Funding / OI divergence
Screen rising quoted carrying cost with shrinking contract-equivalent OI, or falling quoted cost with expanding OI.
Matched recorded snapshots
Up to 50 Binance / Bybit USDT linear contracts, selected by fresh source OI ≥ $1M when the universe was read. Inverse, USDC, HIP-3 and other venues are excluded.
50 considered · 50 valid · 0 matches · 50 below threshold · 0 excluded. Below threshold means a valid comparison meets neither paired filter, including changes in the same direction.
UTC targets: → . Actual recorded anchors appear in the table.
Inputs read at · cache age 0s · universe snapshot age 11s. These ages do not establish each field’s exchange publication time.
Covered result: 50 valid comparisons, but none meet either descriptive filter. Source data is available.
The inclusive cutoffs below are descriptive filters, not calibrated trading thresholds. They measure changes in two inputs and do not identify which side opened positions or predict price direction.
| Exact contract | Funding start / 8h | Funding end / 8h | Funding Δ (pp / 8h) | Raw USD OI Δ | Adjusted OI Δ | Actual start UTC | Actual end UTC |
|---|---|---|---|---|---|---|---|
| Funding up / adjusted OI down Funding change ≥ +0.005 pp / 8h; adjusted OI change ≤ −1%. | |||||||
| No matches in this group among 50 valid comparisons. | |||||||
| Funding down / adjusted OI up Funding change ≤ −0.005 pp / 8h; adjusted OI change ≥ +1%. | |||||||
| No matches in this group among 50 valid comparisons. | |||||||
Funding formula. Δ funding = 100 × (funding_end − funding_start), in percentage points per 8h. Endpoints are recorded quotes already normalized to 8 hours; no second interval scaling is applied. A quote comparison is not a settled payment, realized carry or an annualized delta. Zero funding is valid.
Adjusted OI formula. Δ adjusted OI (%) = 100 × ((OI_end / mark_end) / (OI_start / mark_start) − 1). Dividing USD OI by that same contract’s mark estimates contract-equivalent OI. Proportional growth in USD OI and mark leaves this measure unchanged. Raw USD OI Δ = OI_end − OI_start; its percent change is 100 × (OI_end / OI_start − 1).
Anchors and coverage. The end target is the start of the last closed five-minute UTC bucket. We select the latest row at or before each target, then validate its funding, USD OI and mark together. Both rows must be within five minutes of their targets and exactly 4h apart. Missing, nonfinite or nonpositive marks / OI, and missing or nonfinite funding, exclude the comparison; an older good row never replaces a newer invalid row.
Timestamps identify recorded engine snapshot buckets, not each exchange field’s original publication time. The screen cannot establish synchronization within a bucket or source-field freshness from a recent bucket alone. Exact venue / symbol identities are never substituted across contracts or venues. Within each group, rows rank by absolute funding change, absolute adjusted OI change, then venue / symbol. Contract links open the exact exchange market for further research.
Inputs originate from exchange public APIs through the existing recording service. Each of the three windows uses two bounded anchor lookups per selected contract. Successful reads are cached for up to 120 seconds within the same target bucket; failures for 30 seconds. No chart, backtest or directional signal is implied.