How Flintstat picks work
Every match page shows one pick with fair odds and an edge, and a win chance when the pick is on who wins
Where the prices come from
We collect odds for each match from dozens of bookmakers and exchanges through a commercial odds feed. We build the board once a day, soon after 6 AM UTC, from the prices at that moment. An upcoming match page shows that time as Updated
The best price is the highest odds any bookmaker or exchange offers on the pick, with an exchange price counted after its commission. Each pick also says from which price it is still worth it, and Check your price works that out for the price you have
Exchanges keep a commission from winnings, from 2% to 5% depending on the exchange and the customer's country. We count each exchange at its standard rate for customers outside the UK and Ireland. A pick priced at an exchange shows the price after commission next to the exchange price, and a win there counts in our record at that price
We don't name or link to bookmakers and take no money from them. You get the price and the math
How the win chance is set
The win chance comes from the odds. Each model reads the market its own way, such as removing the bookmaker margin or comparing the best price with the market average. We publish a pick only when its edge clears its model's minimum
The About page covers the bookmaker margin, edge and how to judge a model
Fair odds and edge
Fair odds are the price that matches our win chance with no bookmaker margin
fair odds = 100 / win chance in %
A 40% chance gives fair odds of 2.50. Edge shows, by the model, how far the best price sits above the fair price
edge = win chance × best price − 1
A 44% chance at 2.60 gives 0.44 × 2.60 − 1, an edge of 14.4%. Any price above the fair odds has a positive edge. How to calculate betting edge walks through a full example
Rules and estimates
No-Vig and Sharp Money estimate the chance from the prices of the whole market. Three models apply a fixed rule to the price instead
- Simple Edge rates the chance 4% higher than the price implies, so its edge is 4% on every pick before any exchange commission
- Underdog rates the chance 4% higher than the price implies, 6% higher at odds above 3 and 8% higher above 4
- Spread adds 3 percentage points to the implied chance for each point the best handicap or total line beats the market's average line, up to 15 percentage points in all
A rule sets its own edge, so the edge says little about a single pick. The track record shows whether each rule has paid off
The Fair Price model works on football and tennis. It takes the win chance from the odds of a low-margin bookmaker with the margin removed, taking more of the margin off the longer odds. A pick is the best price of the other bookmakers when it is at least 2% above that chance, at odds from 1.30 to 4.00 in football and 1.30 to 3.00 in tennis
In tennis, if a player retires before the first set is over, the stake is refunded. After the first set, the player who goes through wins the bet
How to read a match page
The verdict compares the best price with our win chance in one sentence. What stands out lists only facts the match has, such as an unusual price, injuries, rest, past meetings or weather. A section with no data does not appear
Start times show in your own time zone. Edge is a long-run average, so a single match can still go either way
After the match the page shows whether the pick won, the price we published and the result of a $10 stake. Updated then shows when the result came in. Every result is in the track record, losses included
On Basic, Elite picks appear after kickoff in History, which lists every pick
What the models leave out
- The models use prices only. Past results, form guides, lineups and late team news are not inputs. Our track record decides only which models, sports and odds ranges we publish
- Injuries, rest, travel, past meetings and weather appear for context. The market price already reflects most of them, and the models do not adjust for them
- Past meetings are a small sample. Rosters, coaches and form change between games, so they do not predict the next one
- Injury lists come from ESPN and can change until the start
- When one bookmaker sits far above the rest, it may be slow, or the market may know something we do not, such as a late injury. Check the news before you bet
Corrections
Before 10 Oct 2026 some hockey picks took their price from the 60-minute line, where a draw is a separate result, while they were settled as a win including overtime. The price shown was higher than the bet really paid, and these picks showed 10 to 13 points more ROI than they should have. The SHL picks in the track record from before that day carry this error and stay as they were published. The NHL, AHL, Liiga and Allsvenskan count from that day only
The board of 10 Oct 2026 was still built before that fix. Its NHL picks, for games of that day and the next, and its Liiga picks were priced on the 60-minute line, so the NHL counts from 12 Oct 2026 and Liiga from 11 Oct 2026
Before 9 Oct 2026, No-Vig and Sharp Money left the draw out when they worked out the chances on soccer match results. The win chances on those picks came out about a third too high, and their edges far too high. Both models now count all three results. The earlier picks stay in the track record as published
On 9 Oct 2026 we found a settlement error. Some picks were graded on another game between the same two teams, and some soccer results counted extra time or penalties. Some wins were recorded as losses and some losses as wins. The error touched 298 of about 14,000 published picks. Almost half were MLB games from April to June 2026. The rest were basketball games in April and May, the Swedish hockey final in April, Japanese and Korean baseball from July to October, 16 soccer matches from April to September and three tennis matches in May and June. We checked each of these picks against the official results of its league and corrected it. Where a game was postponed or a baseball game ended in a tie, the stake is now refunded. No pick was removed from the track record
On 18 Oct 2026 we checked the boxing and cricket results in New Leagues against new sources. Seven picks had a wrong result. Three cricket picks were graded on a game the same two teams played the day before. In boxing one fight had the wrong winner, and a fight ruled a no contest was recorded as a win, so its stakes are now refunded. We corrected them. Boxing picks whose stake was refunded because no result was found were graded again with the new sources, and so were cricket picks with no result. No pick was removed from the track record
On 19 Oct 2026 we checked 11,733 graded picks against the official results of their leagues and tournaments. One was wrong. A tennis match in Shanghai had been graded on an earlier match between the same two players, and its two picks are now losses
On 20 Oct 2026 we restored 186 published picks from April to August 2026 that had dropped out of the history. 161 of them have a result from the official source and 25 have no result
Bet responsibly
You must be 18 or older to bet. Stake only what you can afford to lose, and check that betting is legal where you live. If betting stops being fun, get free, confidential help
Flintstat publishes statistics and analysis, not betting advice