Transaction costs are not a footnote. They are often the difference between a promising backtest and a strategy that cannot survive live trading. The more frequently a strategy trades, the more important the cost model becomes.
Costs include explicit commissions and fees, but those are only the visible part. Bid-ask spreads, market impact, borrow costs, funding rates, taxes, and delayed execution can matter more than the brokerage commission. In less liquid markets, the quoted price may not be the price a real order receives.
Slippage measures the difference between the expected execution price and the actual execution price. A backtest that assumes entries at the close may be unrealistic if the signal is only known after the close. A strategy that trades at the next open may suffer gap risk. A strategy that trades large size may move the market.
Turnover is a useful warning sign. High turnover strategies need stronger gross edges because they repeatedly pay costs. A low turnover strategy can sometimes tolerate wider spreads because it trades less often.
A sensible backtest should test cost sensitivity. If a strategy is profitable only under very low cost assumptions, it may be too fragile. Researchers can run scenarios with higher spreads, delayed fills, or reduced liquidity to understand the margin of safety.
Execution is part of strategy design. A signal with excellent theoretical performance may be unusable if it requires impossible fills. A more modest signal with stable execution can be more valuable.
Costs make research less glamorous, but more real. Ignoring them is one of the fastest ways to confuse a simulated edge with a tradable edge.