Backtested annualized returns of 30%—why might live trading fall far short?

A trading strategy has backtested annualized returns of 30%, a solid win rate, and a maximum drawdown within an acceptable range.

At first glance, this seems like a strategy worth using.

But when you actually start trading, you find that returns are far below expectations, or even that you're losing money.

What's going wrong?

Often, it's not that backtesting is worthless, but that we mistake ideal conditions in a model for guaranteed outcomes in real markets.

Backtesting measures how a set of rules performed on historical data under predefined assumptions; it does not promise future returns.

There are five key areas worth checking.

1️⃣ Overfitting: The strategy may only explain the past

If you repeatedly tweak parameters until the historical curve looks good enough, it's easy to mistake random patterns in historical data for a lasting edge.

The more parameters you use and the more times you test them, the more alert you need to be to this risk.

The solution isn't to keep optimizing on the same historical data, but to reserve data that wasn't used for tuning, conduct out-of-sample validation, and check performance across different market phases.

2️⃣ Execution assumptions: A signal appearing doesn't mean an order will fill as expected

A backtest may assume you can buy or sell at a certain price.

But in real markets, prices can move quickly, orders may be partially filled, and limit orders may never fill at all.

If the backtest ignores these scenarios, it may overstate the strategy's real-world performance.

3️⃣ Trading costs: Small differences can add up to significant losses

Fees, slippage, funding rates, and other applicable costs can all eat into a strategy's returns.

This is especially true for high-frequency strategies or those with slim expected returns per trade: changes in costs can directly affect their profitability.

So when evaluating a strategy, don't just look at gross returns—calculate net returns after reasonable trading costs.

4️⃣ Market conditions: Historical advantages may not last

A strategy may perform well in trending markets but get stopped out frequently in choppy markets.

Liquidity may also change, or market participants may behave differently, gradually reducing the trading opportunities that once worked.

This doesn't mean every decline in performance signals that a strategy has stopped working, but it does mean its suitability under different conditions needs to be clearly tested.

5️⃣ Risk metrics: Don't look only at annualized returns

Even if two strategies have similar annualized returns, their maximum drawdowns, return volatility, capital usage, and performance in extreme market conditions may be completely different.

A high win rate doesn't necessarily mean high long-term expected returns, and a smooth backtest curve doesn't mean real-world trading risk is low.

When evaluating a strategy, you should consider returns, risk, costs, and execution conditions together.

So how can you validate a strategy more reasonably?

You can start with four steps:

* Step 1: Define the strategy's rules clearly and avoid constantly changing trading conditions based on results.
* Step 2: Use independent samples or rolling out-of-sample tests to check whether the strategy can adapt to different phases.
* Step 3: Include fees, slippage, funding rates, and reasonable execution assumptions.
* Step 4: Start with simulation or small-scale testing. Record actual fills, costs, drawdowns, and deviations, then decide whether to scale up.

Also note that out-of-sample testing and simulated trading cannot guarantee future profits. Their purpose is to help identify problems and reduce unreasonable assumptions.

What deserves your trust isn't a pretty backtest curve, but a trading process that can withstand independent validation, cost accounting, and ongoing review.

A question for you:

If a strategy backtests well but performs significantly below expectations in live trading, would you first check for overfitting, execution conditions, or trading costs?

In the next article, we'll continue unpacking a problem that's easy to underestimate:

Why can slippage that seems small gradually erase a strategy's profit edge?