One thing to add is, that some tournaments don’t use a fair handicap. You would sometimes need 7 handicap stones to have a 50 % chance of winning, but you only get two stones. Then it would be expected that you can’t keep something of your advantage until the endgame.
To know whether you need to get better at handicap games specifically, depends on whether you are actually doing worse than expected in those games. Are you winning just 30 % of the games, but your expected winrate is 50 %? Are you winning 20 % of your games, but your expected winrate is just 10 %? If you don’t know the answer, it might be that you can just try to improve your go skills generally, and you don’t have to focus on handicap games.
I think nowadays 9 stones handicap would be appropriate for a competitive game between a 1k EGF and a high ranking pro, while 6-7 stones would be a good handicap for a teaching game between them.
The smaller handicap in teaching games would allow the pro to play more relaxed with proper and educational moves and still win by a small margin of their choosing.
I (3d EGF) have lost several games with 4 or 5 stones handicap against top ranking amateurs, so I think I would need 6-7 stones handicap for a competitive game against a high ranking pro.
“Secret chronicle on handicap go” is a small book in which a pro reach to beat a pro (both same level) with 9 stones handicap.
Because that pro decided to play very conservative black as an experimentation (and failed)
Conclusion: there are no shortcut and the mental is crucial.
Another interesting book on handicap, covering from 9 stones to 2 stones very forcefully
guiding on the spirit to have and the strategies implied is “the breakthrough to shodan”. Quite the book to read the day you decide to give more interest in handicap go.
These 2 books may be better fitted for strong sdks (6k to 1k)
Even further:
“Le jeu a 9 pierres” by maitre Lim (in french only) one if not the one most complete book on 9 stones handicap
Handicap dictionaries in various editions from Asian countries.