There's a limit to how good you can get when the scale of what you've done consists entirely of programs you can accomplish in a day. To get beyond that, you need to work on stuff that's bigger - programs that take years to build, and that take hundreds of engineers, and that explore every nook and cranny of their problem domain.

In the process, you learn about managing large codebases, and refactoring, and scaling, and minimizing complexity, and deleting code, and tons of algorithms and data structures and problem-solving approaches, and communicating with a team, and being patient, and carrying on when you're not sure what the end product you're building will look like or if it'll be any good, and all sorts of other skills. You don't get that by practicing textbook algorithms and language syntax for 10 years.