A screener for crypto perps that detects unusual volume and positioning — and grades itself honestly on whether that means anything.
Where this actually stands. Lynx is good at noticing that something unusual is happening on a specific coin. It has not been shown to know which way to trade it.
The live dashboard measures every tradeable idea at 1h, 4h and 24h. The longer horizon is not treated as second-class: a trade that takes a day is a valid outcome. Statistical claims are clustered by market day so a broad move that triggers many coins is not counted as many independent confirmations.
The original 120-day replay found no directional edge. Live results can differ by regime, so the engine keeps measuring all three horizons, after fees and a disclosed slippage estimate, and says “no directional bet is justified yet” until correlation-aware evidence supports one.
Lynx exists to do three jobs, in a loop that tightens every week:
Every 15 minutes Lynx screens low-float and privacy-coin perps on Hyperliquid for one specific fingerprint:
Only coins that could clear those bars are examined at all: a coin whose typical hour is $5k would need a 200× hour, so it never competes for one of the 16 candle fetches a free-tier run can afford. Among those, the ones running hottest right now go first.
The motivating case: in May 2026, ZEC perp volume spiked with rising OI and zero public news — days later the Orchard disclosure landed. Lynx was built to surface that shape early. Worth knowing: across 120 days and 25 coins, the specific price-down-plus-OI-up pattern that case represents fired 10 times. 96% of flags are new longs instead.
Research scores blend the recorded indicators — volume z, ratio to median, OI rise, funding size and trend, isolation, repeat flags, basis, persistence and measured taker flow — ten in all. Research weights are fitted from past returns by ridge regression. Automatic promotion is paused: reusing rolling holdouts cannot provide fresh confirmation. New paper ideas receive equal size; old scores stay unchanged. A confidence-calibration curve and an equal-weight benchmark keep the sizing honest.
The dashboard shows the benchmark beside confidence sizing at every horizon. Historical versions were promoted under a weaker “beat the incumbent” rule, even when both models lost. The historical diagnostic now purges overlapping labels and includes costs. Promotion requires fresh prospective evidence and an explicit testing budget. A frozen challenger now tests 12 complete weeks at all three horizons. Qualification cannot automatically change live sizing.
Lynx never trades. It suggests; a human decides and executes. That's a design constraint, not a roadmap gap.
Honest caveats, kept on purpose: the thesis is hindsight-selected from one motivating case; most flags will be false positives; volume-spike detection is inherently noisy; and acting on information that looks like a tip can carry real legal risk. A flag from Lynx earns a manual look — nothing more.
Two ideas have already looked significant in testing and turned out to be noise — a “fade the spike” edge, which vanished once the test universe was sampled properly, and a dedicated accumulation detector, killed by a train/holdout split. That is the method working, and it is why nothing here claims an edge it can't show you the t-statistic for.