Bid-ask spread: the cost everyone pays and most traders ignore

At any moment in any market, there are two prices: the bid (the highest price someone is willing to pay) and the ask (the lowest price someone is willing to sell at). The gap between them is the bid-ask spread. Cross the spread to get filled instantly, and you pay it; sit on the bid or ask waiting for the market to come to you, and you might collect it.

The mechanics

Bid side (buyers)             Ask side (sellers)
─────────────────             ────────────────────
$50,001  (3.2 BTC)            $50,004  (1.8 BTC)
$50,000  (5.1 BTC)            $50,005  (2.6 BTC)
$49,999  (1.4 BTC)            $50,006  (4.7 BTC)
$49,998  (8.0 BTC)            $50,007  (3.3 BTC)

Best bid:  $50,001
Best ask:  $50,004
Spread:    $3            (or about 0.006% / 0.6 bps)

# Market buy 1 BTC right now → fills at $50,004 (best ask).
# Market sell 1 BTC right now → fills at $50,001 (best bid).
# Round trip on 1 BTC = pay the $3 spread.

Spread is the market maker's fee. They post both sides of the book and collect the spread on every fill that crosses through them. In return, they take on inventory risk — what they sell to you, they have to buy back later. The spread compensates them for that risk.

What determines spread

• Liquidity. Major pairs (BTC/USDT, ETH/USDT, SPY) have tight spreads because dozens of market makers compete. Small-cap alts can have 50-200 bps spreads because there's almost no competition.
• Volatility. When the market is calm, market makers post tight spreads. When volatility spikes, they widen — sometimes to 5-10× the normal spread for a few minutes.
• Time of day. Spreads in equities widen in the first and last 30 minutes of the session. Crypto is 24/7 but spreads widen during low-liquidity hours (typically 02:00-06:00 UTC).
• News. Spreads can multiply 10× in the seconds after a major announcement, then normalise within minutes.

Spread as a real cost

A strategy that trades twice a day on BTC/USDT pays the spread at every entry and every exit. At 0.5 bps spread and 2 round-trips per day, that's 2 bps/day or roughly 7% per year — comparable to a fee schedule, often larger than the broker's stated trading fee.

On illiquid alts the maths gets brutal. A 50 bps spread on a strategy that trades twice a day round-trips 100 bps per day. Any per-trade edge under 1% gets eaten alive.

Trading around the spread

The classic move: use limit orders instead of market orders where the strategy can tolerate the fill uncertainty. A limit order at the bid (for a buy) or ask (for a sell) sits in the book and waits. If it fills, you got the price you wanted and saved the spread. If price runs without you, you missed the trade.

Many exchanges also pay a maker rebate for limit orders that add liquidity (sit in the book and get filled by someone else crossing). On Binance, the maker rebate is roughly 0.075% vs the 0.10% taker fee — switching from market to limit on a high-frequency strategy can flip a marginal-loss strategy into a small-profit one.

Modelling spread in backtest

# Conservative default for liquid majors on crypto perp markets.
execution:
  fill_model: spread_at_close
  spread_bps: 1.0        # 1 bp half-spread (so 2 bps round-trip per leg)
  fees_bps:   5.0        # taker fee, applied per leg → 10 bps round-trip
  # For illiquid alts: spread_bps: 20  or higher

Most strategies that survive these defaults are real. Strategies that depend on filling at the mid or close without spread costs are almost always artifacts of an over-optimistic simulator.

Next steps

Spread, fees and slippage are three pieces of the same execution-cost problem. See the slippage and trading costs post for the maths of how the three combine, and order types for the controls you have over which side of the spread you sit on.