Offline
Decision Center
What does the available data say?
Transparent probabilities, with their source and limits — no live news, no promises.
Predictions are probabilistic and may be wrong. Do not bet more than you can afford to lose; this application does not execute any bets.
Your Recent Analyses
Results appear with their source tag and calculation time.
No matches yet
Add a match or upload a file you are authorized to use to start the analysis.
Your Inputs
Add Match for Analysis
Use your own data or demo mode; we do not fill gaps with guesses.
01
Match Details
Required fields are mandatory; others improve data completeness.
Analyze match list from CSV
Upload a file you have the right to use. Supported columns: home, away, league, date, home_elo, away_elo, home_gf, away_gf, home_ga, away_ga.
لم يختر ملفSource Transparency
Data Sources and Connection
We do not assume a subscription exists. Any number without a trusted source remains incomplete.
This interface does not connect automatically to any provider. Real integration requires a secure proxy server that keeps keys outside the browser, and a review of the terms of use.
◈Not configured
Public or licensed source
Add the provider name and endpoint for documentation purposes only. The key is not sent from this interface.
▧Active locally
Demo Data
Static examples for trial only; not live results and do not represent an actual team or market.
02
Provider Authentication
The key is not saved here. Use a backend server to manage secrets.
Connection Status
No connected sources. Live data is unavailable.
Pre-claim measurement
Historical Test
Time separation between training and testing, with a baseline comparison.
We do not display results for an unexecuted test. Upload a historical CSV containing actual results to enable measurement.
لا توجد عينة اختبار كافية
أضف بيانات تاريخية مع أعمدة home_score و away_score، ثم شغّل الاختبار بعد تحديد نسبة التدريب.
Brier—
Log Loss—
Calibration—
How do we calculate?
We use Poisson for goal distribution, with home advantage and Elo adjustment when available. We derive 1X2, over/under 2.5, and both teams to score probabilities from the same distribution. xG, injuries, and lineups are only included when received from a reliable source; there is no trained machine learning in this version.
Does not capture all injuries or lineupsSensitive to sample quality and tournament homogeneityIndependence between goals is an approximation
Track no forgotten predictions
Predictions Log
Log the odds before knowing the result, then add the actual result for comparison.
Log is empty
Every saved result will appear here with its source and creation time.