Funding Rate

The periodic payment exchanged between longs and shorts on a perpetual futures contract that tethers the perp price to the underlying spot price. In a bull market it is usually positive — longs pay shorts. The single biggest perp-specific cost the simulator does NOT model.

In plain English

A perpetual future (perp) is a crypto futures contract with no expiry date. Because it never settles, the exchange needs a mechanism to stop the perp price from drifting away from the real (spot) price. That mechanism is the funding rate: every few hours, one side pays the other.

• When more traders are long (betting price rises), funding is positive → longs pay shorts. The normal state in a bull market.
• When more traders are short, funding is negative → shorts pay longs.

If you hold a leveraged long through a hot bull market, you bleed funding continuously — a real, recurring cost of carrying the position, separate from price movement.

Formula / mechanic

Funding is charged on the notional (leveraged) position size at each funding tick (commonly every 8 hours, ~3×/day). A baseline rate of 0.01% per 8h annualizes to roughly:

0.01% × 3 ticks/day × 365 ≈ ~11% / year

In euphoric regimes the rate runs far hotter — multiples of baseline.

Why it matters for this fleet — and why it does NOT rescue the buy-and-hold gap

The engine models zero funding (confirmed: no funding references in source; on the fidelity ledger in simulator fidelity). The intuitive hope is "we don't charge funding, so we're flattering buy-and-hold unfairly." That is backwards.

• The buy-and-hold benchmark (buy and hold) is a spot-like 1× hold. A real spot hold pays no funding — so the benchmark number is honest as "vs holding spot."
• The strategies hold perp positions across funding ticks but are charged nothing. In reality a leveraged perp strategy held through hot funding would do worse than the sim shows; spot hold would be unchanged.

Net: real funding would widen the gap against hold-heavy strategies, not narrow it. Both sides are funding-free in the sim, so the comparison is internally consistent — but funding is not the "missing variable" that makes the numbers less stark. (The variable that does is risk adjusted return.)

Funding's bias is non-uniform — it scales with average holding time, hitting slow trend-followers hardest and fast scalpers least. Direction: every position held across funding ticks overstates PnL.

Examples from the live fleet

• The dossier's edge rows are exactly the ones funding would punish. The only strategies in run 83 with any measurable edge are the long-hold 4h and 1h swings. The fidelity check (§7) measures their hold durations: survivors hold 1.8 days (1h signals) to ~9 days (4h), and the 50/200 pairs hold ~120 days. At ~3 funding payments/day, a single 4h trade carries ~27 unmodeled funding payments — and a 50/200 hold carries hundreds. The simulator charges none of it, so the rows that look best are the ones whose real-account results are most inflated by the gap.
• id522 (ETH 4h 50× long) — the one edge-significant hold-beater. As a 4h swing held ~9 days at 50× notional, it is precisely the profile a real funding charge would erode most; its in-sim alpha reads before any of that cost.

Related

• simulator fidelity
• buy and hold
• basis
• liquidation
• leverage

Sources

• wiki/qa-sessions/2026-06-22-session.md#q1 (first formal entry)
• growth/content/dossiers/ema-cross/1-analysis.md §7 (hold-duration funding check, run 83)
• wiki/concepts/simulator-fidelity.md §3 "No funding rate — the load-bearing gap"