The process

The work before the whistle.

A clear process, a published price and a record that includes the bad days. This is how the card is built, staked and settled.

Last updated 17 Nov 202618+ · No guarantees

Start with the match, not the odds.

The card starts with the fixture list, not a hunch. Each game is considered on its own before the market price is allowed to shape the view.

The first pass looks at:

  • Recent form, home and away splits, and the quality of recent opposition.
  • Expected goals for and against, chance volume and shot quality.
  • Set-piece strength and weakness at both ends.
  • Team news, rotation risk, travel and schedule congestion.
  • Table position, motivation and the shape of the match.

If the read is not clear, there is no selection. A blank card is better than a forced one.

The price is part of the read.

A view only becomes a pick when the available price makes it worth taking. A team can be likely to win and still be a poor bet at a short number.

Prices are checked across major books before publication. The odds shown on the site are the best available at the time of posting. They can move afterwards, and a price that disappears is not a price you could have taken.

Confidence is not probability.

The confidence rating on the editorial cards is an analyst score out of 100. It combines the strength of the read, the amount of uncertainty and the price. It is not a modelled win probability. The live model’s figure is different: it is a calculated probability, labelled “model probability” wherever it appears.

How to read the rating. A 76/100 confidence rating means the read is strong, not that the selection wins 76% of the time. No rating removes variance.

Ratings are deliberately conservative. A lower rating is not a suggestion to stake more, and a higher rating is not a guarantee.

The live model is a Poisson read.

The model on the homepage exists to make the reasoning checkable, not to sound certain. It uses the current ESPN table for each competition: goals for and against, games played, and the league average. Those rates are shrunk toward the league average over the first eight games, then adjusted for home and away advantage.

From there it builds a Poisson scoreline distribution for the fixture. Summing the scorelines gives a 1X2 probability and an over/under probability. The DraftKings price is converted from American odds to decimal, the margin is removed across the three-way market, and the model probability is compared with the market probability. The difference is the edge.

  • A pick needs at least four points of edge, a model probability of at least 30%, and a price between 1.35 and 4.50.
  • An edge above 12 points is treated as a data problem, not value, and is not published.
  • One selection per fixture, up to four on the card. The model does not force a card when nothing qualifies.

What it cannot see. The model does not know lineups, injuries, travel, weather, motivation or late team news. It is a starting point for the match, not a substitute for the read. ESPN and DraftKings data can also be delayed or incomplete.

Model probabilities are not confidence ratings and are not tips from the Telegram channel. They are a live calculation that changes when the table or the price changes.

Flat stakes, no chasing.

The record uses a flat EUR 100 stake so it is easy to audit. That is a bookkeeping convention, not a recommended bankroll plan.

If you bet, use a stake you can afford to lose and keep it consistent. Never increase a stake to recover a previous loss. A staking plan cannot turn a losing edge into a winning one.

Published before kick-off.

Selections are posted on the Telegram channel before kick-off with the reasoning and the price at publication. The site is updated alongside the channel.

Picks are not edited after the match starts. If a pick is void because a fixture is postponed or abandoned, the record notes the void and the stake is returned.

How results are settled.

Results settle on the full-time outcome after 90 minutes plus stoppage time. Extra time and penalties do not count unless the market explicitly says otherwise.

  • 1X2 and double chance: settled on the full-time score.
  • Over/under and both teams to score: settled on the regulation-time score.
  • Asian handicap: a half-ball line cannot push; a whole-ball line can.
  • Push or void: the stake is returned and P/L is recorded as zero.

Cash-out values are not used in the record. The record measures the published pick at the published stake.

Corrections stay visible.

If a price, fixture or result is entered incorrectly, it is corrected with a note. Losing picks are never deleted to improve the record.

The goal is a record that can be checked, not a highlight reel.

What this is not.

  • It is not financial advice or a plan for making a living.
  • It is not a guarantee of profit, a tip-off or inside information.
  • It is not a substitute for your own judgement or local legal advice.
  • It is not suitable for anyone under 18.

Betting carries risk. Read the responsible gambling guidance before you stake anything.

Questions.

The next card

Read the process. Then read the record.

The methodology explains the work. The public channel shows what it has returned, including the losses.