First off, this is going to be a long post. We love that our users provide suggestions and we draw new features from your suggestions a LOT. However, this is a request that needs more detail, because there isn't a good, obvious way to handle minimum exposure. If we implement minimum exposure the way we implement maximum exposure, the optimizer will have a high chance of giving undesirable results.
We chose to handle this situation indirectly through groups and calculate more rather than directly. Minimum exposure would be enforced with players who wouldn't otherwise get into your lineup due to their lower projections but get forced into your lineup. So, what you can do is create a group of "risky" plays and force all lineups to take at least one of them, and you can actually get minimum exposure if you set maximum exposures carefully. For example, if you have a group of 4 players but want someone's minimum exposure to be 20%, set the sum of the other 3 players' exposures to 80%. Of course, this annoying to calculate and not easy to do compared to having a box for minimum exposure like we have for maximum exposure. Alternatively, you can increase the exposure of a player by locking the player and doing a "calculate more" then removing a few lineups from the previous crunch. However, if you have a lot of minimum exposures, that you want to enforce, trying the calculate more method would make it clear how difficult this problem actually is.
So why are we so resistant to implementing minimum exposure? Because it's so easy to do, it's dangerous for new users. There is no unique way to enforce minimum exposure, so when we create an algorithm we need to come up with a way that makes sense to the user. With max exposure we have two ways to enforce it: group and each. With max exposure, group acts in the following way: look at the number of lineups requested, what ever percentage the exposure is set to, that is the limit, remove the player from the pool once that limit is reached. If you want 50% exposure to a single player and you're asking for 100 lineups, that player gets removed from your pool and won't get picked again. How would this work with minimum exposure? It would work the opposite, if you want 50% minimum exposure, you limit the number of lineups without the player, once this is reach, you lock the player for the remainder of lineups. This is fine if your exposure is only set on a small number of players, but if it's set on a lot of players then what happens is, you get lineups that you normally would without exposure, then when you reach the limit of lineups you can have without the players that were left out, then suddenly you're going to be locking all of these players together for the next few lineups. Remember, minimum exposure would only be enforced with players whose projections are on the low end and would normally not be used in many lineups, so the lineups toward the end would just be stacked with multiple "lower projected" players and the early lineups would just be the lineups you would've had without exposure.
Then we have the situation with the each exposure setting. Under max exposure, what it does is it keeps the proportional exposure constant as lineups are being generated. If a player has a max exposure of 50% under each, it means that if a player is used in the first lineup, they are removed for the second, added back into the 3rd lineup, if they are used in the 3rd, they are removed in the 4th. With minimum exposure, it would work the same way except backwards. If they are not used in the first lineup, they are locked in the second etc...
If we look at an example of what would happen with "risky" plays (I'm only choosing risky plays, because again, minimum exposure would only be enforced with "suboptimal" lineups), say we have 4 "risky" players set at 25% exposure. Because of their low projection, they would not be picked in the first 3 lineups, then the 4th lineup they will all be locked together (assuming they fit). Again, you're getting 3 lineups that you would've received anyway, and a 4th lineup that you would never want to use.
Of course, if you are fairly experienced and you're very, very careful about setting up your pool of players and their projections, you may be able to trick the optimizer into doing what you want. However, if you can do that, you can probably do it using groups and/or calculate more. The difference with groups is that it's much more difficult to end up with the bad situations you find with minimum exposure. We don't want to push out a feature that is very difficult to use properly. However, if anyone has an idea on how to implement minimum exposure that doesn't cause low projection players to get lumped together just to meet the requirements we are open to suggestions.
NBA Chat Room
