The trick here will be to find a list length that can be calculated based on the count of the full list of followers. Maybe 10%? Maybe 1%?
The goal is to generate both the full and recent lists, then combine them to add weight to the recents while maintaining the scope of all follower counts.

Although the auto ranging feature was cool to figure out and implement, there needs to be a different mechanism that can select the recent list lengths based on the full list length. The lists vary, such as the number of direct followers is usually longer than the skip28 followers, so it is here that the ratio should be calculated and extracted. Also, the length of the recent list does not flow to the jackpot games like it does to the dailies. There are less than a few hundred results from something like the Power Ball vs the PA evening pick 3 which has nearly 17,000.
Percentages can be easily calculated and rounded (can't add 0.25 of a drawn number...)

First to imagine the solution, figure out the algorithm, then the implementation. There is definitely a cyclic nature to this coding stuff when applied to a problem such as a best guess at a lottery draw.

Turns out the code to do just that was relatively easy to implement, was even able to reuse some variables.

The first setup uses a hard coded 10%
That means if there were 2,000 observations of the full list, there would be 200 for the recent list, and when weight is applied, it  rings the list to 2,200 followers.
Because it uses a percent, it is still auto ranging to the number of full observations, which varies per draw position and drawn number.

Before undertaking the next step (converting the entire script) I will pass the percentage when calling each function.

There are 9 functions...
Pick 3 (works for day or night, depending on the .csv file called in)
Pick 4
Pick 5
And the most widely recyclable, MatchSix, as this can be used for M6, MM and PB.

As the script length grows, it is through attention to detail that these changes are made without causing errors.

Passing in a percentage will be via decimal, for example, 10% will be passed in as 0.1

There us one master module that handles the process for one position. Pick 3 has 3 modules. There are 18 modules in total that must be modified which is why I always start with the pick 3.

I am currently in the pick 3 making changes and testing. I thought it would have taken longer. Now the tweakable parameter becomes the percentage passed in, and each game may do better with a different value, that is the hard part fornthere is no guide for such tasks.

Project updated! All games!
The first live test of the weighted follower system will be Monday with the PB and Tuesday with the MM. Paper play for pick 3 this weekend.
I would clean up the old print lines that were commented out, but you never know when you want to break down the stats, so for now this version will be copied and archived, and a newer, leaner version crafted when time permits.

Summary of changes...
1. The weighted pick was added, achieved by simply combining the full and most recent lists
2. Auto ranging based on the last drawn number for each position is no longer active.
3. Auto ranging based on the number of total followers for each position, with a passable percentage was added. Currently set to 20% on pick 3/4/5 and 10% on Match6, PB and MM.
4. The decision to strip down the code was scrapped in favor of archiving the verbose script and editing a new one, just in case the output of stats is needed.
5. Here is hoping it works at some point, because the stripped down version will be the one that gets back tested.
6. Hoping I have no more "bright ideas" because then 2 versions of the script (current one with 2,200 code lines) have to be edited.