funding-rate-arbitrage

Strategy: Funding · Negative-Rate Harvest

When to use

A user wants to capture funding payments by being on the side that gets paid:

• Long a perp when funding APR is deeply negative (shorts paying longs).
• Short a perp when funding APR is deeply positive (longs paying shorts) — variant below.

This is the most profitable of the six standard templates in our audit and the engine supports it natively. Promote this template when a user asks "what's a strategy that actually works?".

Backtest reference (the real one)

Window	BTC/USDC:USDC 1h, 2026-01-01 → 2026-05-01 (BTC −13% over the window)
Trades	55
Win rate	58.2%
Wallet PnL	+1.38% / +$13.76
Profit factor	1.57
Sharpe	1.52
Max drawdown	0.58%
Avg holding	9h 40m
Backtest ID	01kqyz3ejgy5b7tdemhb6gj9nf

~+4% APR on a single pair through a market that fell 13%. A multi-pair scan (e.g. top 20 perps) compounds this.

The Freqtrade primitive that makes this work

The DataProvider exposes funding-rate candles directly. No Hyperliquid REST call from inside the strategy is needed for backtest — Freqtrade auto-downloads funding history when it sees a candle_type="funding_rate" request:

funding = self.dp.get_pair_dataframe(
    pair=metadata["pair"],
    timeframe="1h",          # Hyperliquid funds hourly
    candle_type="funding_rate",
)

The returned dataframe has the same shape as OHLCV — date, open, high, low, close, volume — but open is the funding rate at the start of that hour, expressed as a fraction (-0.0000135 = -0.0014% per hour). Annualize as funding_rate * 24 * 365.

The naive v1 (placeholder column filled with 0.0) produced 0 trades. v2 with dp.get_pair_dataframe(...) produced 55 trades and Sharpe 1.52.

Reference implementation

from freqtrade.strategy import IStrategy
from datetime import datetime
import pandas as pd
import talib.abstract as ta


class FundingHarvestStrategy(IStrategy):
    minimal_roi = {"0": 100.0}   # let funding work; no profit-target exit
    stoploss = -0.05
    trailing_stop = False
    timeframe = "1h"
    process_only_new_candles = True
    startup_candle_count = 30
    can_short = False

    def populate_indicators(self, dataframe: pd.DataFrame, metadata: dict) -> pd.DataFrame:
        # Hyperliquid funds hourly — request 1h funding-rate candles.
        try:
            funding = self.dp.get_pair_dataframe(
                pair=metadata["pair"],
                timeframe="1h",
                candle_type="funding_rate",
            )
        except Exception:
            funding = pd.DataFrame()

        if not funding.empty and "open" in funding.columns:
            f = funding[["date", "open"]].rename(columns={"open": "funding_rate"}).copy()
            dataframe = dataframe.merge(f, on="date", how="left")
            dataframe["funding_rate"] = dataframe["funding_rate"].ffill().fillna(0.0)
            # Annualize hourly funding: APR = rate * 24 * 365.
            dataframe["funding_apr"] = dataframe["funding_rate"] * 24 * 365
        else:
            dataframe["funding_rate"] = 0.0
            dataframe["funding_apr"] = 0.0

        dataframe["atr_24"] = ta.ATR(dataframe, timeperiod=24)
        return dataframe

    def populate_entry_trend(self, dataframe: pd.DataFrame, metadata: dict) -> pd.DataFrame:
        # Long when funding APR is deeply negative (shorts paying longs).
        dataframe.loc[
            (dataframe["funding_apr"] < -0.10) & (dataframe["volume"] > 0),
            "enter_long",
        ] = 1
        return dataframe

    def populate_exit_trend(self, dataframe: pd.DataFrame, metadata: dict) -> pd.DataFrame:
        # Exit when funding flips back to non-negative (no more carry).
        dataframe.loc[(dataframe["funding_apr"] >= 0.0), "exit_long"] = 1
        return dataframe

    def custom_exit(self, pair: str, trade, current_time: datetime,
                    current_rate: float, current_profit: float, **kwargs):
        # Hard timeout — the entry condition was wrong if we're still in
        # after 24h without an exit signal.
        elapsed_h = (current_time - trade.open_date_utc).total_seconds() / 3600.0
        if elapsed_h >= 24:
            return "timeout_24h"
        return None