Evaluation Score (Eval) #
What It Is
- Rather than just summing up projected points, plans are evaluated based on all the influencing solver settings (FT value, time decay etc.).
Why It Matters
- The evaluation score and solver settings are used to find the strongest root move. Your plan for the upcoming GW is a lot more important and certain than your transfer plans in 10GWs
Considerations
- It’s possible to make evaluation literal (rewrite)
FT Value #
What It Is
To be rewritten following 5FTs
Why It Matters
- Helps generate robust sparse plans, which focus on keys moves
- A setting of 0 will create intricate & fragile plans that are far too speculative
Suggested Settings
To be rewritten following 5FTs
Solver Speed #
What It Is
Balances computation speed vs. solution thoroughness.
Setting | Typical Solve Time | Best For |
---|---|---|
Fast++/Fast+ (7/6) | 5-30 sec | Quick Tests/Mobile Use/Sensitivity Analysis |
Fast/Medium/Heavy (5/4/3) | 30-90 sec | More definitive checks |
Heavy+/Heavy++ (2/1) | 90- sec | Generally no longer recommended |
Considerations
- The gap in the best evaluation score found between fast and heavy solves is typically small, and often 0. However on occasion a better line may be found.
- Solve time is heavily influenced by hardware. An powerful CPU will be multiple times faster than a typical laptop and mobile devices will be slower.
- To ensure the best performance, keep the solver tab open in your browser and avoid switching tabs during a solve, as this may slow down computation
Time Decay #
What It Is
A decay rate used in the evaluation of a plan, that typically reduces the weight of future gameweeks.
Setting | GW1 | GW2 | GW3 | GW4 |
---|---|---|---|---|
0.85 | 100% | 85% | 72% | 61% |
0.70 | 100% | 70% | 49% | 34% |
Why It Matters
Reflects increasing uncertainty in long-term projections.
Suggested Settings
- Default Setting: 0.85 (~0.8 to ~0.9 generally considered a fine default)
- Short Term Focus: 0.70-0.80
- Long Term Focus: 0.90-1.02
Solve Depth #
What It Is
How many gameweeks deep the solver is permitted to make transfer decisions.
Why It Matters
- Avoid unrealistic “false dead-end” moves in later GWs
- Balance complexity vs. computation time. Distant planned moves become incredibly speculative and of proportionally lesser interest
- Supposing GW10 is the upcoming gameweek and 10 weeks of projections have been loaded in. A solve depth of 6 to 8GWs would generally be suggested
- Supposing GW32 and the remaining and projections for the remaining 8 weeks
Suggested Settings
- Generally it would be suggest to set this relative to your default horizon
- For a projection horizon of 6GWs, a solve depth of ~4GWs would be generally suggested
- For a projection horizon of 8GWs, a solve depth of 5-6GWs would be generally suggested
- For a projection horizon of 12GWs, a solve depth of 6-10GWs would be generally suggested