Demo first: why forward testing beats backtests before real money

Why backtests flatter every strategy

A backtest replays history against your rules and reports what would have happened. It is useful — and systematically optimistic. Historical fills are assumed perfect; real spreads widen at news events; slippage appears exactly when volatility makes it costly. None of that is in the replay.

The deeper problem is overfitting: tune parameters against the same historical data often enough and you will always find a combination that 'worked' — because you fitted the noise, not a real edge. A backtest is evidence that a strategy is not obviously broken. It is not evidence that it works.

What a demo forward test actually reveals

Running live-market decisions on a demo account, in real time, with the same software that would trade live, answers questions no backtest can:

• Does the system behave sensibly in conditions nobody selected — the dull weeks, the news spikes, the sessions you would not have chosen for a marketing screenshot?
• Do the operational pieces hold: data feed, broker connection, order placement, stop and target handling, restarts?
• How often does it decline to trade — and are you comfortable with that rhythm? (A system that trades constantly is usually a red flag, not a feature.)
• Can you watch a losing streak without wanting to intervene? Demo is where you learn what the drawdowns feel like while they are still free.

What demo cannot prove — being honest about the gap

Demo fills are friendlier than live fills: no meaningful slippage, no partial fills, no requotes at the worst moment. And demo money removes the psychological weight — watching a real 3% drawdown is a different experience from watching a simulated one.

So a good forward test is necessary, not sufficient. Its real function is filtering: most flawed systems reveal themselves on demo within weeks — errors, nonsense decisions, death-by-costs. Whatever survives has earned a cautious live trial, not your confidence.

When a live transition is actually justified

There is no magic duration, but there is a reasonable checklist. Going live starts being defensible when all of these hold:

• The demo run covers at least a few weeks of varied conditions — including news events and dead periods — not a hand-picked good stretch.
• You have seen it lose, repeatedly, and the losses matched the risk rules: stops respected, sizing consistent, no runaway behavior.
• You understand every decision it made, or at least can review the reasoning for each one.
• The live step is deliberate: smallest viable size, same rules, and a pre-committed drawdown level at which you stop and return to demo.

Treat 'demo-first' as a filter for tools, too

The same logic that applies to your strategy applies to any trading software you evaluate. A tool built for demo-first operation — where the default path is signal review or demo execution, and live trading requires an explicit, deliberate opt-in — is structurally aligned with your survival.

A tool that pushes you toward funding a live account quickly is telling you where its incentives sit. The pressure to skip the demo phase is itself the strongest reason not to.