Our Methodology

Each broker’s overall score is arithmetic you can reproduce. Take the published sub-scores, apply the fixed criterion weights and round half up, and you should land on the same overall number we publish. The sub-scores themselves are editorial judgements, and I set out their evidence base below.

We cover 33 brokers, 28 of them tested on a live account. Every one of those 28 was opened at the broker’s Australian ASIC entity and funded with our own money, in Australian dollars at every broker except eToro, whose account is in US dollars. Our scoring model holds eight weighted criteria, from trading costs to funding, and each broker’s overall score is rescaled over the criteria that broker carries. The whole programme puts real capital at risk so the results are not a paraphrase of a sales deck. That is the only way to get trustworthy data, and the measured figures you see on the reviews of the brokers we tested live are built from that real-money capture. Reviewed by David Levy, Head of Content.

Summary

One weighting model applies to every broker, spanning costs, trust, platforms, experience, service, market range, education and funding, though a broker missing a criterion has that weight rescaled across the criteria it does hold. The numbers run through a fixed weighting model and the rating is published with a date stamp. For the 28 we tested live, those data come from a live account at the broker’s Australian ASIC-regulated entity, funded with our own money. The rest have no funded account of ours and are scored on the same criteria without one. Each finished rating then feeds our ranked guide to the best forex brokers in Australia, so the broker at the top of that list earned its place through the same unsparing process as the one at the bottom. That consistency is what makes the guide credible, and it means you can compare any two brokers and know the same yardstick was used.

The weights were last rebalanced in 2018 and have not changed since. We never change them mid-year, because doing so would let us tilt the rankings. From 1 October 2026, every overall score on the site is calculated directly from the published weights and the sub-scores shown on each review. The recalculation moved some scores and ranks with the weights unchanged. Totality received sub-scores for the first time: five categories scored from a desk assessment of its own published records, each with a written basis, and three categories left unscored. I like that you can now take any overall score on the site, check it against the weights table and the sub-scores on the review, and get the same number we publish.

The verification log

Testing is a team effort. Noam Korbl (RG146) runs our live spread captures, David Levy leads the editorial team’s source checking, and Justin Grossbard leads the research desk. Since July 2026, every review page publishes a dated verification record from the current cycle: masked funded-account references, archived spread captures, ASIC register checks and PDS versions. The record lists the latest artefacts behind that review, showing when a number was last confirmed rather than relying on a publish date alone. That verification record is the most important part of the page, because it shows you exactly when each number was last confirmed.

The 8 weighted criteria

Each broker is scored 0 to 10 against eight criteria. The overall score multiplies each sub-score by its criterion weight, sums the results, divides by the total weight of the criteria the broker carries, and multiplies by 10, giving a score out of 100 that we publish rounded half-up. Where a broker has no score for a criterion, most commonly Education or Funding, the weights are rescaled over the criteria it does carry. Brokers are ranked on the unrounded score, and the current rank breaks only an exact tie. Every sub-score is recorded in our scoring files as measured (our own capture in a named window), documented (a dated primary record) or desk-assessed (an editorial judgement).

A measured input is a dated capture record: spread, commission and similar figures our team recorded on a stated date. It is distinct from the editorial sub-score, which weighs those inputs alongside other evidence. Three classes underpin our category numbers: measured inputs, documented inputs from dated primary records, and desk assessment where no adequate measured record exists. A complete numeric rubric mapping inputs to every category score has not been adopted. Current Trading Experience and Customer Service ratings are desk-assessed.

Why these weights

Trading costs lead because they are the largest controllable variable in a retail trader’s P&L, and in many P&Ls I see the broker took more in spread than the strategy returned in edge. That is why cost gets the heaviest weight on the page, so you can see which broker leaves more of your edge intact. Over a year of regular trading, the difference between a 1.0 pip and a 0.2 pip total cost on EUR/USD is the difference between profit and loss for many strategies.

Trust and regulation come second because the largest unrecoverable loss is broker insolvency, not trading loss; an edge does not matter if the broker is not there to pay out. Twenty per cent of the score going to AFSL standing, AFCA membership, parent backing and enforcement history is deliberately heavy, so you can see that risk before you fund an account.

Trading experience and platforms each get 15% because they capture the day-to-day reality of trading: a tight spread on a platform that freezes during news is not a bargain, it is a liability. Slow execution and unreliable platforms cost real money on slippage and missed entries, and that is one of the most underrated risks in retail forex. It is the reason you see platform stability weighted alongside spread.

Customer service and the smaller-weight criteria (markets, education, funding) round out the score. They matter, but not as much as the top four, until you need to withdraw funds quickly or a stop fails to fire. In those moments, those criteria stop feeling small and become the difference between a broker who answers the phone and one who hides behind a ticket queue. The weight on this column reflects the cost to you when that happens.

Live-account testing process

Each of the 28 brokers we have tested live, out of the 33 we rank, has a live account at the broker’s Australian ASIC-licensed entity, opened by a member of our research team in their own name through the broker’s own online application, the same one a reader would use. The application frames show the broker’s identity checks, a driver’s licence or passport as each form asked, and, where the broker asks them, its appropriateness questions. Each account is funded with our own money, in Australian dollars at every broker except eToro, whose account is in US dollars. No account is on a pricing tier a reader could not open, because that is the only way I would trust a test.

Our spread figures come from two captures: May to June 2026 on IceFX SpreadMonitor on MT4, and September 2026 on our own logger on MT4 and MT5 over 48 hours from 22 to 24 September 2026. Each published spread is dated to its capture, so you can see how fresh the number you are reading is. The logger reads only MT4 and MT5 accounts. Each broker’s application, funding and platform screens are kept in an evidence pack, and each review carries a dated verification record you can check.

Each live account is used for:

• Spread capture across eight pairs on the MT4 and MT5 accounts, most recently over 48 hours from 22 to 24 September 2026 (see the spread-testing section below)
• A small live trade, 0.01 lot, where a commission needed measuring, with the charge read from the platform
• The deposit, recorded with its method, its amount and the time it took to credit

Spread testing with our own spread logger

Spreads are the headline cost on most retail forex trades and the most comparable metric across brokers, but a broker that quotes 0.0 on EURUSD at 3 AM Sydney time is not the same as one that holds 0.1 through London open. Our capture window is built to expose exactly that difference, and that is what I care about.

Tool

Since our September 2026 capture we measure spreads with our own logger, CFB-SpreadLogger 1.1.0, running on MT4 and MT5. It reads each broker’s live bid and ask once per second and keeps one record per minute. From those records we calculate the average and p95 spread for the testing window, so you can see both the cost you will typically pay and how wide your worst minutes run. What appears on our spread table is the average of that capture, not a snapshot from a quiet Sunday afternoon, so the number you see reflects the spread paid during real trading hours. Before this capture, from May to June 2026, we used IceFX SpreadMonitor on MT4, and those tick-median figures stay dated to their own window.

The capture runs from a London VPS connected to each broker’s own MT4 or MT5 server. Most ASIC brokers route execution through LD4 in London or NY4 in New York, so the VPS sits where the servers are; spread readings are venue-independent tick data, so the location does not tilt the results.

Symbols

Eight pairs: EURUSD, AUDUSD, GBPUSD, USDJPY, EURGBP, EURJPY, AUDJPY, USDCHF. The average across these eight is the ranking key on our lowest spread forex brokers table, so the same eight are captured for every broker. Australian dollar pairs matter most to our readers, which is why two of the eight carry AUD. Some global brokers price AUDUSD competitively for their AU clients. Others widen it because it’s not their core flow. The test catches that, and it is often the pair that separates the genuinely competitive from the marketing claim. This pair is the litmus test I use for a broker’s commitment to the Australian market.

Window

48 hours starting at 2pm Brisbane time (AEST), in two 24-hour cycles split at the same hour. That captures:

• The end of the New York session (low-liquidity drift)
• The Sydney open and Sydney session (the AU-relevant window)
• The Tokyo session (AUD/JPY-relevant)
• The London open and full London session (peak liquidity)
• The London/New York overlap (peak volume)

A spread quoted at a single hour is a guess. By covering all four major sessions we get a representative spread profile rather than a snapshot from one quiet hour, so the number reflects what a trader would actually pay across a trading day, and that is the cost you can expect.

Account types tested

For brokers offering multiple account types, we test both:

• Standard / commission-free account (e.g. Pepperstone Standard, IC Markets Standard, FP Markets Standard)
• RAW / ECN account (e.g. Pepperstone Razor, IC Markets Raw Spread, FP Markets Raw, Fusion Markets Zero)

The score combines spread and commission to produce a ‘true cost’ metric. At the pip values read from our September 2026 capture, an A$7.00 round turn per standard lot converts to 0.50 pips on EUR/USD, so a 0.1 pip spread plus that commission comes to 0.60 pips all-in. A 1.0 pip standard-account spread is, simply, 1.0 pip. RAW typically wins for active traders, while Standard typically wins for occasional traders. For a trader doing two lots a month, the maths points one way; for fifty lots, it points the other, and the true-cost number shows which side of the line applies. That number is the quickest way to compare brokers, and it lets you see at a glance which account type costs less.

How we rate trading experience

Our current Trading Experience category is desk-assessed: our research desk judges each broker’s platform and trading conditions, and no category test logs underpin that rating. A reproducible execution benchmark would need dated order and fill records for each broker, noting account type, the Australian legal entity, server and location, plus a written mapping from those inputs to a score. Connection latency and order execution measure different things and would be reported separately. Until such records and a mapping exist, the category stays desk-assessed.

Trust and regulation scoring

This criterion captures the hardest part of broker selection: what is invisible in normal operation. The score breakdown: A spread can be tested in five minutes. You cannot test what happens to your money if the broker fails, which is why this column carries the second-heaviest weight on the table and why I insist we check the AFSL register ourselves rather than taking the broker’s word.

• AFSL status (40%): Current AFSL with ASIC, no current suspension, conditions or enforceable undertakings. Licence status, suspension and conditions are verified against ASIC’s Professional Registers Search at every test cycle. Enforceable undertakings sit on a separate ASIC register, which you can search as well.
• Years in operation under the current AFSL (20%): Brokers with 10+ years on the same AFSL score full marks. New entrants score lower until they build a track record.
• AFCA membership (10%): Confirmed current member of the Australian Financial Complaints Authority.
• Parent backing (15%): Listed parent (ASX, LSE, NASDAQ) or large privately held parent with audited financials scores higher than thinly-capitalised standalone entity.
• Enforcement history (15%): ASIC enforcement actions, surveillance warnings, court enforceable undertakings or fines in the last 36 months reduce the score.

Brokers with current ASIC enforcement actions are flagged at the top of the review intro, because I believe you should see them before comparing costs, and may be removed from the comparison shortlist depending on severity.

We scale that reduction to how settled the action is. A proven finding or a sustained enforcement outcome counts in full. A live, contested proceeding counts for less, because no breach has been established and the broker is entitled to defend it. An interim order that ASIC revoked after the broker remediated is disclosed on the review but does not move the score. We only score actions documented against a named licensee: when ASIC reports a sector-wide breach without naming the firms involved, as it did for the CFD product intervention order in its January 2026 review, the absence of a named action against a broker is not treated as evidence of clean conduct (in my view, a sector-wide finding is a cloud, not a verdict, and we do not pretend otherwise).

Platform scoring

15% of the score, broken down across each platform a broker offers:

• MT4 (Pepperstone, IC Markets, FP Markets, Eightcap, Fusion Markets, Blueberry, ThinkMarkets, Axi, Vantage, Global Prime, ACY, GO Markets, FXCM, MultiBank, AvaTrade)
• MT5 (most of the above plus a handful of MT5-only brokers)
• cTrader (Pepperstone, IC Markets, FP Markets, Fusion Markets, GO Markets)
• TradingView native integration (Pepperstone, OANDA, FP Markets, Eightcap, Vantage, Capital.com, Global Prime)
• Proprietary platforms (CMC Next Generation, IG web platform, Plus500 WebTrader, eToro app, Capital.com platform, easyMarkets, Trade Nation, Mitrade)

Platform assessments consider charting tools, order types, automation support, mobile parity, stability and availability. There is no adopted published numeric mapping from those features to the current sub-score, so treat it as our editorial judgement.

Customer service scoring

Our current customer service ratings are desk assessments rather than timed benchmarks, and no category test logs underpin them. A future test protocol would record each support question with its date, the channel used, when the response arrived, whether the answer was correct and which legal entity responded. Repeated observations across brokers would be needed before response times could be compared as benchmarks. That protocol is prospective work, not something already completed, so the ratings you see today rest on our desk assessment of each broker’s support arrangements.

Range of markets

A flat count is misleading, because in my view the number on the table rewards useful breadth, not padding. A broker offering 10,000 micro-cap US stock CFDs is not “more diverse” than a broker offering 80 forex pairs and 200 well-chosen index/commodity CFDs.

We score the breadth across categories:

• Forex pairs (majors, minors, exotics)
• Indices (major and minor)
• Commodities (energy, metals, softs)
• Share CFDs (ASX, US, UK, EU, Asia)
• ETFs
• Treasuries / bonds
• Cryptocurrency CFDs

Brokers covering all seven categories with reasonable depth in each score full marks. Specialists score lower on this criterion but may make up the points on platform or trading costs, and in my view, a specialist can still be the right answer for a trader who focuses on a single asset class.

Education and funding

Smaller-weight criteria but they matter for specific reader segments.

Education (5%): Demo account quality, in-platform training, webinars, written guides, and YouTube/video output. Beginner brokers (Plus500, eToro, easyMarkets, CMC) tend to score highest here; in my view, they are the ones that actually teach, not just onboard you and leave you to the charts.

Funding (5%): AUD deposit methods (bank transfer, PayID, BPay, Osko, debit card, Skrill, Neteller, PayPal where offered), withdrawal speed (same-day vs 1-3 business days), fees, and base currency support (AUD as a base currency is non-trivial for cost reasons: in my view, a broker that charges a currency conversion on every deposit and withdrawal quietly erodes trading capital).

Update cadence

The frequencies listed below are targets, not proof that every test has been completed. The latest dated evidence record shows what was actually checked. A missing log does not prove a scheduled test failed or never occurred.

• Spreads: Monthly snapshot for the top 10 brokers. Quarterly for the long tail.
• Platform features: Within 30 days of a broker’s release.
• Pricing changes: Within 7 days of a broker’s PDS amendment.
• Full broker rating: Annual cycle, every broker re-rated from scratch.

The “Last updated” date at the top of each review is the actual date of the most recent edit. Reviews older than 12 months are flagged in the page header until the next full re-rate, so you know the data is stale and when fresh numbers are due.

What we don’t do

A few methodological constraints worth being explicit about:

We don’t accept payment for placement. No “premium listing” tier exists.

We don’t rely on broker-supplied data; the broker’s marketing team does not get a say in the spread numbers we publish. Spreads come from our own logger, which Noam Korbl ran for our 48-hour capture from 22 to 24 September 2026. AFSL details are checked against ASIC’s Professional Registers Search. We use a broker’s PDS for pricing structure but confirm the live numbers ourselves.

A broker without a current AFSL is not reviewed. Offshore-only brokers (Vanuatu, Seychelles, Saint Vincent) are excluded by design. If a broker is not answerable to ASIC and AFCA, it is not on this site, full stop.

No “best broker” award based on user voting. Reader testimonials are useful for sentiment but not for scoring. The rankings come from the methodology. In my view, a popularity contest tells you who markets well; a live-account capture records the prices each broker actually quotes.

This site does not issue financial advice. Information on this site is general in nature and does not take personal circumstances into account. In my view, the rating tells you what the numbers are and how we weigh them; the decision about whether a broker fits your situation is yours alone.