Demo vs Live
EA Results
A demo account has no counterparty, so it fills at the quoted price and often charges nothing. Every cost it waives is charged on a funded account.
Updated:
Key takeaways
- Demo fills at the quoted price. Live fills at the price still available when the order lands.
- Spread, commission and swap are charged on live accounts and often waived on demo ones.
- A strategy earning 2% a month can lose its whole edge to two pips of cost.
- Ask for a live verified account with trade history, not a demo screenshot.
- Run a small live test yourself. One month of real fills beats any published curve.
What Changes Between
Demo and Live?
A demo simulates the market. A funded account participates in it.
A demo account simulates the market. A live account participates in it. Everything expensive lives in that difference.
Demo proves one thing well: the code runs without errors. That is a useful result, and a small one.
On demo
- Orders fill at the quoted price, because no counterparty is needed.
- Spread is often fixed and narrower than the live account shows.
- Commission and swap are frequently switched off entirely.
- Requotes and rejections almost never happen.
On a funded account
- Orders fill at the price still available when they arrive.
- Spread widens around news, rollover and the session handover.
- Commission is charged per lot, and swap accrues every night.
- Large orders move through several price levels before completing.
None of these is a broker trick. A demo server has no order book to consult, so it quotes and fills.
The mechanics of a real fill are traced in anatomy of an order.
Scalping robots suffer most. The tighter the target, the larger the share that execution cost takes.
How Much Does
Cost Eat?
Four charges per trade, multiplied by how often the robot trades.
Cost is charged per trade, so a frequent strategy pays it frequently. The rate matters less than the count.
Four charges
- Spread: paid on entry and again on exit, every single trade.
- Commission: charged per lot on raw-spread accounts, both ways.
- Swap: charged nightly on any position still open at rollover.
- Slippage: the gap between your requested price and your fill.
What that adds up to
- A robot taking 400 trades a year pays the spread 800 times.
- At 1 pip per side that is 800 pips of cost before any profit.
- On a 0.10 lot EUR/USD position, one pip is roughly one dollar.
- So the same year carries about $800 of execution cost per 0.10 lot.
Now compare that figure with the gain the robot claims. A strategy earning 2% a month has a thin margin.
Two pips of cost per side can remove the whole edge. The strategy did not fail; the arithmetic never worked.
Cost per trade on your own broker is worth measuring first. The overnight fees guide covers the swap side.
Run pip value for your account currency in the pip value calculator.
What Counts as a
Verified Result?
Five marks of a real record, and a one-month test you can run yourself.
A screenshot proves nothing. It shows a picture of numbers, produced by whoever wanted to sell you something.
A verified record is different. An independent service reads the account and publishes what it finds.
Even then, read the label. Tracking services display demo and contest accounts alongside real ones.
Five marks of a real record
- The account type is stated as real, not demo or contest.
- The broker is named, and the account number is visible.
- Trade history is downloadable, not just a summary image.
- The tracking service confirms the connection independently.
- The record covers at least six months of continuous trading.
When a seller will not show all five, the missing one is usually the answer. Ask which is absent and why.
Our own audit framework applies the same list. It is set out in analyzing EA risks.
Run your own month
- Open the smallest live account your broker allows.
- Run the robot at the smallest lot size for one full month.
- Record every fill price against the price you expected.
- Compare your month with the same month on the vendor record.
One month of your own fills outranks any published curve. It is measured on your broker, at your latency.
This is the method behind every test in our robot rankings. We funded the accounts ourselves.
The distinction between a rehearsal and a performance is covered in backtest vs forward test.
Frequently Asked Questions
Why do forex robots perform better on demo?
Demo orders fill at the quoted price, and spread, commission and swap are often switched off. Live accounts charge all of them.
Are demo results completely useless?
No. They prove the code runs without errors and show the intended logic. They say nothing about profitability.
How can I tell a demo record from a real one?
Check the account type label on the tracking service. Ask for the broker name, the account number and a downloadable trade history.
How much does execution cost actually matter?
A robot taking 400 trades a year pays the spread 800 times. At one pip per side that is 800 pips before any profit.
How long should my own live test run?
One month at the smallest lot size is enough to compare your fills against the vendor record. Six months is enough to judge the strategy.
Continue Your Validation Cluster
Backtest vs Forward Test
The validation stack: what a rehearsal can prove, and what only live money can.
Slippage and Latency
Where execution friction comes from, and how to measure your own fill quality.
VPS for Forex Robots
Latency to the broker server, uptime, and what actually changes your fills.
What Is a Forex EA?
The fundamentals, and what seven funded tests returned on real accounts.
Robot Rankings
Every test on this site was funded with our own money, not run on demo.