13 backtesting myths, and what’s actually true.

It can still be luck, over-tweaking, or one kind of market.

Length helps only when the test is honest. The strongest evidence is how the strategy does on data it was never tuned on.

Adding conditions until the past looks good is exactly how curve fitting happens.

Each filter that removes a few past losers is another chance to fit noise. Fewer, simpler rules are usually more trustworthy.

Every tweak fitted to the past makes the backtest less trustworthy, not more.

Decide what you’ll test before you see results, and stop when a change has no reason behind it other than a better backtest.

The more versions you try, the more likely one looks great by pure chance.

Get enough people to flip a coin ten times and someone will get eight heads. The bar for believing a result has to rise with every try.

Look at it, change the strategy, look again, and it isn’t out-of-sample anymore.

A sealed test only gives an honest answer if it’s opened once, on the finished strategy.

Those are rules of thumb, not proof.

How many trades you need depends on how big the edge is compared with the noise, how many versions you tried, and whether the trades span different market conditions.

Only if the test itself is honest.

Bad data, look-ahead mistakes or unrealistic fills make a ten-year test just as misleading as a short one.

Not without realistic costs, slippage and fills.

Some tools assume no costs and perfect fills by default. Check what yours assumes before you trust the result.

A high win rate can still lose money.

What a win earns compared with what a loss costs matters just as much. Look at expectancy, not just how often it wins.

Costs, fills and execution always differ.

The question isn’t whether live trading matches the backtest exactly. It’s whether the edge survives the difference.

Its biggest value is showing what doesn’t work, before real money finds out.

A backtest can’t promise the future, but it’s the cheapest way to rule out a strategy that never had an edge.

A challenge tests one path through the firm’s rules.

Luck can pass a challenge, and a good strategy can fail one on a single bad day against the daily loss limit.

Only if you can see the evidence.

A tool that shows no trades, no costs and no checks gives you nothing to verify. Ask to see them.