A backtest answers one question: did this logic behave sensibly on bars that already happened? The next question is harder and it only has live answers: does it behave the same way when the bar is still forming, the spread is real and the clock is running? That is forward testing.
What a backtest cannot see
- The forming bar. A backtest judges finished candles. Live, the rule sees a bar still being drawn, and a trigger can read it very differently.
- The real spread at that moment. A simulation charges a stated typical spread. Live, spread widens at rollover, news and the open.
- Timing and data. Feed gaps, late ticks and restart behaviour do not exist in history.
- You. A backtest never tempts you to switch a rule off after two losses.
This is not a criticism of backtesting; it is a statement of its job. A backtest finds broken logic and absurd assumptions. Forward testing finds the difference between the logic and the world.
Three ways to forward test
| Paper lane (Ghost Mode) | Broker demo account | Small live size | |
|---|---|---|---|
| Money at risk | None | None | Real, but small |
| Prices | Live feed, fills simulated at the touch price | Broker's demo feed, which can differ from live | Real fills |
| Execution quirks (slippage, rejections) | Not modelled | Partly | Fully |
| Needs a broker account | No, when run on the public feed | Yes | Yes |
| Emotional realism | Low | Low | Real |
They answer different questions. Paper tests whether the logic behaves live. Small live size tests whether you and the broker do. Doing them in that order means the expensive lesson comes last.
How Ghost Mode works
A rule in Ghost Mode is armed exactly like a live one: same triggers, same conditions, same fire policy, same cooldowns. The only difference is the last step. Instead of sending an order to a broker, it opens a simulated position at the touch price and tracks it to its stop or target. Every fire is recorded with the values that caused it, so you can read the rule's behaviour afterwards rather than remembering it.
Because it is the same rule, you can promote it to live without rebuilding it. That removes a classic source of error: testing one thing and trading another.
What to compare against the backtest
- Fire frequency. If history says "about three a week" and live says "one a month", something about the live conditions differs — usually a spread or session filter.
- What blocked it. The record shows which condition held a rule back. A guard that blocks almost everything live is a sign it is mis-sized.
- The spread at the fire. Compare it to the spread you assumed in the simulation.
- Fire timing. Did it fire on the bar you expected, and once? Duplicates or late fires are logic bugs worth finding while it costs nothing.
- Restart behaviour. Redeploy or restart during the test; a rule should not re-fire as if it were new. Fire state survives a restart for this reason.
How long is long enough?
Longer than you want, and the honest answer depends on how often the rule fires. A rule that triggers a few times a month will not show you much in a week. A week on paper is a smoke test: it catches broken logic, silly spreads and duplicate fires. It is not evidence that a rule has an edge, and a small number of paper trades says almost nothing about future results either way. Decide the number of fires you need to see before you start, and do not move the goalposts after seeing the first few.
Then what
If the paper record matches the history, the next step is the same rule at a size you can shrug off, on one account, with the daily loss brake set. Scale only when the live record has earned it — and treat every stage as a way to be wrong cheaply.
Common questions
What is the difference between a backtest and a forward test?
A backtest replays a rule over historical bars. A forward test runs the rule on live prices as they arrive, without risking money, so you see how it behaves with a forming bar, real spreads and a running clock.
Is paper trading the same as a demo account?
No. A demo account trades against the broker's demo feed and execution. A paper lane simulates fills on a live feed and needs no broker account when run on a public feed. They test different things.
How long should I forward test a strategy?
Until the rule has fired enough times to judge it, which depends on how often it triggers. A week is a smoke test for broken logic, not evidence of an edge.
Can I move a Ghost Mode rule to live?
Yes. It is the same rule with a different last step, so promoting it does not require rebuilding anything.
Run it on paper first
Arm a rule in Ghost Mode on live prices and read what it did before a single order is sent.
Create a free accountRead next
Risk note. This article is educational material about how chartTrigger works. It is not investment advice, not a recommendation to trade any instrument, and nothing here forecasts results. Trading leveraged products carries a high risk of loss. Any historical simulation referred to is exactly that — a run over past bars under stated spread and slippage assumptions, not an indication of future performance.