Ouch! That Shoe Doesn't Fit!!
If you scan up for the selected 10_GM, you'll notice that 24 Mocking Tongues requires Diamonds this week, which clobbers our highest value cell, the 3, if we're trying to avoid the use of Diamonds. As a consequence we have to overload the remaining 8 Ingredients.
The second best Ingredient for 10_GM, the 2, is ALSO the best Ingredient, the 3, for 08_GS. So no matter WHAT we do to Maximize the Success Rate for 10_GM, we'll never be able to catch up with 08_GS. This isn't at all uncommon. The Recipe Library uses a Green Conditional format to catch your attention when you bet on the 2nd place horse, and we flag the checkboxes as well.
The way we've set things up, the only thing you EVER need to do with the Solver is to click on the Solve button. "Old" Witch Point Only configurations aren't useful as Brewing Recipes, but they're a dandy way to look at the importance of each Ingredient.
The GOOD New is that the Witch Point / Diamond ratios don't change much from Week to Week, because a Least Cost Recipe necessarily has a lot of small amount ingredients, and STILL does a decent job of delivering a respectable Success Rate.
So, with 25*12=300 or so Least Cost Recipes to work with, plus the various Witch Points Only recipes to consider, you should be able to find SOMETHING that suits your fancy.
Also, every column in the Recipe Library can be Filtered and Sorted, and the Graphs are dynamic, so it's pretty easy to compare stuff.
Note that we are NOT optimizing the Total Number of Witch Points, they're calculated after the optimization is done, but the Total Number of Ingredients will get you close, unless you went wild on a single Ingredient or two.
If you do get interested in the Solver, ALWAY start with blank Ingredients so that you'll get repeatable results. The Solver (and the Brewery for that matter although it's less obvious) is sensitive to preloaded configurations, so if you're sequencing, and you mistype, you'll get some pretty baffling results that are nearly impossible to replicate, because you mistyped "something." Save yourself the aggravation; select the _Initialize range and then use the Delete Key so that you'll always have a clean start.
Now that I have some data to work with, I'll be looking at the symmetry in the Recipe Space, and will hopefully be able to show that we actually have ONE basic series of Recipes, and ELEVEN shadow Recipes. If that works out then I'll have a Recipe State Network that I can use to chase down the Shortest Global path, rather than the current approach which calculates a likely path, and then goes for a spline fit.
If you're the least bit intrigued by the heavy duty math behind the Sum of the Squares stuff, you can bend your mind at
Least Squares - The Gory Details and
The Art of Problem Solving.