For a long time, our ELO system in Seven – Card Game followed the traditional approach used by many competitive games: ratings changed based on whether a player won or lost, adjusted by the rating difference between both players.
While this worked reasonably well, it didn’t always reflect what actually happened during a game.
A narrow loss after excellent play was treated almost the same as a complete collapse. Likewise, a player who won through strong memory, concentration, and strategy received the same reward as someone who simply scraped by.
To better support the goals of Seven – Card Game, especially our focus on memory improvement and skill development, we’ve introduced a major update to the rating system.
## Why We Changed the Formula
Traditional ELO only looks at:
– Your rating
– Your opponent’s rating
– Whether you won or lost
This means the system ignores important details such as:
– Outstanding play
– Serious mistakes
– Time-outs
– Dominant victories
– Performance throughout the game
We wanted ratings to better reflect actual player performance.
## The New Hybrid ELO Formula
Our updated system still uses classic ELO as its foundation.
The traditional expected score is calculated using the standard formula:
Expected Score = 1 / (1 + 10^((Opponent Rating – Player Rating) / 400))
This produces the probability of winning based on the rating difference between the players.
The classic ELO adjustment is then:
ELO Change = K × (Actual Result – Expected Result)
Where:
– Win = 1
– Loss = 0
This part remains unchanged.
## Adding Performance-Based Adjustments
On top of classic ELO, we now calculate a Performance Modifier.
The modifier rewards strong play and penalizes significant mistakes.
### Positive Performance Factors
Players receive bonuses for exceptional achievements:
– Winning with a 16er
– Winning with a 24er
– Winning with 4 points plus the last trick
These achievements indicate particularly strong performance and are rewarded accordingly.
### Negative Performance Factors
Players receive penalties for:
– Bad moves
– Time-out failures
– Allowing the opponent to achieve major milestones
This helps the rating system distinguish between:
– A well-played loss
– A poor loss
– A dominant victory
– A lucky victory
## Encouraging Good Play Even in Defeat
One important design decision is that players can still earn performance bonuses even when they lose.
Why?
Because we want the rating system to encourage:
– Learning
– Good decision-making
– Memory improvement
– Strategic play
A player who loses but performs well should not be treated the same as a player who repeatedly makes mistakes.
This makes the rating system more motivational and more accurate.
## Time-Outs Matter
In multiplayer games, time management is part of the skill.
A time-out often means:
– Loss of concentration
– Memory failure
– Poor decision speed
For this reason, time-outs carry one of the strongest penalties in the system.
However, we intentionally reduced the penalty compared to earlier experimental versions to avoid overly punishing occasional mistakes.
## Dynamic K Factors
The system uses different K factors depending on player rating.
| Rating Range | K Factor |
|————-|———-|
| Below 1200 | 40 |
| 1200–1799 | 32 |
| 1800–2199 | 24 |
| 2200+ | 16 |
This allows:
– Faster movement for newer players
– Greater stability for experienced players
– Reduced rating volatility at higher skill levels
## Full Transparency
Perhaps the biggest change is not the formula itself.
It’s the transparency.
We recently launched our ELO Explainer system, which allows anyone to inspect exactly how a rating change was calculated.
The explainer shows:
– Expected score calculations
– K factor selection
– Bonus calculations
– Penalty calculations
– Performance modifiers
– Final ELO changes
– Formula validation checks
– Long-term rating projections
No hidden calculations.
No black box.
Everything is fully visible and verifiable.
## Looking Ahead
The current formula is already producing more realistic rating changes than the previous system, but we consider this an ongoing process.
Future improvements may include:
– Additional performance metrics
– Statistical validation against large player samples
– Formula versioning
– Predictive rating simulations
– Community feedback adjustments
Our goal is simple:
Reward skill, memory, consistency, and strategic play as accurately as possible.
As always, we’ll continue to monitor results and refine the system based on real gameplay data.
See you at the tables!
— The Seven – Card Game Team