When to take control yourself
The signals that mean you should be setting rules, and which control fits which situation.
Manual control is worth using exactly when you have information the software does not. That is a narrower set of situations than most people assume, and a much more valuable one.
| Situation | The right control |
|---|---|
| News says a player is being worked into a bigger role than his projection reflects | Minimum exposure — a floor that still lets the build use him more if it agrees |
| You will not roster someone at any price | Exclude |
| You believe in a player but not enough to be heavy on him | Maximum exposure |
| You need a precise structure for a specific contest | Exact exposure |
| You are certain about one player and building around him | Lock — but be aware every lineup now shares him, which reduces your diversification |
| You believe in a particular game breaking open | A stack rule with a bring-back, or an exact-players stack |
| You want your single-game portfolio to hinge on one player's big day | Captain/MVP exposure, not a plain lock |
| Your lineups look too similar to each other | Max shared players under Advanced settings — carefully; it is a hard rule |
Rules to follow when setting rules
- Fewer is better. Every rule removes freedom the build was using to diversify. Three good rules leave it room to work; thirty leave it none, and the portfolio ends up more uniform, not more expressive.
- Prefer maximums and minimums to exacts. An exact says you know the right number. A minimum says you know the direction, which is usually all you know.
- Lock sparingly. A lock is a 100% exposure and it is the fastest way to turn a portfolio into one bet. A high minimum usually expresses the same conviction with far less collateral damage.
- Check what your rules cost. Save the build before and after, and Compare. If a rule moved nothing, remove it; if it moved more than you intended, you have learned something about how tightly the slate was constrained.
If a rule set becomes impossible the build refuses and names the rules that disagree. That is the system working, not a failure — but it is a signal you have over-specified. Relax the least important of the named rules first.