Fees are charged on volume, not on profit. A bot that trades often can have a real edge and still lose, because every fill pays the exchange whether the trade worked or not.

The math, with one example

Exchanges charge a fee proportional to each fill's notional value (price x quantity), at one of two rates:

• Maker rate: your order rested on the book and someone filled it (limit orders that don't cross).
• Taker rate: your order filled immediately against the book (market orders, crossing limits).

A round trip (open + close) pays the fee twice:

fee per round trip = fee on the entry fill + fee on the exit fill

Take a small bot. The fee below is a made-up round number, written as dollars per $1,000 traded; check your own exchange's schedule.

per round trip: $0.50 entry + $0.50 exit = $1.00
per day:        20 x $1.00              = $20
per 30 days:    30 x $20                = $600

That bot has to make more than $1.00 per round trip before it earns a cent. If its average gross edge is $0.80 per trade, it books a profit on paper and loses $0.20 on every round trip after fees, about $120 a month.

Now change one thing: post limit orders and get filled as maker, at a made-up $0.20 per $1,000 per side.

per round trip: $0.20 + $0.20    = $0.40
per 30 days:    20 x 30 x $0.40  = $240

Same strategy, same edge, and the monthly fee bill drops from $600 to $240. The $0.80 edge is now +$0.40 net per trade instead of -$0.20. Nothing about the signal changed; only how the orders hit the book.

Three things move this number more than strategy tweaks usually do:

1. Maker vs taker share. Market orders are convenient and, on most schedules, the more expensive side. (Some spot schedules charge the same rate for both at the base tier; check yours.)
2. Trade frequency. Fee drag scales linearly with round trips.
3. Your fee tier. Most exchanges lower rates with 30-day volume, and some also require a token balance. Check which tier you are actually on, not the headline rate.

Measure your own fee drag

Most major exchanges let you export trade history as CSV. Column names differ, so rename yours to time,side,price,qty,fee,realized_pnl (fee in quote currency, realized P&L before fees, 0 on opening fills) and run this:

"""fee_drag.py - how much of your trading P&L went to fees.

Usage: python fee_drag.py trades.csv
CSV columns (rename yours to match): time,side,price,qty,fee,realized_pnl
  fee          = fee paid on that fill, in quote currency (positive number)
  realized_pnl = P&L booked on that fill before fees (0 for opening fills)
"""
import csv, sys
from collections import defaultdict

def load(path):
    with open(path, newline="") as f:
        return [{k: (v if k in ("time", "side") else float(v)) for k, v in row.items()}
                for row in csv.DictReader(f)]

def report(rows):
    volume = sum(r["price"] * r["qty"] for r in rows)
    fees = sum(r["fee"] for r in rows)
    gross = sum(r["realized_pnl"] for r in rows)
    by_month = defaultdict(float)
    for r in rows:
        by_month[r["time"][:7]] += r["fee"]
    print(f"fills:              {len(rows)}")
    print(f"traded volume:      {volume:,.2f}")
    print(f"fees paid:          {fees:,.2f}")
    print(f"fees per $1,000:    {1000 * fees / volume:,.2f}")
    print(f"gross P&L:          {gross:,.2f}")
    print(f"net P&L:            {gross - fees:,.2f}")
    if gross > 0:
        print(f"fees ate:           {fees / gross:.1%} of gross profit")
    for month, fee in sorted(by_month.items()):
        print(f"  {month}: {fee:,.2f} in fees")

if __name__ == "__main__":
    if len(sys.argv) != 2:
        sys.exit(__doc__)
    report(load(sys.argv[1]))

Output on a synthetic file of 400 fills (made-up data, about 70% taker fills):

fills:              400
traded volume:      479,999.82
fees paid:          197.88
fees per $1,000:    0.41
gross P&L:          1,254.50
net P&L:            1,056.62
fees ate:           15.8% of gross profit
  2026-09: 96.24 in fees
  2026-10: 101.64 in fees

The line to watch is fees ate X% of gross profit. If that number is high, a better entry signal is not your cheapest fix. Moving fills to maker, trading less often, or reaching a lower tier usually is.

A few gotchas when you export:

• Some exchanges report fees in the asset you received (or in a fee token), not the quote currency. Convert before summing.
• Futures exports often split realized P&L and fees into separate rows or files. Join them on order id.
• Funding payments on perpetuals are a separate cost. This script doesn't count them; add them as another column if your bot holds positions across funding times.

I work on ReturnFee, a fee-cashback site. More plain explainers on how trading fees work: https://returnfee.com/learn/?utm_source=hashnode&utm_medium=post

1. Find the right table

Most exchanges have several schedules on the same page. Before you read any number, confirm which one applies to you:

• Spot or derivatives. Spot and futures/perpetuals almost always have separate tables, and futures rates are usually lower per side.
• Which futures. USDT-margined and coin-margined contracts can have different tables.
• Region or entity. Some exchanges run different legal entities per region with different fee pages. The page you land on from a search engine may not be the one your account uses.
• Account type. Regular, VIP, market-maker programme and sub-account rates can differ.

If in doubt, your account's own fee page (usually under account settings or "My fee rate") shows your live rate. Trust that over the public table.

2. Maker vs taker

Each tier lists two rates:

• Maker: you added liquidity. Your limit order sat on the book and someone else's order filled it.
• Taker: you removed liquidity. Your order filled immediately: a market order, or a limit order priced to cross the spread.

What decides it is whether your order rested on the book, not whether you clicked "limit". A limit buy above the best ask fills instantly and pays taker.

Most exchanges offer a post-only flag on limit orders. It cancels the order instead of letting it fill as taker, which is the simplest way to ensure the maker rate.

3. Read the tier table

A typical table looks like this. The numbers are made up for illustration; use your exchange's own.

Questions to answer from the footnotes:

• What counts as volume? Spot and futures volume are often counted separately. Some exchanges count only one side, some include sub-accounts.
• "Or" vs "and". Many tables let you qualify by volume or by a held balance. Some require both.
• When is your tier recalculated? Usually once a day on a rolling 30-day window, so a big trading week moves your tier the next day, not instantly.
• Is the tier per account or per user? Sub-accounts may or may not pool.

Most retail traders sit on the lowest tier, so tier 0 is usually the line to read.

4. Discounts that change the real rate

The table is rarely the final number. Look for:

• Fee-token discounts. Paying fees in the exchange's own token often cuts the rate by a fixed share. Check whether it applies to futures as well as spot, and whether it needs a toggle in settings.
• Promotional zero-fee pairs. Some pairs run at zero maker or zero taker for a period. Check the end date.
• Negative maker fees (rebates). Some higher tiers pay makers. You will rarely see this below the top tiers.

5. Things the fee table does not show

• Funding on perpetuals. Paid or received between longs and shorts every few hours. It is not an exchange fee, but it is a cost of holding a position.
• Spread and slippage. A market order on a thin pair can cost more in slippage than in fees.
• Withdrawal and network fees. Separate page, flat per withdrawal.
• Conversion fees. "Convert" or "quick buy" screens often price the fee into the quote instead of listing it.

6. Turn it into one number

Once you know your table, tier and discounts, your monthly fee bill is:

monthly fees = volume x (maker share x maker rate + taker share x taker rate)

Worked example with the made-up tier 0 above, $200,000 of monthly volume, 70% taker:

200,000 x (0.30 x 0.020% + 0.70 x 0.050%)
= 200,000 x (0.0060% + 0.0350%)
= 200,000 x 0.0410%
= $82 a month

Shift the split to 70% maker and the same volume costs:

200,000 x (0.70 x 0.020% + 0.30 x 0.050%) = 200,000 x 0.0290% = $58

Same trades, same tier, $24 a month apart, only because of how the orders hit the book.

Checklist

• Right product (spot / USDT futures / coin futures)
• Right entity / region
• My current tier, from the account page
• Maker and taker rate at that tier
• What counts toward the next tier, and when it updates
• Fee-token or other discounts, and whether they are switched on
• Funding, spread and withdrawal costs noted separately
• Monthly fee estimate = volume x blended rate

I work on ReturnFee, a fee-cashback site. More plain explainers on trading fees are at https://returnfee.com/learn/?utm_source=hashnode&utm_medium=post.