My players just finished Curse of Strahd and are ready to move on to Rise of Tiamat. I have two players with two characters each, and they left Curse at 10th. One player is happy with his characters and will be keeping them. The other player wanted to try something different, so I am allowing him to make two new characters with the same XP total as the ones he is swapping out.
As I work with the player to create his new characters, I have just now realized something. When increasing ability scores upon leveling (ASI), all points are equal - the 'cost' to go from 10 to 12 is the same as to go from 18 to 20. However, for a character's initial ability scores at first level, the 'point buy' method imposes a premium on high scores - raising a score from 10 to 12 costs two points, but raising it from 13 to 15 costs four.
For a character starting at first level, perhaps the most important consideration is having as high an ability score as possible in their prime attribute - both for survivability and for the fun of being effective in their party role. Despite the cost in point buy, there is a strong incentive to have at least one score much higher than the others.
But for a character starting at 10th level, who does not have to worry about being effective at first level, the more relevant consideration is the cap of 20 on their prime ability. If you know you are getting to start at high level, the optimal strategy would be to deliberately set your initial distribution low, for maximum yield from point buy, and then 'spend' your ASI all at once on your prime score.
One issue is CR and what modules assume about the average efficacy of characters of a certain level. A more pressing concern is fairness among players - if some characters started at first and others are permitted to start at tenth, the ones who started at tenth will have higher scores, particularly in classes that have more frequent ASI (one of the proposed PCs in my case is a rogue).
If you have allowed players to start at higher levels, and have allowed this 'backward adjustment' of ability scores, has this caused problems? Either tension between players or encounter-balancing issues?
I am not interested about what is fair or what I should do. Answers along those lines are opinion-based and off-topic for this site.
I am interested in the actual experience of DMs and what sort of difficulties, if any, this has engendered.
I believe this is a general phenomenon, but at Medix2's request I will include some specific numbers. If this character was actually started at first level, suppose the player wanted a human rogue character that maximized Dex for
15 (Dex) with 3 x 12 and 2 x 11 with a 27 point buy.
After standard human +1 to all, at first level this would be
16 (Dex) with 3 x 13 and 2 x 12.
By playing to 10th level, they would have had 3 ASI events for 6 points, and might be then at
20 (Dex) with 2 x 14, 13, and 2 x 12.
If the same character was allowed to start at 10th level and 'reverse buy', they could start with
13 (Dex) with 3 x 13, 12, 11 with a 27 point buy.
After racial bonuses, at first level this would be
14 (Dex) with 3 x 14, 13, 12.
By starting at 10th level, they would have 3 ASI events for 6 points, and could be then at
20 (Dex) with 3 x 14, 13, 12.
The first-level character would start with a lower prime (in this case, Dex). By not having to actually play that character at first level, they could then have the benefits of having a character at 10th level with equal Dex but better in another ability. By focusing their ASI on lower starting scores, they would avoid the premium placed on point buy scores of >13, without having to have paid the cost of a sub-optimal first level character.