Across 103,878 settled bets from 145 accounts, bets whose records were created a month or more after the bet was placed returned 12.09% to 12.97%. Bets recorded within a day of being placed returned 6.52%.
Same platform, same settlement rules, same users. The only difference is when the record was written down.
Every settled bet carries two dates: when the bet was placed, and when its record was created. The gap between them separates betting that was tracked as it happened from history imported afterwards.
| How the bet was recorded | Lag | Bets | ROI |
|---|---|---|---|
| Logged as it happenedRecorded before the result was known | 0–1 days | 55,014 | 6.52% |
| Added within a weekMixed | 2–7 days | 6,703 | 4.92% |
| Added within a monthMixed | 8–30 days | 6,447 | 19.72% |
| Backfilled historyImported after the fact | 1–12 months | 27,042 | 12.09% |
| Deep backfillImported after the fact | over a year | 5,763 | 12.97% |
The 8–30 day band is out of pattern and we cannot explain it. It stays in the table. Dropping the row that complicates the story would be the same behaviour this study documents.
Nobody has to be dishonest for imported history to look better than live betting. Importing is a series of small choices: which bookmaker account to export, which months to bother with, which screenshots were kept. Winning bets are more memorable and more likely to have been photographed in the first place.
A record built as the betting happens removes those choices. The bet goes in before the result is known, so nothing about the outcome can influence whether it gets recorded.
That is the practical lesson, and it applies well beyond this platform. A tipster's published history, a screenshot of a winning accumulator, and your own spreadsheet were all assembled by somebody who knew the results before deciding what to include.
One aggregate query over settled bets on betr.pro, grouped by the number of days between the bet date and the date its record was created. ROI is total profit and loss divided by total turnover within each band. No individual bet, account, or user is identifiable in any figure published here.
Settled bets
103,878
Accounts
145
Tipster bets excluded
44,275
Read these before quoting anything above.
Everyone here chose to pay for betting analytics, which selects heavily for matched bettors, arbitrage bettors, value bettors, and disciplined tipster subscribers. That population can genuinely run a profit. Nothing in this study supports a claim about bettors in general, and the +6.52% figure for live-logged betting should not be read as one.
At +19.72% it is higher than every other band, including deep backfill, on 6,447 bets. We do not have an explanation. It is published because deleting the row that complicates the story would be the same behaviour this study documents.
We infer that a bet was backfilled because its record was created long after the bet date. That is a reasonable inference, not a direct observation of intent. Someone diligently catching up on a month of honest records looks identical to someone importing only their good months.
Nothing here was randomised. The gap is consistent with selective import, but it is also consistent with people betting differently in periods when they happen not to be tracking. We cannot separate those with this data.
A single UK-focused product, 145 accounts, and 103,878 settled bets. Large enough for the gap to be well outside sampling noise, far too narrow to be a general finding about betting.
FAQ
A live-logged bet had its record created within a day of the bet being placed. A backfilled bet had its record created a month or more afterwards, usually by importing past history from a spreadsheet or screenshots. The measure is the lag between the bet date and the date the record was created.
The most likely explanation is that importing is selective. People reach for the accounts, periods, and slips they remember, and winning bets are more memorable and more likely to have been screenshotted. Nobody has to be dishonest for this to happen. It is what an unaudited record looks like when a human decides what goes into it.
It means an unaudited record assembled after the fact should be treated as weaker evidence than one built as the betting happened. That applies to a tipster's published history, a screenshot of a winning accumulator, and your own spreadsheet equally.
This cohort is, but that is a fact about 145 self-selected users of a betting analytics product, not about bettors. People who pay to measure their betting skew heavily toward advantage play. We are not claiming, and this data cannot show, that betting is profitable in general.
The methodology, the exclusions, and the denominators for every band are on this page. The figures come from one aggregate query grouping settled bets by the lag between the bet date and the record creation date. No individual bet, account, or user is identifiable in any published figure.
A bet logged before the result is known cannot be filtered by the result. betr.pro reads slips from screenshots so recording one takes seconds, which is the only reliable way to keep a record you did not curate.