I've noticed that for a number of freshmen on my teams, the color-coded potentials for certain attributes have gotten worse even though their numeric rating hasn't changed. For example, I have one player who started the season with 75 speed and yellow potential, and now has 75 speed with red potential. Based on the pinned guide to HD 3.0, I was under the impression that the colors mapped directly to the maximum potential rating in that attribute:

Green is High High (28+ Points)
Blue is High (20-27 Points)
Black is normal (8-19 points)
Yellow is High-Low (4-7 points)
Red is Low Low. (0-3 points)

So I interpreted "75 speed with yellow potential" to mean that this player (assuming sufficient work ethic and practice minutes) will cap out at 79-82 speed. However, now it seems like he'll end up with 78 at most. Is this just a rounding issue, where under the hood he actually has improved from, say, 74.8 to 75.2, and that was enough to cross the threshold into red potential? Are the maximum potential ratings static?

All ratings have a hidden decimal and the maximum potential for each player is static. Your player probably improved his speed by 0.1 or 0.2 and that was enough to change from yellow to red.

two things:

1) player ratings are decimal values. starting and maximum ratings are, however, integers (i believe - although something may have changed in 3.0 - maybe ratings are rounded up instead of truncated, i think they used to be truncated - which wouldn't change the prior comment. but i haven't taken the time to sort that out). so if a player starts with the minimum potential (max rating - minimum potential) for any category, they'll show a color change before the rating shows +1 increase - because the player has basically gained a fractional point in that rating, enough to put them in the lower category color wise, but not enough to show a +1 increase.

2) yes, the max ratings are static. but also the chart you have is a little weird. its kinda hard to think about but the reality is - while every player starts with an integer amount of growth (i think), the potential range has to cover the entire double value range. in your chart above, what is a 3.5? its left out of the chart (same with 7.5, 19.5 or 19.2 or 19.8, etc). fyi the black should be 7-20 if you are gonna break it down that way though, then yellow would be 4-6 and blue 21-27. anyway i don't honestly know what a 6.5 is, or a 6.8? i am relatively positive a 6.2 is a yellow. i think yellow either goes up to 6.5 or 6.999. i think i used to know for sure but i just don't pay enough attention anymore. basically you can check this by tracking a bunch of the black ratings for the first couple days of a season, identifying the subset that was a 7 minimum, and tracking which % of them go yellow on the very first day, versus the second day, along with the minutes they practiced and played (and WE) and when they actually ticked up to the higher rating point. in short if everyone is ticking on practice 1, then probably yellow goes up to 6.99. if some folks are taking say 2 days to change color and 3-4 days to tick to the next rating, then yellow only goes up to 6.5.

edit: i actually cannot remember if 0-3 on red is off by a point or if that is the correction. so i might not be 100% on that bit. someone somewhere knows the exact answer...

It happens sometimes w/ rounding. Not really anything you can do. They could theoretically round the numbers and prevent this from happening, but it would have to be done consistently.

Oh, I mean it's not a problem at all; I just wanted to make sure that I understood correctly and that it was in fact an artifact of rounding, rather than actual changes in true potential. Thanks everyone for clearing that up!