Backtesting: what the MarketLens simulation assumes
A backtest applies a stated rule to historical observations. It can help examine whether a rule behaved as expected in one sample, but it is not a replay of the trades a person would necessarily have received and is not evidence that the strategy will work in the future.
Strategies and the simulation
MarketLens currently offers three illustrative rule sets: its trend-and-trigger algorithm with ATR-based exits; a 20/50 SMA crossover; and an RSI mean-reversion rule that buys below 30 and exits above 55. The exact displayed rule names and parameters should be checked in the tool because software settings can change. The simulation is long-only: it does not model short selling.
The tool states that signals form at a bar's close and orders fill at the next bar's open. This avoids assuming that a signal calculated from a closing price could also be filled at that same close. It is still a simplified fill model: it does not simulate intraday price paths, queue priority, market impact, partial fills, exchange outages or changing liquidity.
Costs and position assumptions
The interface lets visitors specify fee per side, spread, slippage and risk per trade. The initial fee, spread and slippage values are example inputs—not claims about a broker or a particular asset. Costs are applied on entries and exits according to the tool's displayed assumptions. A static percentage cannot reproduce every commission schedule, changing spread, borrow charge, funding payment, tax, currency conversion or market-impact cost.
The tool describes its baseline position behavior as fully invested while in a position, with an optional risk-per-trade setting. The separate chart overlay illustrates ATR-based stops. Real position sizing depends on capital, leverage, liquidity, gaps and the trader's own risk constraints. A stop price is not a guaranteed execution price.
What is not modeled
- Taxes, dividends, corporate actions or reinvestment of distributions.
- Short selling, borrow availability or financing costs.
- Intraday execution, market impact beyond the entered cost assumptions, partial fills or order-book depth.
- All possible delisted instruments and point-in-time index constituents. A currently visible symbol list is not a survivorship-bias-free historical universe.
- Changes in a data provider's historical records, missing bars, exchange calendars or every trading halt.
- The practical difficulty of choosing parameters without looking at the test sample.
These omissions can make simulated results materially different from live outcomes. A strategy can appear profitable because of chance, selection bias, overfitting, a particular market regime or unrealistic costs. Repeatedly adjusting rules until a chart looks attractive does not create independent validation.
A disciplined way to inspect a result
- Verify the selected symbol and the tool’s data label. Do not interpret simulated data as historical market evidence.
- Record the tested period, strategy, indicator settings, fees, spread, slippage and risk input.
- Check trade count and individual periods, not only the total return. A few unusually strong trades may dominate the result.
- Consider whether the rule has a plausible explanation that was specified before examining the results.
- Test assumptions that are less favorable, and do not assume a historical return is repeatable.
- Compare the rule to a simple baseline and account for risk and drawdowns, not only gains.
MarketLens is a learning tool, not a brokerage or investment adviser. For broader investor education, consult Investor.gov.