How we record and grade tennis picks
How Hunchmeter records and grades tennis picks: match winner, game and set handicaps, total games, retirements, priced and result-only picks.
Tennis on Hunchmeter
We record public tennis picks from outlets, tipsters and channels, note when each one was seen, grade it against the official score, and link every pick back to the page or post it came from. Tennis is played every day across dozens of tournaments, so pages are organised by match day (Eastern Time) and by tournament.
Which picks we grade
Match winner, handicap and total — each in games or in sets when the pick says which. A handicap or total without a clear unit (for example “+1.5” that could mean games or sets) is listed as Not graded. Set betting, correct score in sets, single-set markets, aces and other player statistics are listed but not graded yet.
Retirements and walkovers
Handicap and total picks on a match that is not completed — a player retires or withdraws during play — are graded as a push, unless the total was already decided before the retirement. A match that does not start (a walkover) voids every pick on it.
When picks enter the record
Recorded before first serve: we captured the pick from the source's public page before the scheduled start. Timestamped before first serve: we read the page after the start, but the platform's own post time (a public Telegram post or a YouTube publish time) is before it. Picks read after the start without a pre-start timestamp are kept in the recording history and never graded.
Prices, points and CLV
Where we have the market price at the time of the pick, the pick is Priced: its return at that price counts towards ROI, and market-adjusted points reward wins at longer prices (+(1−p)/p for a win, −1 for a loss, where p is the no-margin market probability). CLV compares the price at the pick with the last price before the start. Picks without a recorded price count as Result only.
Models
Sites that publish a probability for every match are prediction models, not experts. Their projections are graded on a separate Models page and never appear in expert records or leaderboards.
One pick per source per market
If a source posts the same market on the same match more than once, the earliest pick with the best time proof counts; later ones are shown in grey as Repeat.