AI funding signals that reduce scanner noise
The goal is not to show every possible pair first. The goal is to rank the pairs that deserve review before the market regime changes.
What the AI signal is for
AI ranking helps compress a noisy funding universe into a smaller review list. It does not replace risk management or execution checks.
A useful signal should explain why a route appears, show the route mechanics, and point to the next validation step.
- AI Top Pairs ranks current route candidates.
- AI Reco provides a separate recommendation view.
- Risk tags prevent high APY from hiding weak trade quality.
How to use the signal
Treat the signal as triage. Review the highest-quality candidates first, backtest routes that still look valid, and then decide whether the execution venue and size fit your account.
- Prioritize high score plus clean risk labels.
- Avoid routes with unexplained spikes or unstable funding.
- Keep execution and monitoring outside the model's assumptions.
Why ranking by APY alone misleads
The highest funding rate on the board is usually the one nobody wants. Extreme rates appear on thin markets, on assets in the middle of a violent move, and on venues where the rate is high precisely because holding the position is dangerous. Sorting a scanner by APY sorts by risk as reliably as by reward.
A useful ranking has to combine the rate with the things that decide whether you can hold the position: how deep the book is, how long the rate has persisted, how far the two legs can drift apart, and what it costs to get out if it turns.
- The top of an APY-sorted list is a list of the most dangerous positions.
- Persistence matters more than magnitude: a modest rate that holds beats a spike that does not.
- A rate you cannot exit is not income.
How to read a score without trusting it blindly
A score is a compression of several measurements into one number, which makes it easy to compare routes and easy to be wrong about why. Use it to shorten the list, then look at the underlying values before committing: the spread history, the depth on the thinner leg, and the funding interval on both venues.
The most valuable thing a signal does is remove the noise: the dozens of routes that look attractive for a few minutes and are gone. What it cannot do is tell you that the next hour resembles the last one.
- Use the score to shortlist, not to decide.
- Always check the thinner leg: the route is only as good as its worst side.
- A signal describes what has been true. It makes no claim about what happens next.
Questions
Does AI Reco place trades?
No. It ranks and explains candidates. Trade execution remains user-controlled.
Why not just sort by APY?
APY alone ignores entry gap, stability, liquidity, sign flips, and venue-specific risk.
What data goes into a funding signal?
The rate history on both venues, how long the current sign has persisted, the depth available on each leg, the spread between the two marks, and the cost of a round trip including fees. The individual inputs matter more than the composite: they are what you check before committing.
Should I trade the highest-ranked route automatically?
No. A ranking is a filter, not a decision. It tells you which of several hundred routes deserve a look; whether the top one is tradeable depends on your size against the depth on the thinner leg, which no ranking can know.
How often do funding signals change?
Continuously, because funding is recomputed on every venue's own schedule and the book moves between those points. What matters is not the freshness of the number but whether the condition it describes has persisted long enough to still be there when you finish opening both legs.