A pick is locked before kick-off together with the odds of the reference line. Once published it can be neither edited nor deleted, and the result is settled by the system from the match protocol. That is why a track record here means exactly what it shows.
Before the match starts the author picks an outcome and a bankroll share. The odds are filled in by the server from the reference line — the form cannot submit a value of its own.
When the match starts the pick is sealed: edits are barred by code, not by promise. A draft that was not published in time stays out of the statistics — and is not hidden either.
After the match the outcome comes from the protocol, not from the author's word. CLV is calculated alongside it — how much better the taken price was than the closing line.
In a channel the author decides what to show. Here a pick is locked before the match together with its odds and bankroll share, and that snapshot cannot be changed afterwards. A losing pick cannot be deleted either: it stays in the history and in the ROI calculation.
So the numbers on an author's page are not their claims about themselves, but what the system computed across every one of their bets.
From the event's quotes for the chosen market and selection at the moment of publication — that is the reference line. The author never types them in, so inflating them to make the result look better is impossible.
If there is no reference line for the chosen outcome, the pick is not published at all: with no price to lock there is nothing to verify.
CLV is the difference between the locked odds and the odds at the line's close. It shows whether an author systematically beats the market rather than whether they guessed one match right.
On a short history, return is mostly noise: thirty picks can produce any ROI. A consistently positive CLV cannot be noise. The formulas are on the rating methodology page.
You need a feed with a history: over 30 days the rating counts authors with at least 10 settled picks and an average price of no less than 1.3. The price threshold guards against betting near-certain outcomes: they produce a pretty accuracy and prove nothing.
The score is trimmed further while the sample is small: +40% over a dozen picks will not lift an author above someone with +8% over three hundred. Hidden feeds and backdated picks do not compete on equal terms — details in the methodology.
Yes. The model publishes through the same mechanism and within the same deadlines: before the whistle, at the reference line's odds, with no right to edit. In the rating it is an ordinary row and gets no allowance for counting faster.
The point is to give a human pick an honest benchmark: beating the model is a measurable achievement, while "I am up" is not.
What is public is the result: what the author bet and how it ended. A paid feed delivers the pick BEFORE the match — that is, while it can still be acted on. After the match such a pick is revealed and joins the overall statistics like any other.
A subscription is taken out for 30 days and does not renew itself: there is no recurring charge, access simply ends. The author commits to a number of picks for the paid period — if that commitment is not met, the money is due back to the subscriber.
No. Nothing here accepts bets or moves money onto a match outcome. The bankroll share is the unit in which a pick's result is measured, not an amount anyone deposits.
The platform shows which tipsters are actually right over the distance, and guarantees nothing: a positive result in the past promises nothing about the future.
Matches of the day show the authors' consensus, the stream shows every pick at the moment it is locked, the rating shows who holds a result over the distance, and Leagues and markets shows who is strong exactly where you care.