First off, this is a lovely idea for a handbook. Being able to figure out how much initiative is enough at a glance would be very handy.
Now this may be nitpicking about math and wording, but this handbook has math in the title, so I'm just going to go ahead. "Diminishing Returns" here are slight, if they diminish at all.
If you have 50% win rate on initiative, you lose 1/2. Increasing 50 percentage points here cuts your losses in half.
If you have 75% win rate, you lose 1/4. Increasing 25 percentage points here cuts your losses in half.
If you have 87.5% win rate, you lose 1/8. Increasing 12.5 percentage points here cuts your losses in half.
The thing about percentages is that higher percentages are worth more. Moving from 50% to 75% is generally worth as much as moving from 0% to 50%. To put it another way, consider if you had % damage reduction. If you had 50% DR, you effectively have 2x health. If you have 75% DR, you effectively have 4x health, NOT 3x. This is why, for instance the armor formula for League of Legends (100 / (110+armor)) has perfectly linear returns. A character with a 50% initiative win rate wins 1/2 initiative rolls, while a character with 100% wins *all* of them. That's increasing returns for the same number of percentage points. In order to have merely even returns, you need to be earning less percentage points with each bonus.
Also, with the popularity of solutions like Ring of Anticipation, it would probably be good to include data on how rerolls affect probability. Without such figures, the handbook can't really help you much with figuring out how much initiative is enough.
You could also combine this with optimization by the numbers and add in the expected initiative (average to max) for different CRs