Assumptions and Approximation Risks

Algorithmic trading backtests often rely on simplifying assumptions to make simulations tractable. Common approximations include:

• Zero transaction fees
• Perfect execution at next-day open price
• No slippage
• No market impact
• Ignoring dividends
• No capital constraints

While these assumptions may be reasonable in some cases, they can be dangerously misleading — especially for strategies with small expected returns per trade.

🔍 Why This Matters

If your per-trade expected return (edge) is small, even minor sources of error or friction can completely erase profitability.

For example:

• A 0.1% edge per trade can be wiped out by a 0.05% slippage and 0.05% commission.
• If dividends are excluded, total return vs. price return divergence can mislead long-term performance.

In short: the smaller the edge, the more dangerous approximation errors become.

🛡️ Safer Approach: Design for Robustness

To mitigate these risks:

• Design strategies with large per-trade edge where possible.
• Run sensitivity analysis: How does performance change with ±0.1% slippage or added transaction cost?
• Model execution explicitly, including realistic fill probabilities and price slippage.
• Include transaction fees and taxes, even if only as estimates.

A robust strategy should remain profitable even when assumptions are relaxed.