Why there is no win rate, and no ROI

There is no win rate on this site, no ROI, no yield and no profit figure. Not because they are pending — because they are not coming. The archive publishes counts and nothing else.

A rate from a small sample says more than it knows

Thirty settled days is not thirty measurements of skill. Nineteen wins in thirty reads as 63%, and 63% is entirely ordinary for a fair coin — you would see it, or better, often enough that it tells you almost nothing. But printed on a page it does not read as "almost nothing". It reads as a property of the product, and the reader takes a decision on it that the arithmetic underneath cannot carry.

Nor is it symmetrical: a run that flatters us gets published, and by the time the sample is large enough to mean anything the early number has already done its work. The honest version of that number is one nobody wants to read — a wide interval around a figure that will move a lot — and the flattering version is one we should not print.

A claim like that needs evidence gathered before it is published

A strike rate that means something has to be designed before the data exists, not computed afterwards from whatever happens to be in the table: a period stated in advance, a rule fixed in advance for what counts and what is excluded, and a record that nobody involved could have edited. We do not have the last of those yet — the archive is timestamped by our own server, which is stated plainly in what the hash does not prove — so any rate we printed today would inherit exactly that weakness, and it would be the most persuasive thing on the page. Evidence first, number afterwards. Or no number.

And it is not ours to claim in the first place

Presenting a product as something that improves your chances at a game of chance is prohibited outright. That is not a formatting preference we could quietly change our minds about: the public record is built from a single column — whether each main pick won, lost or voided — and never loads odds, stake or any amount at all. There is no number in scope for a later edit to divide by another one, which is what stops the cheap version of this mistake ("it is only one more field").

What is published instead

• Counts: won, lost, void, and how many settled days there are in total.
• The current run at the newest end of the record, in either direction. Voided days are skipped rather than counted or treated as breaks — a void never happened, so ending a run on it would be inventing an event and extending one would be worse.
• The first and last settled date, so you can see the denominator you are working with.

A total is a fact. A ratio is a claim. You are welcome to divide ours yourself, with the sample size in front of you and our name off the result.