Every broker comparison is written from documentation — spreads from a marketing page, execution from a policy statement, support from a chat window before an account exists. None of it is wrong, and none of it tells you how the firm behaves toward a funded client. That requires funding, and the cheapest version of it is a deliberate test rather than a full commitment.
What a two-week test establishes
- The spread you actually pay, at the hours you actually trade.
- How orders fill: slippage, rejection, and behaviour around news.
- Whether support answers a specific question with a specific answer.
- How long a withdrawal takes from request to arrival.
Fund the minimum that lets you trade properly
The temptation is to deposit the account minimum, but a balance too small to place a normally sized position tests nothing useful. Fund enough that you can trade your usual instrument at your smallest normal size with a normal stop, and no more. For most methods that is a few hundred units of the account currency.
Fund from the account you would actually use long term. Anti-money-laundering rules return funds to their source, so the test of the withdrawal route only means something if it is the route you intend to keep.
Week one: measure the cost
Record the spread on your main instrument at fixed intervals across a full session, then again at the hour you normally trade, then once during a scheduled announcement. Three sets of numbers, taken by hand, are worth more than a month of reading about average spreads.
Place trades at your normal size and log the difference between the price on screen when you clicked and the price you were filled at. Do this for at least twenty entries. The distribution matters more than any single fill: consistent small negative slippage is normal; occasional large negative slippage with no positive slippage at all is a pattern worth noting.
What to log for every trade
- Instrument, time, and direction.
- Displayed price at the moment of the order.
- Actual fill price and the difference in points.
- Whether the order was requoted or rejected, and how many times.
- The spread at the moment of entry and at the moment of exit.
- Any commission and any financing charged.
Week one, in parallel: test support
Ask one question with a specific, verifiable answer — the exact conversion markup applied to settlements, the day the triple swap falls on a named instrument, or the tiered margin thresholds on the instrument you trade. These have factual answers that exist in the broker's own documentation.
What you are measuring is not friendliness. It is whether an agent produces the number, or produces a paragraph that avoids it. A firm whose front line cannot state its own charges is telling you how a later dispute will go.
Ask the same question a second time, days later, through a different channel. Inconsistent answers to a factual question are more informative than a slow one.
Week two: test the exit
This is the part of the test that matters most and the part most people skip, because the account is working and there is no reason to interrupt it. Request a withdrawal of a portion of the balance and time it from request to arrival.
Note whether documents are requested at this point rather than at signup, how quickly they are reviewed, whether the processing time matches what the broker published, and whether anyone contacts you to discuss the request. That last one is the single clearest signal in the entire exercise: a withdrawal should be processed and dated, not discussed.
Any pressure to keep the funds ends the test
An account manager who asks why you are withdrawing, suggests waiting, or mentions an opportunity you would miss has answered the only question that mattered. Withdraw the remainder and close the account. This behaviour does not improve as the balance grows.
What to do with the result
Compare the measured cost against what the broker advertised, and against the same measurements at your current broker if you have one. A difference of a fraction of a pip is within noise and should not drive a decision. A consistent gap between advertised and actual spreads, or a fill pattern skewed one way, is a finding.
Weigh the withdrawal result heavily. A broker that is marginally more expensive but pays reliably is a better counterparty than one that is marginally cheaper and takes two weeks with three document requests. Cost differences are measured in fractions of a pip; access to your own money is not measured in anything comparable.
Testing two brokers at once
Running the same test at two firms simultaneously removes the largest source of error, which is that market conditions change between tests. The same instrument, the same hours, the same order sizes, on the same days, produce a comparison rather than two measurements.
The cost is two minimum deposits and two withdrawal fees, which is trivial against the cost of choosing wrongly and discovering it with a full balance. Remember to close the account you reject rather than leaving it dormant, since inactivity charges will otherwise consume the residue.
Common questions
How much should I deposit for a test?
Enough to trade your usual instrument at your smallest normal size with a normal stop — typically a few hundred units of the account currency. A balance too small to trade properly tests nothing.
Is testing a broker with a demo enough?
No. Demo servers fill instantly at the quoted price and do not model slippage or rejection. They test the platform and the instrument list, which is useful, but not how the firm handles a funded client.
Why does the withdrawal matter more than the spread?
Because cost differences between reasonable brokers are fractions of a pip, while access to your money is binary. A firm that pays reliably at a slightly higher cost is the better counterparty.
Should I close the account if I decide against the broker?
Yes. A dormant account with a small balance is subject to inactivity charges that recur until the balance is exhausted, and the funding card or wallet needs to stay open until the money is out.
