The figure being advertised is the best one anybody recorded, and how good it looks depends on how long somebody sat there. Only one of those two ever gets printed.
The word reports a minimum, and readers hear an ordinary price
Start with the kind of quantity being advertised, which is settled before any question of whether the advertising holds. Lowest reports the least a figure ever was. What somebody deciding where to open an account wants to know is roughly what a fill will cost them on a Tuesday. Those are two different quantities off one series, and any moving quantity has a best reading and a usual one, the best being the flattering half of the pair and therefore the half that gets printed. The trouble specific to a dealing price is how far apart the two can sit, and how little in the phrase warns anybody that they have been handed the wrong one.
The stretch somebody watched is doing half the work
A price series manufactures a fresh number continuously, which makes a floor a property of a window as much as of a firm. Watch a screen through a quiet fortnight and a screen through a fortnight containing a rate decision and the same pricing yields two different records, neither of them dishonest and neither of them the other’s rival.
Windows also get chosen after the numbers are in rather than declared before them. Name one and you have handed a reader something to check and yourself something to be wrong about. Name none and every window stays available afterwards, including the ones nobody thought worth mentioning, and a reader has no way of telling a fortunate stretch from a representative one.
A minimum also behaves unlike the averages people are used to reading, and this is the part that does the damage. An average settles as observations accumulate: gather more and the figure moves less, which is why anybody trusts one. A minimum never settles. Each further observation can only leave it where it was or push it down, so the number keeps sinking, slowly and indefinitely, with the quantity of watching done.
So a floor is a joint product of what a firm charges and of how many times somebody looked, and only the first ever gets stated. Two firms advertising different floors may be pricing almost identically and differing in how thoroughly each sampled its own screens. The reader believes they are comparing two prices. In front of them are two prices and two unstated counts.
Almost every version of the claim is true
The familiar complaint about advertising is that it says false things, and this one rarely does. Across enough instruments and enough moments, at a firm passing a raw price through and charging separately for the service, a reading as narrow as the phrase implies will almost certainly have occurred. Approach it as a lie and the audit fails at once, because the screenshot exists and the firm can produce it.
A figure earns its place by ruling something out, and this one rules out very little. Excluding so few states of the world, it changes almost nothing about what anybody should expect to pay. That is a different defect from dishonesty and it wants a different response: not prove it, which they can, but state it, which they have not.
An unarguable sentence can still be aimed at a genuine difference, and this one is. A firm that states a commission and passes the market price through untouched has split its charge into a part it sets and a part it does not, and for a wide class of dealing patterns that structure genuinely comes out cheaper all in. Nothing measured here prices the two arrangements against each other across firms, so read that as reasoning about how a charge is assembled rather than as a finding. It is also the half of the argument a reader can act on, since which way a firm charges is settled by the account terms and holds for every trade placed through it, which is true of no reading taken off a screen.
| What the printed floor depends on | Whose side of the screen it sits on | Ever stated beside the figure | What a reader puts there instead |
| The price the firm was willing to make | The firm’s | Yes, that is the figure | Nothing, this part arrives |
| The instrument the reading came off | The firm’s | Sometimes, in small type | Whichever pair they deal in themselves |
| How long the screen was watched | The observer’s | Almost never | An assumption that it was long enough |
| How many readings were taken in that time | The observer’s | Never seen | An assumption that it does not matter |
| Which of the readings got printed | The observer’s | Never seen | A price they might reasonably expect |
The inputs to one printed figure, sorted by which side of the screen each of them sits on.
The more attentive party reports the lower floor
The method that ought to settle all this cuts the other way. A bill records what one account was charged on one occasion, and a printed floor reports the best of many occasions. To The Investors Centre, which opens and funds live accounts with its own money to test UK trading platforms rather than compiling rankings from providers’ published fee schedules, that gap is a working constraint rather than a debating point, since the funding buys occasions one at a time and never the best of them.
The sharper trouble follows from the third section. A programme watching a screen closely will honestly record a lower floor than one glancing at it, so between two scrupulous parties the smaller figure belongs to the more attentive rather than the cheaper. Any number of that kind carries the effort behind it as well as the price in front of it, and no published count of readings separates the two.
The version a firm could plausibly publish
A complaint that the whole shape of a price series goes unpublished has to reckon with what publishing it would involve. A quantity changing many times a second does not go in a table, and a firm attempting it would be maintaining a data product instead of a page. Asking for the entire distribution is asking for something nobody would read, so the request should be dropped.
One extra fact does nearly all of the work. Print the count of readings the figure was drawn from, and the number stops being a floor of unknown depth and becomes a floor somebody can weigh: one out of a hundred is a different object from one out of four, and a competitor watching the same screens can contest it.
Which half of the cost can be checked before you deposit
All of which has a practical consequence. Cost on a currency trade arrives in two parts: one printed as a fixed figure in the account terms, the other generated by a market while you deal.
The advertised phrase describes the moving half and says nothing about the fixed half, though the fixed half is the part anybody can verify before sending money and the part that holds still while they read.
Which inverts the order of the questions. Settle the structure first, since it is stated, checkable and applied to every trade placed through the account. Take the advertised floor last, and read it as an indication that a firm runs raw pricing at all, which it usually is, rather than as a measurement of what your dealing will cost, which it never was.
…