The problem is rather the EA roster.
I have now simulated over 20x12 years, most with TFG roster, but also some times with the EA roster as a comparison, and everything read out in excel (PC, madden20, portet EA m21 week17 roster, porter TFG roster).
Almost all positions have the problem that the average, especially the young players, is too high. This leads exactly to what has been mentioned here more than once: The new rookies of the gen classes don't get playing time and can't get better.
With the HBs the problem is really bad, which is probably why you noticed it so much.
Here is the graphic from the EA simulation of the HBs. The first one shows the OVR development of the position 1-64 (not of the player) of signed players over 12 years. The second graph shows the average OVR of the top half (top 32) and the third graph shows the average age of the top 32. The fourth graph shows the spread of the age by position 1-64 over 12 years.
It can be clearly seen that the OVR decreases for several years and the age continues to increase. The halfbacks are there almost only with players from the EA roster, which have developed even more over the years, and are now only very slowly exchanged for new rookies. They are then all significantly worse and the OVR drops massively as a result. The problem occurs more or less on every position.
An increase of the XP slider unfortunately brings nothing, because the EA Roster players become even stronger. This increases the OVR curve in the first few years and partly causes even more problems.
What helps is a roster like TFG, which has a lower average OVR and fits the draft classes.
How many classes did you look at for this, and how exactly did you compare it to make such a confident conclusion?