Normalizing Funding Rates Across Venues
Raw rates, settlement intervals, predicted rates, and settled rates are not interchangeable. This note defines a common-period normalization and identifies contract and execution differences that a single number cannot remove.
A common interval is not an annual return
If venue A settles 0.010% every eight hours while venue B settles 0.006% every four hours, comparing the raw numbers understates B's holding cost. Kieran Lab converts eligible pairs to a common eight-hour window: normalized rate = raw interval rate × 8 ÷ interval hours. This does not assume later settlements stay fixed.
Interval eligibility is evaluated pair by pair rather than by excluding all hourly quotes. A 1h quote can pair with 1h / 4h, a 4h quote can pair with 1h / 4h / 8h, and an 8h quote can pair with 4h / 8h. Only direct 1h-versus-8h pairs are excluded because their reset frequencies differ too much. Simple annualization likewise communicates scale rather than expected return.
Predicted, settled, and asynchronous observations
Most venue screens show an estimate for the next settlement, which can change until the event; historical analysis should use settled funding. A cross-venue snapshot can still compare one contract one minute from settlement with another four hours away.
The comparator retains each venue's native interval and next settlement. Asynchronous legs create cash-flow and path risk: one side can settle before the other, while the subsequent rate may erase the original spread. Normalization should expose, not conceal, that risk.
Contract differences beyond funding
Matching tickers do not guarantee identical economic exposure. Index composition, mark-price rules, collateral, contract multipliers, funding caps, and auto-deleveraging can differ. Stablecoin quotes are also not risk-free cash equivalents.
An executable comparison must still validate depth, slippage, fee tier, order constraints, margin usage, and basis. Kieran Lab's spread ranking is a research starting point, not an executable yield quote.