# You Were Right About the Direction and Still Lost Money: How Funding Rates Drain Perp Positions
*Educational content, not financial advice. This is a breakdown of a cost mechanism on perpetual futures — not a call to trade them.*
The call was right. The chart did what you thought it would do. And the position still lost money.
Not to a fakeout. Not to slippage. To a line item most traders never read: the **funding rate**. On June 23, 2026, BTC funding sat near **+0.0043% per 8 hours** (CoinUnited research citing Coinglass data). Sounds like dust. On $100,000 notional — just $1,000 of margin at 100x — that dust is **$4.30 every 8 hours**, about **$12.90 a day**, while price goes absolutely nowhere. Hold through a crowded market instead, where rates spike to 0.05%–0.15% per interval (a March 2026 Bitcoin.com guide annualizes that at roughly 55%–164% of notional), and a two-week hold can burn 5–15% of notional in funding alone.
Right direction. Wrong carry. The market paid your thesis nothing and charged it rent anyway.
## What you actually agreed to
Funding is the periodic payment that keeps a perpetual future tethered to spot. Every 8 hours (typically 00:00, 08:00, 16:00 UTC), whichever side is crowded pays the other: positive funding means longs pay shorts, negative means shorts pay longs. It's purely peer-to-peer the exchange processes the transfer and keeps none of it. The concept dates back to BitMEX's perpetual swap in 2016, and every perp market since has inherited it.
The detail that hurts: **funding is calculated on your full notional, not your margin**. Margin decides where you get liquidated. Notional decides what you pay. Leverage connects the two, and traders size the first while forgetting the second.
## Three mistakes that turn right calls into red P&L
**Mistake 1: Ignoring funding on multi-day holds.** The bill compounds quietly cost scales with full notional while your profit cushion scales with margin. At +0.15% per interval, that's $450/day on $100k notional. A "temporary" hold through nine settlements is a second, invisible position fighting your first one.
**Mistake 2: Assuming funding is an exchange fee.** It isn't. Another trader collects every dollar you pay and in crowded-long regimes, simply holding the short side has been the collecting position. Funding is income for one side and expense for the other. Knowing which side of the crowd you're standing on matters more than the rate itself.
**Mistake 3: Treating extreme funding as an entry signal.** Sustained readings above ~+0.10% per interval signal crowded longs; strongly negative readings below about -0.05% signal packed shorts and squeeze fuel. KuCoin News reported (Aug 1, 2026) that retail now tracks funding nearly as closely as spot price. But extremes are context, not timing triggers by themselves crowded longs can stay crowded for weeks. Read funding alongside open interest and price structure; never alone.
## Check funding before you enter - two minutes, no excuses
Open a Coinglass-style funding dashboard (or your venue's own funding page) and note three things before any multi-day perp position:
- **Current rate and sign** which side of the invoice would you be on?
- - **Settlement times in UTC** how many payments does your planned hold cross?
- - **Rate history** is this regime calm (~0.004%) or spiking (0.05%+)?
If your expected hold crosses several settlements during a spike regime, the honest question isn't "will my direction work?" It's "can my thesis clear the carry?"
## The sub-8-hour loophole
Funding is charged at settlement snapshots but only on snapshot-settlement venues. Open and close inside a single 8-hour window and you pay **zero funding**, regardless of the rate. Continuous-accrual venues close that loophole: their meter runs from the moment you open. That's structurally different economics from a multi-day hold scalpers and swing traders live in different cost universes on the same instrument.
One more wrinkle: the classic 8-hour windows created timing games (flip after the snapshot, dodge the payment), and newer venues are killing them. Through 2026, markets have been shifting toward continuous, per-block funding accrual enabled by millisecond-scale oracle feeds like Pyth's. Hyperliquid pays hourly and caps funding at 4%/hour (per its official docs); dYdX v4 TWAP-smooths its premium index; GMX v2 accrues funding continuously per second instead of snapshot windows and layers a pool-utilization borrowing fee on top. Same coin, different rent depending on where you hold it.
## Do the math once it takes five lines
# funding_drag.py - what does carry really cost?
INTERVALS_PER_DAY = 3 # standard 8-hour settlements
def funding_bill(notional, rate_per_8h_pct, days):
return notional * (rate_per_8h_pct / 100) * INTERVALS_PER_DAY * days
notional = 100_000 # $1,000 margin at 100x
rates = [0.0043, 0.05, 0.15] # calm / trending / crowded (BTC, June 2026)
for r in rates:
bill = funding_bill(notional, r, days=14)
annual = r * INTERVALS_PER_DAY * 365
print(f"rate {r}% per 8h | 14-day bill ${bill:,.2f} | {round(annual, 1)}% of notional per year")
# rate 0.0043% per 8h | 14-day bill $180.60 | 4.7% of notional per year
# rate 0.05% per 8h | 14-day bill $2,100.00 | 54.8% of notional per year
# rate 0.15% per 8h | 14-day bill $6,300.00 | 164.2% of notional per year
Run it with *your* notional and intended hold. If the 14-day figure makes you flinch, that flinch is information about sizing, not about direction.
## A journal template that catches the leak
Date / pair / venue:
Direction & size (notional):
Funding at entry (+/- % per 8h):
Settlements crossed: ___
Funding paid / collected: $___
Net of directional P&L: $___
Thesis note:
Fill it on every close longer than two days. After ten entries you'll know exactly how much carry has been skimming from your edge — most traders genuinely don't know their number.
The direction call gets all the attention. Carry quietly decides whether being right actually pays. Check yours before the next settlement does.
---
*Tags: trading, crypto, perpetual futures, funding rate, risk management, education*
