top of page

October 2023: Why 3% is a misuse of statistics

The Gambling White Paper, High Stakes: Gambling Reform for the Digital Age, published by the Department for Culture, Media & Sport on the 27th April 2023 proposes affordability checks, now renamed financial risk checks, on online gambling with a £125 net loss within a month or £500 net loss within a year prompting a supposedly frictionless check for CCJs and bankruptcy and enhanced financial checks, where banks statements and payslips will be requested by gambling operators, when a gambler has a £1,000 net loss within 24 hours or £2,000 within 90 days. DCMS states:


These enhanced checks are narrowly targeted and we estimate only around 3% of online gambling accounts will be affected.[i]


This article argues that this 3% figure has been misused by the Gambling Commission in an attempt to claim that these overly intrusive financial checks will only affect a tiny proportion of people and thus any objection to them would be overblown. This article will show that such a claim is untrue and suggests that the Gambling Commission is wilfully misusing statistics to achieve its own policy goals.


While the figure of 3% is used by DCMS in the White Paper, the original proposal for affordability checks and the source of the 3% figure is the Gambling Commission in their submission to the Gambling Review which led to the White Paper, Advice to Government - Review of the Gambling Act 2005.[ii] Subsequent to the publication of the White Paper, the Gambling Commission has only ever referred to the affordability proposals as the government’s proposals even though page 24 of the Commission’s submission clearly outlines the same thresholds and time periods as found in the White Paper. DCMS fully accepted all of the Gambling Commissions proposals for affordability checks.


Page 25 of the Gambling Commission’s Advice to Government outlines their estimates on the number of online gambling accounts that would be subjected to checks. This is re-created here:


Key financial             Category of           Proposed threshold   Percentage of    Number of

risk for                       assessment            for consultation          accounts             accounts

consumers                                                                                       subject to           subjected to                    

                                                                                                           checks                checks 

                                                                                                           (estimated)        (estimated)


Significant                 Simple check          £125 net loss              21.2%                 6.1 million

financial                     using opensource  per month

vulnerability              data


Binge gambling        Enhanced                £1000 net loss            2%                      600,000

                                    checks                      per 24-hour


Significant                  Enhanced                £2000 net loss           3.2%                  1,000,000

losses over                 checks                      in rolling 90

time                                                               days


The source data for these estimates is given as: ‘The binge gambling and significant losses over time information has been estimated following an industry data request which covered approximately 19 percent of all active remote gambling accounts in the May 2020-April 2021 period. The significant financial vulnerability information came from data obtained from consultation responses to our call for evidence’.[iii]


In my previous article, which can be found here, I show how the supposed evidence for the need for these affordability checks was sparse to say the least.[iv] The evidence for these estimates is similarly scant. The industry data request is referenced to a page on the Gambling Commission website that provides for a downloadable excel spreadsheet so the data can be reviewed.[v] That data is simply eleven cells of data. The total number of active accounts is given as 5,867,022 and then for each affordability measure, the number of active accounts that exceed a range of loss figures is given: The table is recreated here:


Total number of active customers that reached the following net              Number of active customers

expenditure during any rolling 24hr period

Loss exceeding £750                                                                                           60,047

Loss exceeding £1000                                                                                        116,880

Loss exceeding £1250                                                                                        80,722

Loss exceeding £1500                                                                                        68,835


Total number of active customers that reached the following net              Number of active customers

expenditure during any rolling 90 day period

Loss exceeding £1000                                                                                        364,800

Loss exceeding £1500                                                                                        250,822

Loss exceeding £2000                                                                                       189,384


Total number of active customers that reached the following net              Number of active customers

expenditure in 3 consecutive months

Loss exceeding £300 in three consecutive months                                      139,062

Loss exceeding £500 in three consecutive months                                       83,454

Loss exceeding £700 in three consecutive months                                       55,895


Simple maths can provide how the estimates where arrived at. This data is a sample of 19% of all active accounts. Therefore:


Total number of active accounts = (5,867,022÷19x100)= 30,879,063.


We then come up against some minor data issues about what the Commission’s own statistics say about how many active online accounts there actually were. Firstly, the time period for the sample is May 2020-April 2021, while the Commission gives population numbers for April 2020 to March 2021. Not a massive issue admittedly but worth noting. The main issue is that in the July 2022 version of the industry statistics, itself an update on the November 2021 version, the total number of active online accounts is given as 32,000,000, while in the November 2022 version, the total number is 31.88 million.[vi] One expects that revisions are happening the whole time and thus this reduced figure is more accurate. However, it is still 1 million more accounts than that suggested in the affordability estimates, 3.1%, again not a major difference but worth noting.


The number of active accounts having a net loss of over £1,000 in a 24 hour period is 116,880. This equates to (116,880÷19x100) = 615,157 of the total active accounts or (615,157÷30,879,063x100)= 1.99% or a rounded up 2% and a rounded down 600,000 checks as stated in the Commission’s Advice to Government.


The number of active accounts having a net loss of over £2,000 in a 90 day rolling period is 189,384. This equates to (189,384÷19x100) = 996,757 checks, rounded up to 1,000,000 by the Commission or (996,757÷30,879,063x100)= 3.2% as exactly stated in the Commission’s Advice to Government.


As stated above, the significant financial vulnerability information came from data obtained from consultation responses to the Commission’s call for evidence and as these are never published there is no way of checking them.


What we have is the sparsest of data sets that are impossible to check in order to prove or disprove the estimates made by the Commission. There are minor deviances from the maths and from what the Commission has stated, but none of this is significant enough to argue that there has been a misuse of the statistics even though we cant check the data to see if its real.


There are three issues that make up the case for arguing a misuse of statistics and the first two can be found in the notes of the Gambling Commission’s industry statistics.


The first is that the time period used, May 2020-April 2021. As most will remember, the first national lockdown was announced on the 23rd March 2020 which lasted until the end of June. The second national lockdown came into force on the 5th November 2020 and ends on the 2nd December 2020. The third national lockdown starts in January 2021 and ends in phased manner starting on the 8th March 2021.[vii] The impact on live sport was considerable, for many months throughout this period, live sport did not happen.[viii] Online gamblers were forced either to not gamble, to gamble on foreign sports or to gamble on gaming products. Even though the Gambling Commission brought out extra regulations to prevent what it thought would be a tsunami of problem gambling, which never happened, the changes to gambling behaviour were not significant enough to be critical to the nation’s health but were definitely abnormal and unlike any year in living memory. The Gambling Commission’s own research showed that lockdown meant many gamblers spent more.[ix] For the Commission to use this time period appears odd as other data sets exist and with only the sparsest of data supplied, can only lead to a suspicion of data manipulation.


The second issue is expressed quite clearly in the Commission’s industry statistics: ‘Customers may have accounts with more than one operator and therefore the data relates to accounts rather than the individuals’.[x] While the Commission’s own data on the average number of accounts only goes up to 2019, it states that ‘44% of online gamblers were registered online with one gambling company, however, the average number of accounts held in 2019 was three, which has remained stable compared to 2018’.[xi] It would be fair to assume that during Covid-19 lockdowns, with gamblers having more time to play on their phones, they would have signed up to more accounts, but using these numbers, the issue of the 3% being misused comes to the fore. While the Gambling Commission has been keen to always state that only 3% of accounts would suffer enhanced checks, they have then made the mistake of conflating this with the number of account holders, as they did in their extraordinary open letter to the Racing Post who have been campaigning against the new affordability measures. The Commission stated:


The financial risk checks consultation is Racing Post readers' chance to engage in the development of policy, and we would invite your views on how the 0.3 percent of account holders could have their financial risk assessed if they are not asked to directly provide the additional financial information.[xii]


Obviously if each gambler has 3 accounts on average, this means that the number of account holders is  (30,879,063÷3)= 10,293,021 using the Commission’s estimated numbers for the number of account holders. If there are to be a million enhanced checks this implies one in 10 account holders will be impacted – 10%. Obviously this would only apply if the distribution of checks was evenly spread, which it wont be, it will be focussed on those who spend more, so more likely to have all three of their accounts checked, which leads to the next point found in Commission’s notes to the statistics:


Active accounts are those that have been used by customers in the last 12 months[xiii]


The exact definition is that an active account is one used by a gambler at least once in 12 months. A better term may well be ‘near-dormant’ for a large number of accounts. What we know from a report which the Commission and DCMS cite most frequently, Patterns of Play by Nat Cen and the University of Liverpool is that firstly the majority of betting volume comes from a small percentage of players and secondly, ‘a large majority of bettors either finished ahead or else lost only modest amounts (when compared with spending on other common leisure pursuits)’ and gives the example:


  • 4.4% of accounts lost more than £1,000 over the year

  • 2.2% lost more than £2,000

  • 0.7% lost more than £5,000[xiv]


The report continues with ‘The proportion of accounts with a ‘large’ loss is small but still represents a significant number of individuals. For example, our estimate that 2.2% of accountholders spent more than £2,000 over the year implies that more than 190,000 bettors with the seven operators incurred that level of loss’.[xv] While Patterns of Play looks at player losses over a 12 month period and the Commission considers a 90 day rolling period, the numbers for a £2,000 loss look similar, 190,000  v 189,384, but they cannot be assumed to be from the same data set. Also, as can be seen, Patterns of Play states firstly that 2.2% of accounts lost more than £2,000 then says 2.2% of accountholders spent more than £2,000. The same conflation between accounts and accountholders as the Gambling Commission. Whilst in both cases, there is no reason to believe in malicious intent and it could just be simple typos, the difference is a still a significant factor of three and for the sake of good policy making there needs to be some clarity, especially when the Gambling Commission is making repeated claims that just 3% of accounts would receive enhanced checks, when it is about 10% of customers at least who will be affected on this basic calculation.[xvi]


When it comes to gaming, Patterns of Play provides less insight. Similarly, a small number of players provide the majority of stakes, and also similarly ‘3.1% lost more than £2,000’, it is not defined whether these are accounts or accountholders but the implication is that it is accountholders as the accompanying graph provides Percentage of Customers as an axis label.[xvii] Again it would be useful if the evidence used to make gambling policy was provided in full, so it can be analysed transparently and without ambiguity to what data is being considered.


The third issue, concerns what has already been mentioned about Patterns of Play, that of the Pareto effect, where a small number of customers provide the majority of stakes. What this means is that there is not an even distribution of expenditure amongst betting accounts, it is concentrated amongst a relatively small number of account holders, who not only stake and lose more than the majority but will have undoubtedly have more accounts than the average of 3, as they will be more price conscious and thus choosing the best odds between a large number of accounts. This means that for the more robust gambler, the chances are that not only that they all will have to suffer enhanced checks, but probably have to suffer them on the majority  of their accounts. The Gambling Commission is stating in its consultation on the new affordability checks that gamblers will be able to gamble their remaining account balance but will not be able to deposit any more until the enhanced financial checks are completed.[xviii] What that will mean is that for a robust gambler, they will rapidly end up in limbo, unable to gamble as various operators conduct their enhanced checks. Many of these robust gamblers are semi/professional gamblers who make either a majority or a substantial part of their income from gambling, as this income can not be considered as income according to the Gambling Commission, not only will these enhanced checks take a long time to complete but the likelihood is they will end up with a negative conclusion.


How many of the customers will be impacted is a simple answer, practically all of them but we just don’t know. The Racing Post, the only media outlet brave enough to campaign against these unwarranted and intrusive checks conducted a survey of 9,000 of its readers and found that 16.6% had already been subjected to affordability checks, which is under the system introduced in 2019, where it was up to the operator to set a limit – the Commission suggested a £550 loss per month.[xix] This can be easily critiqued as a self-selecting and referring to a different regime but does give some indication of numbers. More prescient is the forecast by William Woodhams, chief executive of Fitzdares, a top end bookmakers catering to big punters, who stated to the Post, ‘My estimate is that 80 per cent of Fitzdares customers will trigger checks’.[xx] The question of how many punters will have to provide documents to continue betting, was also looked at in the newspaper with Bahar Alaeddini, a partner and gambling law specialist with law firm Harris Hagan stating, ‘An educated guess is very significantly higher [than 3%]. As it stands, only the Gambling Commission will have the answers in the 2CV research (unpublished) it has commissioned in tandem to the consultation’.[xxi]


All we do know is that the 3% is being used politically -  a misuse of statistics – to suggest that just a very small number of people will be impacted by these unprecedented, unwarranted and unevidenced affordability checks. I will be writing to the  Office for Statistics Regulation to ask that they write to inform the Gambling Commission:


  • To ensure that when they refer to the 3% of accounts they don’t conflate this with accountholders and clearly state this is for all online gambling accounts the majority of which are rarely used

  • That they mention that at least 10% of account holders will be impacted

  • That they release the data which provided the 3.2% figure so it can be scrutinised

  • That they release the 2CV research data so a true figure for the number of people who will be impacted by enhanced checks will be provided


Obviously, not of relevance to the Office for Statistics Regulation is the expected outcome of these affordability checks, which will in effect stop the robust gamblers from providing the much needed liquidity in the betting markets and the Levy money that subsidises horseracing prize money. They will do what any normal consumer would do when confronted by overzealous and unnecessary regulation, go elsewhere, they will go to black market and gamble un-taxed, un-leviable and unprotected. British racing will collapse and not a single problem gambler will be helped.


[i]   page 4


[iii] Ibid p. 25












[xiv] Ibid p.30

[xv] Ibid


[xvii] p.60



[xx] Ibid


bottom of page