fair price, a USDT-quoted trade to a USDT one.
Only trades quoted in USDT or USDC are included, so the price is the real on-chain stable leg rather than a derived USD estimate — and a SOL-for-memecoin swap is never mistaken for a SOL price.
Why WBTC on Ethereum is compared to WBTC's price, not BTC's
The same problem as the stablecoin one above, one level up: none of these chains trade actual BTC, they trade a wrapper — BTCB on BSC, WBTC on Ethereum, cbBTC on Base and Solana. A wrapper is only worth the underlying if it actually trades at parity, and WBTC does not. Binance's own WBTC/BTC book had it 2.5 bps rich on 2026-08-21, and Kraken 4.8 bps.
Left uncorrected, that premium is charged to Ethereum's execution quality: BTC-on-Ethereum read 3.25 bps, of which 1.78 was the wrapper and only 0.64 was the actual half-spread a taker paid. So we record Binance's WBTC/BTC price every second too and score WBTC fills against a WBTC fair price. The basis is not a constant — it swung between −11 and +3.4 bps over the last nine weeks — so it has to be sampled, not assumed.
The other wrappers need no such correction: BTCB measures +0.09 bps from BTC, cbBTC +0.05 and +0.06, and WETH −0.36 — all inside this method's noise floor of about 0.4 bps. WETH is a useful check on that claim, since it is 1:1 with ETH by construction, so its reading is the noise floor rather than a real basis.
Why Solana trades are matched one second later than their own timestamp
Matching a trade to "the Binance price at its exact second" assumes the chain's own timestamp is the real time. On BSC, Ethereum and Base it is. On Solana it isn't: Solana doesn't stamp a block with a clock reading. It takes a time anchor from validator votes and then projects forward by multiplying the slot count by the nominal 400ms slot duration, so whenever real slots run slower than nominal, the timestamp falls behind wall clock.
We measured the size of that gap by regressing each trade's execution error against the Binance mid's own forward move — a continuous estimate rather than a whole-second guess. The EVM chains land on zero (PancakeSwap v3 +0.05s, Uniswap v3 −0.03s, Native +0.08s), confirming the method. Every Solana venue lands between 1.25 and 2.04 seconds. So Solana trades are matched at their timestamp plus one second; the EVM chains are not shifted.
Why +1s and not the ~1.9s average: only the part common to every venue is a clock error. The sharpest market maker on the chain (goonfi, 0.67 bps) still shows 1.25s, and a venue that tight cannot be a second and a half stale — that floor is the clock. Venues lagging further (Raydium 1.91s, Whirlpool 2.02s) keep their excess, because that part is genuine pool staleness a taker actually pays, and calibrating it away would delete a real finding. The correction is deliberately conservative: Solana still reads slightly wide, never flattering.
Effect on the numbers: EVM rows moved by at most 0.03 bps, while Solana tightened across the board — SOL at $10k–50k went from 1.62 to 0.96 bps, BTC-on-Solana from 1.87 to 1.40. Match coverage rose rather than fell, so nothing was dropped to achieve it.
This will need re-checking soon. Solana is cutting slot times from 400ms to 200ms in four steps (SIMD-0525), reaching mainnet from mid-August 2026. The nominal slot duration is the exact constant this projection uses, so the offset is expected to move at each step.
Best execution venue — every chain, side by side
The grid above asks which chain, then which venue inside that chain. Neither view ever puts a venue on Base next to a venue on BSC — but that is the comparison a taker actually makes. This table does: one row per venue × chain carrying the asset selected at the top of the page, ranked best-first at the order size you pick. The best cell in every column is boxed, so each size is readable on its own without re-sorting.
What "best" means follows the side toggle, because two honest measures rank venues differently. On both it is the median distance from the Binance price — how closely a venue tracks it in either direction. On buy or sell it is that side's signed cost, where a negative number is real price improvement. Lower is better either way. A venue with fewer than 30 matched trades at a given size still shows its row, but is never ranked there — a number we won't print must not be allowed to win.