Average coverage on the slip
65.7%
of the model's weight
reading a score
Why a score is printed with a percentage next to it
Every name on the slip carries a second number that looks like a footnote and is not. It says how much of the scoring model that bookmaker could actually answer, and on this site that is usually a good deal less than all of it.
Lines scored on the whole model
1 of 7
The heaviest criterion
40 of 100
what the games give back
Gaps filled with an estimate
none
not one, anywhere on this site
Two scores that look alike
MSport sits on line 1 with 84.0 points, computed over 60 per cent of the model. BetFox sits on line 7 with 53.0, computed over all of it: 690 slots and 191 live tables, counted. Printed as bare numbers those two are directly comparable. They are not directly comparable, and the percentage is how a reader can tell.
| Bookmaker | Slip Index | Coverage | Criteria it could not answer |
|---|---|---|---|
| MSport | 84.0 | 60% | what the games give back (40 points) |
| Wiwi Casino | 76.0 | 60% | what the games give back (40 points) |
| Soccabet | 72.0 | 60% | what the games give back (40 points) |
| betPawa | 67.0 | 60% | what the games give back (40 points) |
| SportyBet | 65.0 | 60% | what the games give back (40 points) |
| Betway | 58.0 | 60% | what the games give back (40 points) |
| BetFox | 53.0 | 100% | none |
Why the gap exists at all
Most Ghanaian lobbies are served only to a browser inside the country, or built so that the catalogue never appears in the page a fetch receives. That is a limitation of where this research was done rather than a failing of the bookmakers, and it lands on the criterion this publication weighs most heavily: what the games give back, 40 of the 100 points.
The alternative was to weigh it less so the gap did not show, which would have made the model dishonest in a quieter way. A publication about what games give back cannot decide that games matter less because it could not count them.
The floor
A record has to answer 60 per cent of the model's weight before it is scored at all. Below that, averaging over what could be read stops being a score and starts being a reward for having a thin record: a book that answers three criteria brilliantly and five not at all would float to the top on the strength of what nobody could check.
The same research that produced these records also rejected 110 circulating figures about these fourteen brands. Each bookmaker's page says how many were rejected on its own record and what kind they were. The model in full.
About reading a score here
- What does the percentage under a bookmaker name mean?
- How much of the scoring model that record could answer. A book scored on 60 per cent of the model has been averaged over the criteria that could be read, and the criteria that could not be read are not in the average at all.
- Would it not be simpler to score a missing criterion as zero?
- Simpler and wrong. A zero says we know and the answer is nothing. Scoring an uncounted lobby as zero would rank a book we could not read below a book we know has no casino, on the basis of our own reading failure.
- Why not fill the gaps with estimates?
- It was tried elsewhere in this project and it was worse. Estimates immediately became the loudest signal in the market and put one bookmaker on top of eight tables out of ten. An invented figure does not stay quiet: it decides the ranking.
- So is the leader really the best?
- MSport scores highest on the criteria it could answer, which is a real statement and a narrower one than it looks. It is printed with its coverage figure attached everywhere it appears, so that a reader can see how much of the model is behind it.