Last week I stumbled upon this article, and tweaking my decks to better fit this template has made them run SO MUCH smoother. Of course I wasn't running enough lands, and my curves weren't as tight as they could be.
The idea behind this template is to curve out the best you can, with the assumption that games end at turn 7.
The template also assumes that you want to play your commander as fast as possible, which is why there is no spell at the mv of your commander in the curve. This also means that if you play a commander with a mana value of less than 4, you shouldn't play 2mv-ramp spells (like signets), because it would throw off your curve. If your commander has a mana value of 3, ramping on turn 2 won't make you cast it faster, so you better spend your mana on something else.
Remember that this is only a template, and it does not cover every possible type of deck. My Aesi landfall deck runs 43 decks instead of the 38 recommended by the template (and should probably run even more), while my Veyran storm deck only runs 31 (but has lots of cheap card draw and plays MDFC cards). For these edge cases I recommend you this article by the same guy, where he explains in more details the formula he used to create this tempate. He also wrote an article about how many color sources you should run to consistently cast colored spells, but none of my decks have more than two colors so I can't talk to quality of it.
Hope this is as informative to you as it was to me !
Wow that's a thorough analysis ! Thanks a lot. It does cover a blind spot of the article, which is that it is based on what people are doing instead of what they should do. I will experiment with your model !
15 cheap ramp spells feels like a tall order if you don't run green though, especially if you don't have the budget for mana crypt/mana vault/moxes. EDIT : there are more mana rocks than I previously thought, even on a budget. See this scryfall search.
Do you have an idea of how fast the output of your model changes when the average cmc of the deck changes ?
Crap, I wrote a long comment and accidentally closed the tab. Well.
In a nutshell: 15 ramp spells is doable. There are already 15 colourless rocks to fulfill this order.
So the analysis above doesn't take average mana value into account. It simply looks at the (subjectively) acceptable starting hands and goes from there. You can indirectly manipulate the assumed average mana value by modifying which hands to accept and which to mulligan.
I have experimented with a variable MV setup... But the results are pretty wild. I'll return to you if I ever get around to doing some more of that. (Yeah, that's going to be tonight lol.)