On-chain deposit data across the top perp DEXs (90d history, trailing 30d window): every stablecoin transfer in/out of each protocol's deposit contract, parsed from chain logs — Hyperliquid via the Arbitrum bridge, Lighter via its Ethereum zkLighter contract, Aster via its BSC vault.
The depositor-momentum signal points in opposite directions for these two:
LIGHTER (long). The only venue with positive AND accelerating net deposits: +$19.7M over the last 30d vs -$33.8M the prior 30d (a +158% swing). New-depositor count held flat (-2.5%) while peers fell off a cliff. The flow is smart-money led — 12 wallets depositing >$1M each account for $374M of $382M gross — and it stays put (81% net-positive retention). Sophisticated capital is rotating in.
ASTER (short). The sharpest deterioration of the three. Net deposits flipped from +$50.9M to -$33.6M (-166%), and new depositors collapsed -62% (29.5k -> 11.2k). The base is retail (14k sub-$1k wallets) — the cohort that leaves fastest once incentives fade. Gross deposits are still large ($454M) but the trend has clearly rolled over.
(For reference, Hyperliquid's new depositors also fell -53% alongside a large net bridge outflow, but that figure is noisy with market-maker/bridge churn, so the cleanest expression is the LIT/ASTER pair rather than a HYPE leg.)
Thesis: net deposit inflows from sticky, sophisticated wallets lead price; retail outflows plus collapsing new-user growth lead underperformance. Long LIT / short ASTER isolates that divergence and is roughly neutral to perp-DEX beta.
Confirmation: LIT net flow stays positive and new-depositor count stabilizes; ASTER prints another net-negative deposit month.
Kill: ASTER deposits re-accelerate (fresh airdrop/incentive) or LIT's >$1M wallets start withdrawing.
Equal-depth method across all three (same transfer signal). Aster is BSC-vault only, so its deposit totals are a lower bound.
LIT
ASTER