What is the difference?
The key difference in concept between the Transfer Solver & Linear Optimiser is how they evaluate any given selection, and following on from that what the output actually means.
Non-linear Evaluation
- Pros:
- Autosub modelling in full detail, a team with several high risk players requires a stronger bench. ie. full autosub contribution probabilities calculated
- Same team GK situation handled. ie. if a team owns 2 keepers from the same team and there is uncertainty about who will
- Cons:
- Guarantee of finding optimality is lost, best found root solution likely to be optimal or very close but cannot be a 100% guaranteed
Linear Evaluation
- Pros
- Output will be optimal solution in the linear evaluation framework
- Cons
- Fixed autosub assumptions & non-handling of same team GKs reduces evaluation quality
- Linear optimal similarly does not guarantee non-linear optimality, but again the best found root solution likely to be optimal or very close but cannot be a 100% guaranteed
Which to use?
The best answer is probably to test both approaches. If they both land on the same root move it should further increase confidence.