Corrections and Updates

Fixed in place, with a date

An error is corrected on the page itself, and the correction carries a dated note so a reader can see what changed and when. We do not delete the old wording and pretend it was never there.

This matters most with figures that carry a date, such as a leverage cap or a margin example. If the figure moves, the page shows the new figure and the date it was confirmed, so the history stays visible.

How to report a mistake

Send the page address and the sentence you believe is wrong to [email protected]. If you can, point to the broker document or source that contradicts us, because that speeds up the check.

You do not need an account with us or any special standing to report an error. A reader who spots a wrong number is doing the desk a favour, and the message is treated that way.

When a source changes without notice, we date the page against the source

When a source changes silently, we treat the page as potentially wrong from the date the source changed, not from the date we noticed. The entry records both dates where they differ: the date the source is understood to have changed and the date the page was corrected. That way a reader can see there was a window in which the page did not match the source.

We do not guess a source change into existence. If a regulator page, a broker's platform list, or a leverage cap appears different but we cannot establish when it changed, the entry says the change date is unknown and the correction date is the date we confirmed it. A stated unknown is more useful than a confident wrong date, especially on a page about margin and risk.

The page text is then rewritten to describe what the figure depends on rather than asserting a number that may move again. For example, a cost that consists of several components is described by those components, not by a single figure that will silently expire. That reduces the chance of the same sentence needing a second correction for the same reason.

The correction record: date, page, old text, new text, and reason

A correction is recorded as a dated entry that names the page, quotes the old text, quotes the new text, and states the reason in one sentence. The date is the day the fix goes live, not the day a reader first reported it, so the log matches what a reader would have seen if they checked the page that morning.

The entry sits at the bottom of the page it affects and in the site-wide log, so a reader who arrives from a search result sees the change without hunting. We keep the old wording visible in the entry rather than overwriting it silently, because a correction that hides what it corrected is indistinguishable from a page that was always right.

The reason field is short and factual: a source changed, a figure was mistyped, a platform name was missing, a regulator caveat was incomplete. It does not apologise at length and it does not speculate about how the error happened. If the cause is unknown, the entry says so. A reader should be able to read the entry in under a minute and know exactly what moved.

Who signs off a correction, and the window before it goes live

A correction is signed off by the editor responsible for the page, and it goes live within two working days of the error being confirmed. Confirmation means the editor has checked the claim against the underlying facts, not just against a reader's message. If the check cannot be completed in two working days, the page carries a visible note that a figure is under review until it is resolved.

The editor who signs off is not the person who wrote the original sentence where that can be avoided, so a second pair of eyes sees the old and new text side by side. For a regulator caveat or a leverage cap, the sign-off also requires the exact wording from the source, because those are the sentences most likely to be paraphrased into something inaccurate.

The two-working-day window is a target, not a promise about every case. A correction that depends on a third party confirming a change may take longer, and the log entry records the date the fix went live rather than the date it was first reported. What does not vary is that the entry is dated and the reason is stated, so a reader can see the gap between report and fix.

Three marks: error, update, clarification, each with its own label

Every change carries one of three labels. An error mark means the page stated something factually wrong, such as a margin figure that did not match the stated leverage cap. An update mark means new or refreshed information was added. A clarification mark means the original wording was accurate but ambiguous, and the sentence has been rewritten so a reasonable reader cannot take it the wrong way.

The labels are not decorative. A clarification is the mildest mark because nothing false was said; an update is neutral because the world moved; an error is the strongest because the page misled. A reader scanning the log can sort by label and see, for instance, that a page has clarifications but no errors, which is a different signal from a page with repeated error marks.

We apply one label per entry, not a combination. If a single edit both fixes a wrong figure and adds a new platform name, it is split into two entries so the error is not diluted by the addition. That keeps the count honest: an error entry always means an error, never an error plus housekeeping bundled together under a softer word.

Why the correction log is public and stays on the page

The log stays public because a reader deciding whether to trust a margin figure or a regulator caveat needs to see whether the page has been corrected before. A private log protects the writer's reputation; a public one protects the reader. The page is about costs and risk, so the history of its own accuracy is part of the information a reader is entitled to.

Keeping it on the page also disciplines the writing. When every fix is visible, there is less temptation to quietly reword a sentence to cover a gap. The log is not a marketing feature and it is not a blog; it is a short factual record attached to the page it describes, and it stays there for as long as the page exists.

The log does not name individual readers who report mistakes, and it does not publish correspondence. It records what changed and why, not who noticed. That balance keeps the record useful without turning a correction into a public argument. A reader can see the page has been checked and fixed without any personal detail being exposed.

How a reader checks a page's history: log, date stamp, and quoted old text

A reader checks a page's history in three places: the date stamp at the top of the page, the correction log at the bottom, and the quoted old text inside each entry. The date stamp tells you when the page last changed; the log tells you what changed and why; the quoted old text lets you compare the version you remember with the version now live.

The log is ordered newest first and each entry is labelled error, update, or clarification, so a reader can filter to the label that matters to them. If you are checking a margin figure, filter to error entries first, because an update may simply have refreshed a figure that was already right. The label tells you which kind of change you are looking at before you read the detail.

If a page shows no log entries, that means no change has been logged since the page was published, not that the page is guaranteed correct. The date stamp and the log together are the evidence; the absence of entries is a statement about history, not about accuracy. A reader who spots something that looks wrong can use the reporting route described elsewhere on this page.

Nothing is deleted: old text stays visible in the correction entry

Nothing is deleted because a reader needs to see what the page said when they relied on it. If a margin example or a leverage cap was wrong, removing the old wording hides the fact that anyone reading last month saw the wrong version. The correction entry keeps the old text quoted in full, next to the new text, so the record is auditable without a archive request.

This also applies to pages that are retired. If a page is removed because the instrument or the broker line-up changed, its correction log is not wiped; the page is marked as retired and the log stays reachable. A dead page with a visible history is more honest than a dead link with no trace of what it used to claim.

The only text we remove is material that should never have been published, such as a personal detail or a name that was included by mistake. That removal is itself logged as an entry, with the reason stated and the removed material not repeated. A reader can see that something was taken out and why, without the original sensitive content being republished.

What we log as a correction: wrong numbers, broken links, misquoted figures

An error worth logging is a statement on the page that is factually wrong or misleading in a way that could change a reader's decision, such as a wrong margin figure, a misstated leverage cap, or a formula that produces the wrong lot size. We log it when the page says something the underlying facts do not support, not when a reader simply prefers different wording.

Broken links, missing platform names, and misquoted regulator details also count because they send readers to the wrong place or leave them unable to verify a claim. A link that 404s is an error; a link that still works but points to a newer page is not, unless the old page is gone and the text still promises it. We log the second case as an update, not a correction.

We do not log a change to a live market price, a reader's own trade result, or a difference of opinion about strategy. The bitcoin reference price of 77000.0 on the page is a stated example, not a live quote, so a reader seeing a different price on their platform is not an error. If we ever label a figure as live and it is not, that is a correction.

A change is not always an error

A condition that has merely changed since we checked it is a re-check, not an error. If a broker alters its terms or its entity details after our reading date, the page is updated after verification rather than flagged as a mistake.

The distinction is simple: an error means we stated something untrue at the time of writing, while a re-check means the world moved after the page was written. Both are handled, but only the first is labelled a correction.

Correction versus update: one fixes a mistake, the other adds or refreshes

A correction fixes something that was wrong when it was published; an update adds new information or refreshes a figure that was right when written but is now out of date. The test is simple: could the original sentence have been defended on the day it went live using the facts then available? If yes, the change is an update. If no, it is a correction.

Updates are marked with the date they were made and a short note on what changed, but they do not carry the same weight as a correction because nothing was misrepresented. For example, if a broker adds a platform to its line-up, the page gains a platform name; that is an update. If the page had named the wrong platform on day one, that would be a correction.

The distinction matters because corrections are the only entries that tell a reader the page was unreliable at a point in time. Mixing the two would let a real error hide inside a routine refresh. We keep them in separate fields in the log so a reader can filter to corrections only and see how often the page has actually been wrong, not just how often it has been edited.