#
Massive Data Model
The Massive Data Model is the site's premium projection system, combining traditional statistical methods and machine learning techniques with extensive feature-rich datasets to project player Expected Value (EV) across future gameweeks.
#
What It Projects
- Expected Value (EV)
- Projected FPL points considering goals, assists, clean sheets, cards, bonus points, and all other point scoring events weighted by their respective probability.
- Extended Horizon
- Projects up to 14 gameweeks ahead, enabling long-term planning and strategic decision-making.
#
Methodology
- Data Sources
- Integrates historical performance data, market odds, tactical analysis, and more.
- Contextual Modeling
- Accounts for team strength & style, player roles, tactical changes, and temporary factors like penalty takers/rotation, as well of course as data recency.
- Validated Approach
- Research-backed modeling that outperforms market-based predictions (see Further Reading).
- Constant Development
- Continuously developed with annual re-evaluation and upgrades.
#
Update Frequency
- Hourly Refreshes
- Market data, news feeds, xMins, background sims and projections update hourly throughout the day.
- Post-Match Integration
- Performance data is processed overnight following fixtures.
- Pre-Deadline Updates
- Press conference information/late team news validated and incorporated before each gameweek deadline.
#
Further Reading
Massive Data: A Goalscoring Model More Predictive Than Bookmakers Odds
../../../articles/massive-data/
Ultimate Truth: How FPL Models Perform Relative to a "Perfect" Model
../../../articles/ultimate-truth/