21 was my test number for a few games. But I like 35 since it will force your team to 2-Long faster.

This gave me an idea a few weeks ago. Idk when I will try to implement it but the basic strategy would be to use the whole system backward .

According to jdb the recommendation is below 10 but you suggest 20 might be better . The scale still seem small in comparison to the 100 available .

Idk why I think backward would work but a setting a 50 or 80 might send stuff toward a desired result. Pending you put all your long plays in the short section and all your short plays in the long section.

Maybe not? But I want to try it...I think I would just need to sit down calculate it first

I wouldn't go above 50. After 50, things get really wonky. Here's why:

First play: 15 yards
New first down short: ((5 * 0.5) + (15 * 0.5)) = 10 yards (Compared to 5)

Next play: -5 yards
New second down short ((10 * 0.5) + (-5 * 0.5)) * .75 = 1.875 (Compared to 3.75)

These are some what extreme, but not really. Basically you ping pong wildly above and below the static values.

35 seems to work well though in terms of immediate course correction without your game planning being a roll of the dice each play.