Stats
Both teams' last 10 league games this season: goals, corners, cards. Expectations are pulled toward the league average.
drag the dot
We look at every match from four angles. The moment the model sees more chance than the market, the pick lands on your phone.
Angles
Both teams' last 10 league games this season: goals, corners, cards. Expectations are pulled toward the league average.
Searches bookmakers' opening odds for patterns that held before. Flags a match when one bookmaker's three prices sit in the same ranges.
At the break it reads the second half from the first-half score and closing odds: is one more goal likely?
During the match it weighs xG and shot flow against the market price. If the model is not above the market, it stays out.
Cards
Every pick reaches your phone as one card: the match, the odds, the probability. When the match ends, the result is stamped on that same card.
↓ scroll to open the deckMarkets
Home team wins.
At least 3 goals in the match.
The match ends with 3 goals or fewer.
At least one goal in the second half.
Yellow cards: Over 3.5, Over 2.5 or Under 4.5.
Total corners, Under or Over against the league norm.
Live
Scroll and the match plays. See how Probs watches a pick and when it sends it to your phone.
↓ scroll · 0' → 90'sample scenario · not a real match
Value check
Sometimes the price rises because the match is turning against the pick. So every time the price moves, the model recalculates: if probability × odds is above 1, the card goes out; if not, it does not.
In the sample, at 9' the odds are 1.31 and the model says 75%: the product is 0.99, no card. At 10' the odds jump to 1.36 while the model holds at 76%: 1.03, the card drops.
0,76 × 1,36The odds reached 1.40 but the model fell to 68%. The product is 0.95: however tempting the price, nothing is sent.
0,68 × 1,40Reading the probability
73%
73% means that if this card came 100 times, about 73 would win. The other 27 happen too; judge the run, not one card.
The white mark on the bar is the probability implied by the odds. When the bar passes it, the model sees more chance than the market.
The percentage blends the model's view with the market price. When the model drifts far from the market, the truth usually lands in between.
Anatomy of a card
A match day in the app
Picks for matches starting at the same time share one list, with a notification.
Minute, score, odds and probability. On your phone the moment the card drops.
The card itself turns WON, LOST or VOID, with the score below.
Every pick of the day on one list, wins and losses together.
Coverage
Times in Istanbul time (GMT+3). We look at hundreds of matches a day and leave out matches with unreliable data (friendlies, youth and reserve teams).
Track record
A losing card stays in place, stamped LOST. At 23:58 every card of the day arrives in one recap, wins and losses together.
App
sample screen
FAQ
No. We take no bets and broker none. We publish picks and probabilities; how you use them is your call.
From "probably". We promise no certainty; every card shows its probability.
Pre-match cards come 5 minutes before kick-off, with matches starting together on one list. Live cards come during the match, the moment we see value.
We do not send a card for a match where we see no value. Few cards means we are selective.
73% means about 73 of 100 such cards win. One card can lose; the run is what counts.
We cannot collect reliable prices for these markets. Those cards show the expected count and the probability instead.
If the match is postponed or cancelled, or card and corner stats are not recorded, the pick is void and the card is stamped VOID.
No, it is a sample built to show the rule. Every card and screen on this page is a sample.
Each card is stamped in place when it settles. At 23:58 every pick of the day arrives in one recap.
iOS and Android versions are in the works. We will announce the launch on this page.
It starts in Turkish and English; more languages are coming.