Week Ahead: Funding Rate Predictions

Two assets dominated the close of last week. ON token posted a 205% price move in 48 hours. SIREN extended its run another 167% on Bybit while carrying an average funding rate of -0.34% per settlement. The OU model ran overnight on 268 qualifying pairs. Here's what it found.

The Week That Was: Big Moves With Unusual Funding

Before looking ahead, the context matters.

ON token's 200%+ pump happened with funding near zero or slightly negative across all exchanges. Binance showed an average rate of -0.00122% over the move. KuCoin was -0.00191%. That's a spot-driven rally — longs weren't crowding in. No funding tail to wag.

SIREN is the inverse story. It pumped 167% on Bybit in 48 hours while sitting on the most deeply negative funding in the market — -0.9799% average per settlement over the past seven days, 0% positive settlements. Shorts have been getting paid consistently throughout this move. The crowd was positioned wrong and kept getting paid to stay wrong. That structural tension rarely holds indefinitely.

Both cases are clean examples of why watching the funding rate alongside price matters. Same headline result — asset up 160-200% — completely different mechanics underneath.

OU Model Update: 268 Pairs, Same Dominant Regimes

The Ornstein-Uhlenbeck fits cleared 268 pairs with R² > 0.5 this cycle, up from 206 last week. More data, more pairs passing the threshold. The top 10 by model quality with finite half-lives:

Three degenerate fits (persistence >> 1, half-life of 99,999 hours) were excluded. Same problem as last week: structural rate caps on illiquid instruments. Not tradeable signal.

The Fast Mean-Reverter: BABYUSDT Still Holds

BABYUSDT on Hyperliquid posted identical parameters to last week — half-life 15.8 hours, R² 0.962, equilibrium -0.003%. The 50% positive settlement rate means it's not directionally drifting. It oscillates. It reverts. The fit has been stable across two model cycles now, which is unusual. Most pairs shift as new data rolls in.

That stability is worth noting. A 15.8-hour half-life on a pair with R² near 1.0 is a rare combination. The signal is there. Whether it survives the next market event is the only real question.

VIRTUALUSDT sits at 72.1 hours with a 65.6% positive rate and near-zero equilibrium. Positive bias, fast-ish reversion. The model says longs get paid here more often than not, and when rates go negative they don't stay there long.

The Persistent Negative Drift Cluster

Seven of the top-fit pairs have structurally negative equilibria. The picture hasn't changed much from last week, but there are new entries worth flagging.

ICPUSDT on Hyperliquid enters the top 10 this cycle. Equilibrium at -0.014%, 19.4% positive rate, half-life 83.3 hours. Shorts have been paid 80.6% of the time historically. The model fits cleanly (R² = 0.875) with a half-life that's long enough to trade but short enough to matter.

RESOLVUSDT on Hyperliquid is new as well. Half-life 62.1 hours, 31.7% positive, equilibrium -0.0181%. Fairly aggressive negative drift for a pair that's only settling on one exchange. Worth monitoring for rate normalization if open interest climbs.

SAHARAUSDT on Binance remains at 0% positive rate, -0.1015% equilibrium, 88.5-hour half-life. No change in structure. Every settlement in its history has been negative. If a single positive settlement appears this week, it's a regime break worth flagging.

The Outlier: INJUSDT's Quiet Positive Bias

INJUSDT on Hyperliquid is the one pair trending the opposite direction: 77.8% positive settlement rate, equilibrium of essentially zero (0.0001%), half-life 77.5 hours, R² 0.867.

This isn't a carry monster. The equilibrium is near flat. But 77.8% positive rate with clean OU fit means rates here are reliably, slightly positive — and when they go negative, they tend to recover within a few days. That's a different profile from the negative-drift cluster dominating this list.

The interpretation: perp traders on INJ are chronically, modestly long-biased. Not aggressively — the equilibrium is too low for that. But the pattern has been consistent enough to show up in the model.

The Extreme Rate Table: Short Sellers Getting Paid

The highest absolute rates over the past seven days, by annualized impact:

All negative. Not one positive settlement for the top four. These instruments are in sustained, heavy short-favoring regimes. The SIREN rate is the extreme: 85.8% APR being paid to shorts, consistently, for a week, while the token rallied 400% from its trough. That combination breaks eventually. It either stops rallying, shorts capitulate, or both.

Three Things to Watch This Week

1. SIREN rate regime break. With the price now far above where the short crowd established positions, funding rate normalization is overdue. Any settlement above -0.1% would be the first in more than a week. Watch the SIREN rates page for the inflection.

2. BABYUSDT consistency test. Two model cycles with identical parameters. If it prints another 15.8-hour half-life next week, the signal is structural, not noise.

3. ON token funding drift. The pump happened with near-zero funding. If new longs pile in now at elevated prices, funding turns positive — and the OU clock starts. Watch whether ON USDT rates start accumulating.

268 pairs in the model. The negative drift regime dominates, but the outliers are where the interesting trades live.

---

Explore the data:

• All Funding Rates — live rates across all pairs and exchanges
• Predict Markets — trade these setups directly on Settled
• Most Predictable Funding Rates — how the OU model works