I've been thinking about the common occurrence of posts complaining about the "Rated Range" not being what it should be. I've have some very important observations that will help make this much more understandable.
1. The first scenario goes like this: I charged my car to 80%, toggled to miles and it says 250 miles. So I took 250 miles / .8 = 312.5 miles. But just the other day I saw 320 miles. So I lost 7.5 miles in one day.
The problem here is all about display resolution. The "power meter" shows you battery percentage in 1 % resolution and miles in 1 mile resolution. There are only 101 possible values for battery percentage. And somewhere around 325 different values for miles (for a LR RWD).
For any given percentage that you read on the power meter there is just about 3 different values possible for the miles display.
As an example, here is a collection of real world percentage and miles for my car observed during a charge:
37% 119 miles
38% 119 miles
38% 120 miles
38% 121 miles
38% 122 miles
39% 122 miles
In this case I actually got 4 different mileage numbers all while the percentage was 38%.
If I use the classic case of dividing range by battery percentage, I get the following:
37% 119 miles. 321.6
38% 119 miles. 313.2
38% 120 miles. 315.8
38% 121 miles. 318.4
38% 122 miles. 321.1
39% 122 miles. 312.8
I can calculate a "rated range" that varies by about 9 miles. That's about 3% of false degradation, just by looking at the display at different times.
As I got near the end of that charge, I saw the following:
54% 174. 322.2
55% 174. 316.4
55% 175. 318.2
55% 176 320.0
55% 177 321.8
56% 177. 316.1
Only about a 6 mile difference in "rated range". This makes sense, in that the variance of the "rated range" calculation decreases as we get close to 100%. By the time you get to about 100% we should be seeing a maximum of 3 miles of variance.
I use Teslafi and I noticed that the data collected by Teslafi has battery percentage in 1% resolution and mileage in .01 resolution. But this doesn't really help because there is still up to 3 or more miles of range that can be accounted for a single percentage number. Having percent battery capacity as a 1% resolution number is a major culprit to why Teslafi's battery health graph is so noisy.
If you use Teslafi, you can go to battery health and change the "Include charges equal to or greater than" number to 100% and you will most likely see that your data only varies by about 3 miles. And the sneaky thing is that Teslafi knows this because they actually give you the option to only include certain charges and they limit that number to at least a 50% charge. That's because the error gets bigger, due to simple math, as the SOC gets lower. If you included charges that stopped around 5-10% you would see huge swings in your "rated range". If you went down to 0% your "rated range" calculation would actually go to infinity.
Changing that percentage to different numbers will basically decrease the volatility of the end result when you limit to higher final percentages.