Printed vs Real

Every funding screener ranks by a number you cannot trade. This board takes that exact number and subtracts what it costs to actually do it.

Updated — · Sep 21, 06:24 PM

Source: venue public APIs, fetched directly — per-figure source on hover

Delta-neutral funding-arb combinations across 11 venues right now

100 / 1,973

are actually tradeable — the rest die on an empty book, on costs, or because the spread is gone before it repays them. 0 of the top 20 by printed APR survive.

The row every screener puts first right now is XMR — short TxFlow / long Lighter, 254% APR. Its thin leg traded $359k in 24 hours. You have to go 9 rows down the printed leaderboard before one has a book behind it at all.

Best that survives

105% net

ENA · short Lighter / long Backpack

Killed by the book

1,148

thin leg under $500k of 24h volume

Observation window

152.6h

240 logged snapshots

Where they die

Each stage contains the next, so the bars are directly comparable. Gates run in order and a row dies once.

1,973 combinations — ranked by gross carryprinted1,973ranked by gross carry825 combinations — thin leg can absorb sizebook825thin leg can absorb size575 combinations — carry clears fees + spreadcost575carry clears fees + spread100 combinations — repays that cost before it diestime100repays that cost before it dies

Not in the funnel: 69 rows sit below the smallest rung we test (5% APR) and 0 sit above a level this pair has never reached in the window. Neither is counted as surviving.

What actually kills them: time

It is not the fees. On the biggest printed spreads that at least have a book behind them — the first appears at rank 9, execution cost takes a median of 2.3 points off the printed APR. That is a rounding error next to the number itself, and drawing it would show two dots on top of each other.

The kill is the clock. The grey dot is how many hours the carry must run to repay its one-time cost; the mint dot is how long a spread that big has actually lasted on that pair inside our window. Whenever grey sits to the right of mint, the trade needs more time than it ever gets — the row is labelled at the end of the line, so nothing here depends on telling two colours apart.

hours of carry needed to break evenhours a spread this size has lasted

0h16h32hGRAM · short Lighter / long Binance needs 8.9h · has lasted 5.4h capture over that life, after cost: -0.063% of notional verdict: dies firstGRAM Lighter→Binancedies firstGRAM · short Lighter / long Bybit needs 9.4h · has lasted 5.4h capture over that life, after cost: -0.073% of notional verdict: dies firstGRAM Lighter→Bybitdies firstJTO · short Backpack / long Binance needs 16.2h · has lasted 5.7h capture over that life, after cost: -0.13% of notional verdict: dies firstJTO Backpack→Binancedies firstJTO · short Backpack / long Hyperliquid needs 15.4h · has lasted 5.7h capture over that life, after cost: -0.12% of notional verdict: dies firstJTO Backpack→Hyperliquiddies firstJTO · short Backpack / long Bybit needs 17h · has lasted 5.7h capture over that life, after cost: -0.14% of notional verdict: dies firstJTO Backpack→Bybitdies firstXPL · short Extended / long Lighter needs 16.3h · has lasted 4.9h capture over that life, after cost: -0.14% of notional verdict: dies firstXPL Extended→Lighterdies firstJTO · short Backpack / long Extended needs 31.8h · has lasted 5.7h capture over that life, after cost: -0.32% of notional verdict: dies firstJTO Backpack→Extendeddies firstXPL · short Binance / long Lighter needs 13.1h · has lasted 4.9h capture over that life, after cost: -0.1% of notional verdict: dies firstXPL Binance→Lighterdies firstXPL · short Hyperliquid / long Lighter needs 12.2h · has lasted 4.9h capture over that life, after cost: -0.09% of notional verdict: dies firstXPL Hyperliquid→Lighterdies firstXPL · short Bybit / long Lighter needs 13.9h · has lasted 4.9h capture over that life, after cost: -0.11% of notional verdict: dies firstXPL Bybit→Lighterdies firstENA · short Lighter / long Backpack needs 11.5h · has lasted 21.3h capture over that life, after cost: 0.12% of notional verdict: tradeableENA Lighter→BackpackrepaysNEAR · short Backpack / long Binance needs 16.9h · has lasted 17.2h capture over that life, after cost: 0.003% of notional verdict: tradeableNEAR Backpack→Binancerepays

What actually survives

PairShort → LongPrintedRealNeedsHas lastedKeepsThin leg
ENALighter → Backpack107%105%11.5h21.3h+0.12%$733k
NEARBackpack → Binance104%101%16.9h17.2h+0.003%$5.2M
NEARBackpack → Extended103%101%16.1h17.2h+0.012%$5.2M
XMRExtended → Bybit79.1%77.0%18.8h33.9h+0.136%$6.9M
XMRExtended → Lighter79.8%76.8%27.4h33.9h+0.059%$6.0M
XMRAster → Lighter75.3%72.9%23.3h33.9h+0.091%$2.3M
XMRAster → Bybit74.6%72.5%20h33.9h+0.119%$2.3M
XMRHyperliquid → Lighter71.8%70.0%18.3h33.9h+0.128%$6.0M
XMRHyperliquid → Bybit71.1%68.7%24.6h33.9h+0.075%deep
ENALighter → TxFlow68.6%67.0%16.6h21.3h+0.037%$914k
XMRBinance → Lighter51.4%49.5%27.3h33.9h+0.039%$6.0M
XMRHyperliquid → Backpack48.3%46.0%34.5h37.7h+0.018%$528k
HYPERISEx → Lighter44.2%43.2%15.9h19.9h+0.02%$7.3M
NEARLighter → Aster34.6%33.4%25.3h50.1h+0.098%$7.8M
ENALighter → Aster34.6%33.4%25.3h29.1h+0.015%$3.1M
NEARLighter → TxFlow34.6%33.0%32.9h50.1h+0.068%$1.0M
NEARLighter → Extended34.2%33.0%25.6h50.1h+0.096%$16.0M
ENALighter → Extended34.2%33.0%25.6h29.1h+0.014%$4.1M
NEARLighter → Binance34.6%32.6%40.5h50.1h+0.038%$24.4M
NEARLighter → Bybit34.6%32.5%43h50.1h+0.028%$24.4M
NEARRISEx → Hyperliquid29.4%27.3%50.7h94.5h+0.147%$1.3M
PUMPHyperliquid → Aster27.1%25.3%48.5h64.3h+0.049%$2.6M
PUMPBinance → Aster27.1%25.1%51.7h64.3h+0.039%$2.6M
BTCHyperliquid → TxFlow26.5%25.0%40.1h119.9h+0.242%$37.1M
PUMPHyperliquid → TxFlow26.6%24.4%59.3h64.3h+0.015%$1.2M

What this board cannot claim. Two limits we are not hiding. (a) The lifetime test is PAIR-level: our logged history stores the best spread per pair, not per venue-combination, so a row is credited with the longest run any combination on that pair achieved — generous, not strict. Per-combination logging starts with this build. (b) Runs are UPPER bounds: snapshots land ~1.4h apart at the median, so a spread that died and came back between two reads counts as continuous, and anything shorter than the gap is invisible. The worker's 15-minute series shows shorter lives than this board can resolve.

Printed = gross carry (apr_short − apr_long) — the exact number funding screeners rank by, with no fees, no spread, no book depth and no history behind it. Real = the same row put through three gates in order: (1) the thin leg must clear a 24h-volume floor; (2) the carry must clear blended round-trip cost (limit on the thin leg, market on the liquid one — the favourable assumption, chosen deliberately); (3) it must repay that cost in fewer hours than a spread of the same size has ever actually lasted on that pair. Verdicts: no_book / costs / too_slow / unproven / tradeable — 'unproven' means we lack the history to judge it and is never counted as surviving.