How the numbers are made.
Every figure on the board can be traced to a sampled quote, a published schedule, or a stated assumption. This page is the full reference.
Live pipeline methodology
Spread methodology v1.0
Last updated 2026-07-04. Changes are versioned, not silent.
The historical dataset is versioned separately on its own methodology page; the PropSpread Score by the scoring engine, see below.
Sample, normalize, rank.
Three steps between a raw quote and a row on the board. This part versions the spread measurement itself.
Sample
Continuous spread ticks from connected feeds, never scraped screenshots.
- Every bid/ask snapshot is stored with its timestamp
- Nothing is transcribed from marketing pages or screenshots
- No connected feed means a waiting state, never an invented number
Normalize
Pips, cents and points converted to dollars per lot, published commissions included.
- Pips, points and cents all become dollars per one-lot round turn
- Each firm's published round-turn commission is added on top
- Slippage stays a separate, labeled assumption and is never mixed in
Rank
A rolling 15 minute window decides the leader, so no single tick can.
- A rolling 15 minute window decides, never a single tick
- One extreme quote can neither crown nor sink a firm
- A silent feed keeps its last values plus a visible staleness note
Sampling cadence: feeds are polled on a short interval (seconds, not minutes) whenever the board is open, and the on-page sample counter is cumulative for the session. These are sampled spreads from the feeds we track, not quotes you can trade on. The order can shift as new samples arrive.
One quote, followed to the dollar.
The Normalize step performed live on this page: the current tightest EUR/USD quote walked through the exact arithmetic the board applies to every firm, every tick.
From a terminal to your screen.
Four hops, each one inspectable. Which firms ride a verified feed at any moment is exactly what the disclosure states below declare.
- Trading terminal
A funded-account terminal at each connected firm quotes bid and ask through the session, the same prices a trader on that account would see.
- Collector bridge
A small collector polls those quotes on a short interval and stamps each snapshot with its time. It adds nothing and averages nothing.
- Aggregation backend
The backend stores the samples, maintains the rolling window and computes ranks. A quote is never edited after the fact.
- Your browser
The board renders straight from that state. When a value changes you see it flash; when a feed stops you see a staleness note, not a guess.
Demo data says demo.
The interface derives its wording from the data source, so a false combination is unrepresentable. These are the three states and the exact source line each one shows.
Demo (simulated)
- An in-process generator produces the figures
- Everything is watermarked as demo data
- Nothing claims to be a live market
Shown in product: “All figures are simulated for demonstration.”
Connected, demonstration data
- The transport connection is real
- The backend is fed by a simulated publisher
- The badge says exactly that
Shown in product: “Connected to a simulated feed for demonstration, not real account data.”
Connected, verified feed
- Figures come from connected account feed snapshots
- Only here does the watermark disappear
- Only here does the word live stand unqualified
Shown in product: “Figures derived from connected account feed snapshots and published schedules, as noted.”
One rating, published math.
The PropSpread Score blends measured trading costs and execution with payout reliability, rules transparency, support, platform stability and community feedback, weighted toward what we can measure live.
Scores publish per firm once its data qualifies; a firm without a published score shows Score pending.
Live methodology details are unavailable right now. Category weights are shown only when served by the scoring engine's active version, never from a stale build-time copy.
Score data sources
Trading cost inputs draw on the platform's own spread sampling. Execution, payout, rules, support and platform metrics come from provider documentation and public sources; community reflects public reviews and sentiment. Current inputs are demo-labeled until verified live feeds ship.
Until a firm's score is computed and published by the scoring engine, its cards show an explicit pending state instead of a provisional number. Scores never override the measured spread ranking on the board, and paying firms cannot buy a higher one.
How historical figures are measured.
Every statistic PropSpread publishes about the past is computed from stored observations by these rules. They are written so a published figure can be reproduced, and so the questions the data cannot answer are stated rather than dodged.
What is observed
- Bid/ask snapshots from each firm's connected feed
- Append-only audit trail, kept at full precision
- Nothing rounded, smoothed or interpolated on the way in
How history is stored
- Raw observations kept for a bounded window
- Hourly and daily rollups kept long term
- Rollups computed from raw only, because a median of medians is not a median
Which observations count
- Zero, negative, crossed or implausible quotes are excluded
- News and rollover spreads stay, since dropping them would flatter every firm
- Exclusions are counted and published, never deleted
Market hours
- Every observation classified open, closed or unknown
- Daylight-saving-aware calendar, per market
- Headline statistics use market-open observations only
Median, average and percentiles
- The average is a sample-count-weighted mean, which is exact
- Median and percentiles come from per-period medians, which is robust
- The median leads because spikes drag averages
Sample counts and coverage
- Every statistic carries its sample count and period
- Too little evidence produces no statistic at all
Commission and all-in cost
- Commissions come from published schedules, dated and sourced
- A past period uses the schedule in force back then
- Unverified schedule: the cost reads unavailable, never estimated
Comparing firms
- Medians compared over the same period only
- Every compared firm must have valid observations, or the period drops
- A tie is awarded to nobody
Time zones
- All periods, buckets and dates are UTC
- Calendars evaluated in each market's own timezone
- Session boundaries stay correct year round
Historical coverage
- Coverage begins when continuous sampling began
- Nothing is backfilled or reconstructed from a current snapshot
- The real span is served live, so it can be checked rather than taken on trust
Reverification
- Commission schedules are re-checked and carry their last-checked date
- Statistics are recomputed from the stored observations
- Every published figure is reproducible from the dataset and these rules
What this data is not.
The list a reader would otherwise have to infer. Every item here is a limit we would rather state than have somebody discover in a live account.
Prices and fills
- Figures are indicative snapshots for comparison, not tradable quotes. Actual fills vary by account type, server, liquidity and time of day.
- Slippage is a normalized, clearly labeled assumption, not a measured value.
Commissions
- Commission schedules for FTMO, FundingPips and FundedNext are read from each firm's own published terms and were last checked on 2026-08-08. Institutional Funding carries a placeholder schedule that no published source confirms. Schedules vary by account type and product, so confirm the terms of the account you would actually trade.
- Metals commission at some firms is a percentage of traded notional rather than a flat fee, so the per-lot figure shown is a projection at a reference price and moves with the metal's price.
Coverage
- Sampling covers the instruments and firms listed on the board; a firm can price differently on instruments we do not sample.
- Historical statistics cover only the period since continuous sampling began. PropSpread does not hold multi-year history and does not reconstruct it.
Rankings and statistics
- The Average Prop Firm benchmark is a computed composite of sampled competitor feeds, not the published pricing of any specific firm.
- Rankings describe the sampling window shown and may change as new samples arrive.
- Period medians and percentiles are computed across per-period values, which is a robust estimator rather than the exact quantile of the pooled distribution.
Full disclaimer
This tool provides an indicative comparison of spreads and estimated execution costs based on periodic snapshots from connected data feeds and normalized trade assumptions. Displayed values are not tradable quotes and may differ from prices available on any provider's live accounts. Spreads vary by account type, server, liquidity conditions, and time of day. Competitor names and marks belong to their respective owners; no affiliation or endorsement is implied. The "Average Prop Firm" benchmark is a computed composite of sampled competitor feeds, not the published pricing of any specific firm. Cost estimates use a normalized trade scenario and do not constitute financial advice or a prediction of trading results.
The questions we get asked.
Short answers, all of them restating what the sections above commit to.
Where do the spread numbers come from?
From bid and ask snapshots collected off connected account feeds while the board is open. Nothing is copied from marketing pages or screenshots, and a firm with no feed shows a waiting state instead of a number.
Why convert everything to dollars?
Pips, points and cents cannot be compared directly, and commissions hide real differences. Converting each quote to dollars per one-lot round turn and adding the published commission puts every firm on one scale, so the comparison is between totals, not labels.
Why rank on a rolling window instead of the latest tick?
A single tick can be an outlier in either direction. The rolling 15 minute window smooths spikes, so a firm can neither win nor lose the ranking on one quote.
Is any of the data simulated?
Some of it, and the interface always says so. Every data source carries one of three disclosure states, and demonstration data is watermarked as such. The exact wording of each state is listed in the disclosure section on this page.
What happens when a feed goes quiet?
The last received values stay on screen with a visible staleness note. Nothing is extrapolated and nothing is silently recalculated.
Can a firm pay for a better position?
No. Board order comes from measured spreads, and the PropSpread Score comes from the published methodology on this page. Sponsored placements, where they exist, are labeled and never change the measured ranking.
Why is a firm's cell sometimes empty?
Because nothing trustworthy exists to show yet. A firm without a connected feed, or without enough samples, gets a waiting state with the real count, never an invented number.
Do weekends and holidays pollute the statistics?
No. Every observation is classified against the market's own calendar, and headline statistics use market-open observations only. A weekend quote never joins an average.
Can a published figure be reproduced?
Yes. Statistics are recomputed from stored observations under the rules on this page, and commission schedules carry the date they were last checked, so any figure can be traced back to its inputs.
This tool provides an indicative comparison of spreads and estimated execution costs based on periodic snapshots from connected data feeds and normalized trade assumptions. Displayed values are not tradable quotes and may differ from prices available on any provider's live accounts. Spreads vary by account type, server, liquidity conditions, and time of day. Competitor names and marks belong to their respective owners; no affiliation or endorsement is implied. The "Average Prop Firm" benchmark is a computed composite of sampled competitor feeds, not the published pricing of any specific firm. Cost estimates use a normalized trade scenario and do not constitute financial advice or a prediction of trading results.