A test deposit is an experiment with a written result, not a small investment. Verify the firm's registration before you send anything, fund close to the minimum, withdraw part of it before you place a single trade, then run four weeks of your normal size and measure the fills. Scale only after the money has come back once.
Most traders choose a broker the way they choose a restaurant — a recommendation, a website, a promotion — and then find out what they actually signed up for eighteen months later, usually on a day when something has gone wrong. A test deposit inverts that. It costs a few hundred dollars of exposure and about a month of patience, and it answers questions no review site can answer for your account, your instruments and your hours.
Verify the registration before you send a dollar
This takes four minutes and it is not optional. An unregistered firm is not a cheaper broker; it is a categorically different kind of risk, because the dispute process, the capital rules and the recovery options all rest on registration existing in the first place.
- US securities brokers — search the firm in FINRA BrokerCheck, which lists registration status, licences held, and disclosure events including customer complaints and regulatory actions.
- Futures and retail forex firms — search NFA BASIC for registration status and regulatory history, and cross-check the CFTC's own registration lookup.
- Check the legal entity, not the brand. The name in the account agreement is frequently a different company, and sometimes a different jurisdiction, from the name on the homepage. It is the entity that holds your money.
While you are there, establish what protection actually applies. US securities customers are covered by SIPC up to $500,000 per customer, which includes a $250,000 limit for cash — and SIPC explicitly does not protect commodity futures contracts unless they are held in a special portfolio margining account (SIPC, What SIPC Protects). That single distinction decides what a firm failure would actually mean for you, and it is unpacked in segregated funds explained.
How small is a small deposit?
Two conditions have to hold at once, and they pull against each other. The deposit must be small enough that losing all of it would not change your week, and large enough to trade your normal position size several times. A test run at a size you would never really trade measures a simulator, not a broker.
When both conditions cannot be met — your real size needs more margin than you are willing to risk on an untested firm — that is itself the finding. The account is too small to produce evidence, and the test waits. It does not shrink. The sizing arithmetic is in position sizing from risk.
Withdraw first. This is the step almost everyone skips.
Before the first trade, request a withdrawal of part of the deposit. It feels pointless and it is the most informative thing you will do all month, because withdrawal friction only reveals itself at the moment you want your money — and by then the balance is no longer trivial.
Write down five things:
- The date requested and the date the money actually landed. Not the date the platform marked it "processed."
- Every fee deducted, including the ones the receiving bank takes.
- Whether new identity verification appeared that was not required to deposit. A firm happy to take money on light checks that demands documents to release it has told you something about its priorities.
- Whether anyone contacted you about it. A message confirming the transfer is normal. A call asking why you are withdrawing, or offering an incentive to leave the money in, is not.
- Whether the route out matched the route in. Most legitimate firms return funds by the method they arrived on, and a firm that will not is worth a direct question.
If that round trip completes cleanly, you have bought a very cheap piece of information. If it does not, you found out for the price of a test deposit rather than the price of an account.
The four weeks: what to measure
Four weeks is the right length because it reliably contains a scheduled news day, at least one quiet range day and a month-end. Those are the conditions that separate a platform that works from a platform that works when nothing is happening.
| What to test | How to record it | What a pass looks like |
|---|---|---|
| Fill quality | Intended price vs filled price, every trade | Slippage clusters near zero outside news |
| Behaviour under news | The same measurement, on scheduled releases only | Wider, but the platform stays responsive |
| Order-type support | Place a bracket, a stop-limit, a trailing stop | All work as documented, including on mobile |
| Statement accuracy | Reconcile your journal against the statement | Every fee and fill matches, to the cent |
| Support response | Ask one real question in week one | A specific answer from a human, not a macro |
| Platform stability | Note every disconnect and its duration | None during the open; none unexplained |
The statement reconciliation is the item people drop, and it is the one that catches quiet problems. Fees that do not match the published schedule, fills recorded at prices you did not see, a financing charge that appears without explanation — all of it shows up in a line-by-line comparison and nowhere else. Keeping a real trading journal is what makes the comparison possible at all.
Reading the fills honestly
Slippage on its own is not a verdict. A market order into a thin book at the opening bell will slip at any broker on earth, and blaming the firm for that is like blaming a taxi for traffic. What you are looking for is asymmetry and pattern.
- Asymmetry: does slippage run against you far more often than it runs for you? Random execution noise should land on both sides.
- Pattern: does it worsen specifically when you are in profit, or specifically as a stop is approached?
- Scale: total the cost per round turn and set it against your average winner, using the method in commission vs spread and the measurement column described in slippage in trading.
One month of data will not settle a marginal case. It will comfortably settle an obvious one, and obvious cases are more common than most traders expect.
What should end the test immediately
Some findings are not data points to weigh against others. They end the experiment on the day they appear.
- Funds not returned on a valid request, or a withdrawal repeatedly "pending" with no documented reason.
- Pressure to deposit more, from anyone, in any framing.
- A bonus with trading-volume conditions attached to withdrawing your own money. Your deposit should never be hostage to a turnover target.
- An assigned "account manager" who gives trade ideas. A regulated broker executes orders. A firm whose staff tell you what to buy has a different business model, and it is one where your loss can be their gain.
- Terms that changed after you funded — particularly spreads, leverage or withdrawal rules.
- No verifiable registration for the entity that actually holds the money.
The same instincts apply to any firm asking for money on the strength of its own marketing — the vetting logic in how to find a legit trading community transfers almost line for line.
Passing the test is not the same as being right for you
A broker can clear every check above and still be the wrong choice, because fit depends on your instrument, your hours and your size. A firm with excellent equities execution may be mediocre in futures. A platform that is superb on desktop may lack the order types you need on a phone. Decide what you are optimising for first — the framework is in how to choose a broker for day trading — and treat the test as confirmation of a decision you have already reasoned through, not as the decision itself.
Frequently Asked Questions
How much should I deposit to test a new broker?
Enough to trade your normal position size at least a few times, and little enough that losing all of it would not change your week. Those two conditions pull in opposite directions, and the second one wins. If your real size cannot be traded at the broker's minimum, that is itself a finding: the account is too small to tell you anything, and the test has to wait until it is not.
Should I withdraw money before I start trading?
Yes, and it is the single most useful step in the whole process. Withdrawal friction only reveals itself when you want your money, which is the worst possible moment to discover it. Requesting a partial withdrawal within the first few days costs you nothing but time and tells you the processing window, the fees, whether extra identity checks appear, and whether anyone tries to talk you out of it.
How do I check whether a broker is actually registered?
For US securities brokers, search the firm in FINRA BrokerCheck, which lists registration status, licences and disclosure events. For futures and retail forex firms, search NFA BASIC, which covers registration and regulatory actions. Verify the exact legal entity name on the account agreement rather than the brand name on the website, because the two are often not the same company.
How long should a broker test run?
About four weeks of your normal trading. That is long enough to include a news day, a quiet range day and at least one month-end, which are the conditions that separate a platform that works from a platform that works when nothing is happening. Anything shorter tells you about the software; four weeks tells you about the firm.
Bottom line
Testing a broker properly costs a small deposit, four minutes of registration checks and one month of ordinary trading — and it replaces a guess with evidence at the only point in the relationship where being wrong is still cheap. Check the registration, withdraw before you trade, measure the fills against your own journal, and write your walk-away conditions down before you have a reason to argue with them. Then read how we answer the same kind of questions about ourselves, because a firm that will not be tested on its own terms is telling you something either way.