#
Transfer Solver vs Linear Optimiser
FPL Review offers two distinct solving approaches that are complementary tools providing valuable cross-checks. Both often find very similar solutions, but on occasion there can be reason to use one or the other.
#
Core Differences
#
Transfer Solver (Full Evaluation)
Algorithm
Chess engine-style heuristic search that explores promising sequences of transfers while pruning obviously poor choices. An in-house built engine designed to tackle the vast solution space for the FPL problem efficiently.
- Full Evaluation: Full probability-based calculations recognising non-linear complexities
- Autosubs: Distinguishes between the autosubs contribution generated from 11 nailed players vs a team with several players who may not play, calculates full autosub probabilities based on xMins, availability and usage
- Same Team GKs: Naturally handles same team GK effect. For example two same team GKs @45xMins each will be similar in value to one GK with 90xMins (slightly worse due to small chance of a sub)
- Optimality: Cannot guarantee total mathematical optimality but will often find it in practice
- Forced Decisions: Works with constraints but can be temperamental with excessive or highly constraining forced decisions (in terms of budgetary effects, or highly "against the model" transfer to "Any Player")
- Speed: Generally solve times are consistent based on the settings, hardware and scale of problem
#
Linear Optimiser (Linear Evaluation)
Algorithm
Formulates FPL planning as a Mixed Integer Linear Programming (MILP) problem and solves it using the HiGHS solver, a state-of-the-art optimisation engine that guarantees mathematical optimality within the linear framework.
- Eased Evaluation: Linearised with fixed sub & vice-captain contributions for mathematical optimisation
- Autosubs: Uses fixed settings for each sub role rather than full probability calculations
- Same Team GKs: Does not handle same team GKs. For example two same team GKs @45xMins each, the sub GK will contribute EV based on the fixed Sub GK setting
- Optimality: Guarantees optimal solution within the linear framework
- Though not guaranteed non-linear optimal (however will often find it)
- Forced Decisions: Handles constraints natively due to mathematical framework
- Speed: Sometimes can solve extremely quickly though tends to be a little more random
#
When Each Might Be Preferred
#
Transfer Solver
- Scenarios where rotation risks and autosub probabilities matter significantly
#
Linear Optimiser
- Scenarios with many constraints where the mathematical approach handles complexity better
#
Practical Usage
Best practice: Use both as complementary tools. When they agree, you can be confident in the solution. When they disagree, determine which plans align to your preference.