Perpetual swaps have no expiry, so exchanges use a funding rate to keep the perp price tethered to spot. When the perp trades above spot, longs pay shorts; when it trades below, shorts pay longs. The payment is periodic — usually every eight hours — and the rate floats with the size of the basis. That single mechanism turns funding into one of the cleanest reads on how the crypto funding rate market is positioned right now.
This guide is about using funding as a directional and positioning input — not a holy grail. We will not quote returns, win rates, or a Sharpe number, because we do not invent them. What we will do is explain what the numbers mean, why cross-venue dispersion matters more than any single venue, and how to combine funding with open interest and the long/short ratio to build a defensible funding rate trading strategy.
What positive vs negative funding actually tells you
Positive funding means longs are paying to hold their position — the crowd is leaning long and willing to bleed carry to stay there. Persistently high positive funding is a crowded-long tell: it raises the cost of being long and the risk of a long squeeze if price stalls. Negative funding is the mirror image — shorts are paying, positioning is bearish, and the setup for a short squeeze builds. The level matters, but so does the change. A funding rate drifting from mildly positive toward extreme, while price goes nowhere, is the classic signature of a trade getting crowded.
Funding is a positioning gauge, not a price prediction. It tells you who is paying to be in the trade and how lopsided the book has become — context for risk, not a standalone buy or sell.
Why cross-venue dispersion is the real signal
Any one venue's funding can be noisy or distorted by local flow. The more durable signal lives across venues. QUANT_API exposes both halves of that picture: funding.funding_avg gives you the blended perpetual funding rate across the venues we ingest, while funding.funding_dispersion measures how far individual venues disagree. We also break funding out by venue type — funding.funding_usdtm for USDT-margined perps and funding.funding_hl for Hyperliquid — so you can see whether a regime is broad-based or one corner of the market.
High dispersion is the interesting state. When funding.funding_avg is elevated but funding.funding_dispersion is also wide, the crowd is not uniformly positioned — some venues are far more stretched than others, which often precedes a venue-specific unwind. When the average is extreme and dispersion is tight, the entire market is leaning the same way, and that is when squeeze risk is most systemic. Reading the two together beats staring at any single exchange's rate.
Combining funding with open interest and the long/short ratio
Funding tells you the price of the lean; open interest tells you how much capital is behind it. Pair funding.funding_avg with positioning.doi_usdtm_30m — the change in USDT-margined open interest over 30 minutes — to separate two very different stories. Rising price with rising OI and climbing funding is fresh leverage piling in (a real, but increasingly crowded, trend). Rising price with falling OI is shorts covering, not new conviction. As an open interest signal, that distinction is the whole game and the reason a raw funding read alone misleads. Add positioning.lsr_top — the top-trader long/short ratio — as a sanity check on who is on which side. When smart-money long/short skews against retail-driven funding, the disagreement itself is information. None of these are signals you trade in isolation; they are a positioning stack you read together. The full set lives in the catalog, addressed as category.signal@window[:transform] across 14 windows and 9 transforms.
- Funding high + OI rising + dispersion tight → broad crowded long; size down, watch for a long squeeze.
- Funding negative + OI rising → fresh shorts stacking; squeeze fuel if price refuses to break down.
- Funding extreme on one venue only (wide dispersion) → local distortion, not a market-wide regime.
- Funding cooling while price holds → leverage flushing without a price break; often a healthier trend.
Resolving the features — point-in-time, no look-ahead
Here is the part that makes a funding study trustworthy. Every value QUANT_API returns is computed strictly from data with a timestamp at or before your as_of — live or historical. So when you build a funding rate signal in a backtest, each row reflects exactly what was knowable at that decision time. No revised funding prints leaking backward, no future OI bleeding into a past bar. The call below pulls a live funding-and-positioning snapshot for BTC and ETH.
curl -G https://api.quant-api.dev/v1/features/live \ -H "Authorization: Bearer fk_live_<secret>" \ --data-urlencode "assets=BTC,ETH" \ --data-urlencode "features=funding.funding_avg@1h:level,funding.funding_dispersion@1h:zscore,funding.funding_usdtm@1h:level,positioning.doi_usdtm_30m@30m:delta,positioning.lsr_top@1h:level"
To study how a funding regime behaved historically, post your own timestamps to POST /v1/features/historical and you get a point-in-time matrix at exactly those instants — the same leak-free guarantee, applied to the past. From there, POST /v1/backtest lets you express a rule over these features without rebuilding the data plumbing yourself. The mechanics are in the docs, and more positioning walkthroughs sit in other guides. The moat is simple: a backtest is only honest if every input was knowable at decision time, and point-in-time funding is the difference between a strategy and an artifact.