53 platforms, GBP 250,000 and 1,100 hours. What does that actually buy? 

A research site publishing its testing scale invites the obvious question, which is whether the numbers are large. Divided out, they are smaller than they look. 

What are the figures? 

The Investors Centre, a UK research site, states that it has tested 53 platforms with live accounts, using more than GBP 250,000 of its own capital across more than 1,100 hours of hands-on work. Those are unusually specific claims for the sector, where most methodology pages describe a process rather than quantify it. 

Specificity is worth something on its own, because it is falsifiable. A site claiming rigorous independent testing is making a claim nobody can check. A site claiming 53 platforms is making one that could be embarrassed by a reader who counts. 

Less impressive, which is the honest way to read them. GBP 250,000 across 53 platforms is roughly GBP 4,700 per platform, and capital deployed into a trading account is not spent, it is at risk and largely recoverable. The real cost is the losses taken and the time. 

1,100 hours across 53 platforms is about 21 hours each. That is a substantial working week per platform, enough to open an account, fund it, place a reasonable spread of trades, time a withdrawal and write it up. It is not enough to observe a platform across a full market cycle, and nobody should read it as such. 

The same figures read as a sceptic would read them. They support a narrower claim than the headline suggests, and the narrower claim is still worth more than a fee-schedule summary. 

Is 21 hours per platform enough? 

For the questions that testing is uniquely able to answer, largely yes. What a round trip actually costs, what the spread does during a data release, whether a withdrawal arrives when requested, how long verification takes, what the platform charges that its tariff does not mention. All of that is observable inside a working week. 

For questions about reliability over years, obviously not. Whether a platform handles a crisis well, whether its support degrades under load, whether its pricing drifts after acquisition, are questions no amount of concentrated testing answers. They need longitudinal observation, and 21 hours is a snapshot. 

How does this compare with the alternative? 

The dominant model is compiling comparisons from providers’ published fee schedules, which costs almost nothing per platform and therefore scales to hundreds. That produces wide coverage and, on the specific matter of what a platform costs in practice, no information at all, because a fee schedule describes intent rather than outcome. So the trade is depth against breadth, and it is a real trade rather than a marketing line. A reader wanting to know whether an obscure broker exists is better served by the wide list. A reader wanting to know what a platform will charge them is not. 

What would make the claim stronger? 

Dates. A count of 53 platforms is more useful with the testing date attached to each, because a platform tested three years ago has since changed its pricing and possibly its owner. Sample sizes per platform would help too, since twenty round trips and two hundred support very different confidence. 

Publishing the losses would be the strongest move available. A site that funds accounts necessarily loses money on some of them, and stating that plainly is both true and more convincing than any assertion of independence. Short of that, the team at The Investors Centre opens and funds live accounts with its own money to test UK trading platforms, rather than compiling rankings from providers’ published fee schedules. 

That goes further than most of the sector manages and stops well short of a full audit trail. 

How quickly do the findings go stale, and what should a reader check? 

Faster than the sector admits. Spreads move with market conditions, commission schedules get revised, inactivity terms change with a month’s notice, and platforms are acquired and repriced. A cost figure recorded eighteen months ago is a historical fact rather than a current one, and readers rarely check the date. 

Which means the maintenance burden on a testing model is heavier than the initial research, and it is the part that quietly gets skipped. A site that tested 53 platforms once and a site that maintains 53 platforms continuously are doing very different amounts of work and can describe themselves identically. 

The date on the specific claim they care about, first, because that single check invalidates more bad information than any assessment of methodology. Then whether the figure is presented as measured or as reported, since those are different and the distinction is often blurred by phrasing. 

And whether the site says anything that costs it something. Coverage limits, findings that favour a platform with no commercial relationship, admissions that a test was inconclusive. A body of research with no inconvenient results in it has either been extraordinarily lucky or is not reporting everything it found. 

Should a reader care about any of this? 

Only instrumentally. Nobody chooses a broker because a review site spent 1,100 hours on something. The figures matter because they indicate whether the numbers in the reviews came from anywhere real, and a site willing to publish a scale it can be held to is more likely to have done the work than one that is not. 

That is a modest conclusion and it is the correct one. Testing scale is a proxy for credibility, not a substitute for checking the specific claim you care about. Take the one figure your decision turns on, find its date, and see whether the site will stand behind it. That single check is worth more than the whole of this article.