A comparison page prices a customer. A spreadsheet prices you. The gap between the two is usually a few hundred pounds a year, and it is entirely arithmetic. 

What is a pattern, once you take the words out of it? 

Ask someone what they intend to trade and you get a sentence. A spreadsheet wants six quantities: which instruments, how much exposure per position, how many round trips a year, how many nights those positions are carried, how much annual turnover is priced outside sterling, and how often money moves in or out. Every charge a platform can levy attaches to one of the six. 

The test is whether each can be written as a number with a unit beside it. Ninety six round trips. Two hundred and twenty position nights. Most people give the first two confidently and shrug at the other four, which is unfortunate, because the shrugged ones hold most of the money. 

A pattern not written in units is an intention, and intentions do not multiply. 

Which rows carry the weight? 

Frequency and nights, comfortably. They are the multipliers, and a multiplier turns a trivial per-unit difference into the largest line on the page. Sixty pence extra on a round trip is not worth arguing about until you write it out ninety six times and it becomes fifty eight pounds. A third of a percentage point on conversion meets forty thousand pounds of turnover and becomes a hundred and forty. 

Neither is visible per trade, which is the unit in which both look harmless. So round the quantities against yourself. If your honest answer sits between sixty and a hundred round trips, put a hundred in the cell. An estimate biased low is worse than none, since it carries the authority of arithmetic while understating what you built it to find. 

The spread column stacks four different units 

Notice first that the spread column on a comparison page stacks four different units. Pips on a currency pair, points on an index, pence on a UK share, a percentage on the conversion. None is money, and turning them into money runs through position size, which the page does not know. A headline spread also describes the calmest hour on whichever instrument the firm quotes best, so if row one points somewhere else it is a number you will never be charged. 

A rate belongs to you only once the instrument is named 

Price each line against the instruments you named, at the hours you expect to be awake, and where a range is quoted put both ends in the cell. Watch the shape of the pricing rather than its labels. One account type folds the dealing cost into the quote, another splits it into a tighter quote plus commission. A commission line looks like a penalty next to a zero until you total the column above it. 

Where do rate numbers come from before you have an account? 

This is the circular bit. The amounts you want are visible only from inside a funded account, and the account is what you are deciding about. Two sources fill the gap. The provider’s schedule, accurate about what it describes and silent about the rest. And somebody who has already paid the bills. A UK investing research site that opens and funds live accounts with its own money to test trading platforms, rather than compiling rankings from providers’ published fee schedules, The Investors Centre puts the FCA-regulated options against each other on the charges that decide the bill. 

Whatever you copy, copy the date with it. Put the source and the month beside every rate, because a rate without a date cannot be rechecked. 

Two rate cards gathered eight months apart compare a platform with a calendar, and the sheet quietly prefers whichever one you researched more recently. 

What does it look like once the multiplication is done? 

Take a worked pattern: ninety six round trips a year, five thousand pounds of exposure per position, two hundred and twenty position nights, forty thousand pounds of turnover priced outside sterling, four withdrawals. Hold those quantities identical across both columns. Price platform A on the pattern you intend and platform B on the one you have, and the totals stop meaning anything. 

Two columns is the ceiling too. Rate cards gathered in the same month stay comparable and eight of them will not, and the shortlist that gets you to two is made on authorisation and on whether row one is offered at all, which is not arithmetic the sheet can do. 

A worked illustration. The rates in the two right hand columns are placeholders chosen to show the arithmetic, not quotes from any named provider. 

Read the gap rather than the totals 

Three hundred and thirty seven pounds separates them, and the dealing lines every comparison page leads with explain two hundred and forty. The rest is conversion, financing, a monthly fee and a charge for taking your own money out. Platform B loses the headline comparison and wins the year. 

The sheet audits the rates and takes your word for the quantities 

Here is the weakness in the exercise, and it is not in the pricing. Row one holds the pattern you intend, and stated intentions have a poor record here. A 2021 study of retail CFD clients found that 10.5% of them held no positions overnight at all. 

Note what that does and does not say: the denominator is every retail CFD client in the sample, not the subset who describe themselves as day traders, so it cannot tell you what share of self-declared intraday traders end up holding overnight. What it does establish is that carrying a position into the next day is the norm across the whole population rather than the exception. A nights row of zero, written by somebody who then carries forty, leaves a year of financing off the sheet. 

The measured half and the self-reported half 

Funded testing has a matching limit, worth stating since the rate column leans on it. Paying for an account establishes what one platform charged on one instrument in a stated month, to the penny. It cannot count your round trips, and no research programme can. Every figure here is a measured number multiplied by a self-reported one, and a product inherits the weaker factor. Order two platforms with the total, do not budget with it. 

Change one quantity and watch the order change 

Run the same two rate cards against a quieter year. Twenty round trips, nothing held overnight, nothing priced outside sterling. Platform A costs a hundred and sixty pounds of spread plus twenty of withdrawals, so a hundred and eighty. Platform B costs twenty, ninety and ninety six, so two hundred and six. The winner has swapped and not one rate has moved. 

The cheaper platform is not a property of the platform. It is a property of your rows. Which means the file is misnamed if you saved it under a provider, because what it documents is a pattern, and the ordering holds only for as long as the pattern does. Save it under the pattern. Rewrite the quantities and the answer is allowed to change without anybody having misled you. 

The one column you can improve for nothing 

Rates are set by others. The quantity column is yours, and on day one it is guesswork. Three months in it need not be. Export the statements and count the closed round trips, the nights any position was open at the platform’s rollover time, the converted turnover and the withdrawals. Four counts, twenty minutes, and the guesses become measurements. 

Then rerun the sheet with the same two rate cards. Almost nobody does, because the decision is made and the file feels finished. The second run earns its keep, since it tells you whether you chose well and, more usefully, whether you are the trader you described in row one. Date the cells, keep the file, open it again in a year.