Fees Optimizations

Freqtrade × Hyperliquid: Fee Optimization

When to use

• A user asks why their backtest PnL doesn't match live PnL.
• A user designs a tight-target scalp / market-making / grid strategy and you want to flag fee-as-a-fraction-of-edge concerns before they deploy.
• A user explicitly asks about maker vs taker, post-only, fee tiers, builder code, or slippage.
• Any time you're about to suggest an order_types/entry_pricing/exit_pricing block, anchor it to this skill's recommendations.

What you're actually paying

Fees on a Superior Trade Hyperliquid deployment have three layers, each tunable:

1. Hyperliquid exchange fees (the floor)

Hyperliquid's perp fees as of writing (verify against https://hyperliquid.gitbook.io/hyperliquid-docs/trading/fees for current numbers):

Spot is typically slightly lower; HLP/staking discounts can shave another 0.001-0.005%.

The big takeaway: maker is ~4.5× cheaper than taker. A round-trip taker trade pays ~9 bp; a round-trip maker trade pays ~2 bp. On a 0.6% scalp target, taker fees consume 15% of gross PnL before slippage.

2. Builder code fee (Superior's cut)

Every order on Superior-managed Hyperliquid deployments goes through a builder code, which adds a separate fee that goes to Superior. Default is 5 bp on volume traded, taker or maker. Verify the actual rate for the user's account in the Superior dashboard.

This fee is always paid regardless of order type. It can't be optimized away with maker/post-only — it scales with notional volume traded. The only knob is reducing turnover (fewer trades, larger size, longer holds).

3. Slippage (the variable cost)

Slippage = the difference between the price your strategy assumed and the price you actually filled at. On Hyperliquid:

• Liquid majors (BTC, ETH, SOL): typically 0–2 bp on $10K notional with tight spread.
• Mid-cap perps: 5-20 bp on $10K notional.
• Thin alts / new listings: can be 50+ bp on a single market order.

Slippage is always taker-flavored for market orders and near-zero for filled limit orders. Backtest engine uses last-trade price assumption, which understates real-world slippage on illiquid pairs.

The four knobs Freqtrade exposes

Knob 1: order_types — limit vs market

{
  "order_types": {
    "entry": "limit",
    "exit": "limit",
    "stoploss": "market",
    "stoploss_on_exchange": false
  }
}

Knob 2: entry_pricing / exit_pricing — where on the book

{
  "entry_pricing": {
    "price_side": "same",
    "use_order_book": true,
    "order_book_top": 1,
    "price_last_balance": 0.0,
    "check_depth_of_market": { "enabled": false }
  },
  "exit_pricing": {
    "price_side": "same",
    "use_order_book": true,
    "order_book_top": 1
  }
}

Default for cost-sensitive strategies: price_side: "same", use_order_book: true, order_book_top: 1, price_last_balance: 0.0. This posts at the best bid (long) / best ask (short), maker-only unless the market crosses you.

Knob 3: unfilledtimeout — how patient is the maker?

{
  "unfilledtimeout": {
    "entry": 10,
    "exit": 10,
    "exit_timeout_count": 0,
    "unit": "minutes"
  }
}

A limit order sitting on the book is great until the market walks away. unfilledtimeout is when Freqtrade gives up and re-prices (or cancels). For a 5m strategy, entry: 5 minutes is sane — give the limit one bar to fill, then chase. For a daily DCA, entry: 60+ is fine.

exit_timeout_count: 3 means: after 3 timeout cycles on the exit, fall back to a market order. Good for ensuring the exit eventually happens.

Knob 4: stake_amount — fewer, bigger trades

The builder code fee scales with notional volume. Two trades at $1000 each pay the same builder fee as four trades at $500 each. If your strategy works at higher position size, prefer fewer larger trades to reduce per-trade fixed cost.

This isn't a Freqtrade knob per se, but an engineering choice: lower-frequency strategies amortize per-trade fees better.

Recommended starting block (cost-sensitive)

For any strategy where fees matter (scalping, grid, frequent entries), start here:

{
  "order_types": {
    "entry": "limit",
    "exit": "limit",
    "stoploss": "market",
    "stoploss_on_exchange": false,
    "trailing_stop_loss": "market"
  },
  "entry_pricing": {
    "price_side": "same",
    "use_order_book": true,
    "order_book_top": 1,
    "price_last_balance": 0.0
  },
  "exit_pricing": {
    "price_side": "same",
    "use_order_book": true,
    "order_book_top": 1,
    "price_last_balance": 0.0
  },
  "unfilledtimeout": {
    "entry": 10,
    "exit": 10,
    "exit_timeout_count": 3,
    "unit": "minutes"
  }
}

For a "fill at any cost" deployment (e.g. funding harvest where missing the entry costs more than the spread), use price_side: "other" and entry: "market".