Backtest — What Is Investment Strategy Backtesting?

Quick Answer

A backtest is testing an investment strategy on historical data — applying precise entry, exit, and position-sizing rules to past prices to estimate how the strategy would have performed: its returns, losses, and drawdown. You define the rules, choose a long enough data window (ideally 20–30 years), run the simulation, and evaluate metrics like CAGR, max drawdown, and the Sharpe ratio. It matters as a sanity check, but watch for pitfalls like overfitting and survivorship bias that inflate results. This is educational information, not investment advice.

---

What is a backtest?

Backtest (backtesting) is testing an investment strategy on historical data. You check how your strategy would have performed in the past — what returns it would give, what losses, what drawdown.

Example: "If I bought S&P 500 every month since 2010, how much would I have today?"

How to conduct a backtest?

Step 1: Define strategy

Precise entry and exit rules:

• When do you buy? (e.g., 1st day of month)
• How much do you buy? (e.g., $1,000)
• When do you sell? (e.g., never / after reaching target)

Step 2: Choose historical data

• Period — longer is better. Minimum 10 years, ideally 20–30.
• Source — Yahoo Finance, Portfolio Visualizer (automatic backtest).

Step 3: Run simulation

Tools:

• Portfolio Visualizer (portfoliovisualizer.com) — free, intuitive.
• Backtest.curvo.eu — European ETFs, includes TER.
• Python + pandas — for advanced users, full control.

Step 4: Evaluate results

Key metrics:

• CAGR — compound annual growth rate
• Max drawdown — largest peak-to-trough decline
• Sharpe ratio — risk-adjusted return
• Win rate — % of profitable trades

Backtesting pitfalls

Overfitting (curve fitting)

Biggest sin: you optimize the strategy until it perfectly fits history. Won't work in the future because you "learned" noise, not signal.

Survivorship bias

Testing on companies that survived. Doesn't include bankruptcies — which inflates results.

Look-ahead bias

Using information not available at decision time (e.g., annual company results in January when report comes out in March).

Ignoring transaction costs

Backtest without commissions, spreads, and taxes gives unrealistic results.

When does backtest make sense?

• Comparing simple strategies — DCA vs lump sum, 60/40 vs 80/20.
• Checking drawdowns — could you withstand a 40% drop?
• Testing allocation — which asset mix gave best risk/reward?

When does backtest mislead?

• For strategies based on short-term signals (market timing).
• When optimizing dozens of parameters.
• When testing on too short period (<5 years).

How Freenance can help?

Freenance lets you track your portfolio's real history — it's your "live test." Compare your results with benchmarks and check if your strategy works in practice, not just on paper.

👉 Track real portfolio performance — freenance.io

Related Articles

• Sharpe Ratio — co to jest współczynnik Sharpe'a?

FAQ

What does it mean to backtest an investment strategy?

Backtesting means applying a defined set of rules (entries, exits, position sizing) to historical price data to estimate how the strategy would have behaved in the past. It produces a return path, drawdowns, and risk metrics like CAGR and Sharpe ratio. The result is an estimate, not a guarantee — markets evolve and historical regimes do not always repeat.

What is overfitting in backtesting?

Overfitting (curve fitting) is tuning strategy parameters so tightly to past data that the strategy captures historical noise rather than a stable pattern. Symptoms include suspiciously high returns, many tweaked parameters, and weak performance on out-of-sample periods. The cleanest defense is reserving an unseen test window and keeping the rule set as simple as possible.

How many years of historical data are enough?

A robust backtest usually needs at least 10 years and ideally 20–30, so the data covers different market regimes — bull, bear, sideways, inflationary, and deflationary. Shorter periods can flatter strategies that happened to fit one regime, like the 2009–2021 stock bull run. For monthly DCA-style strategies even longer windows are preferable.

What is survivorship bias and why does it inflate results?

Survivorship bias is testing only on assets that still exist today, ignoring delisted, bankrupt, or merged ones. The backtested universe becomes a "winners-only" sample and average returns look unrealistically high. Using databases that explicitly include defunct tickers, or sticking to broad index data, materially reduces this distortion.

Should I trust a backtest before investing real money?

A backtest is a useful sanity check, not an investment recommendation. Treat it as one input among many: it should be combined with realistic costs, taxes, slippage, and an honest assessment of your behavior in drawdowns. Even a clean backtest cannot replace risk tolerance review, diversification, and the understanding that past performance does not predict future results.