Here's another thought. Before I start, I need to stress that this is
not a personal criticism of
@Phil V. I hope he knows me well enough to appreciate that, but other people might not.
Anyway...
To qualify my knowledge level, larger spreadsheets, linked across networks, some vba, form building, embedded HTML, application building, a bit of statistical reporting with dashboards, graphics etc.
Obviously it depends on the context, but trying to be too ambitions with Excel can be a very bad idea. The problem is maintainability: checking that a spreadsheet is correct and is giving the right answers can be incredibly difficult; and the more of these "advanced" features you use, the harder it is to check. It can be very difficult even for the person who created the spreadsheet, and much more difficult for anyone else. (And remember, "anyone else" includes future versions of you. If you build a complex Excel model now, it will be hard for you to maintain it in the future.)
My wife is into spreadsheet modelling. She currently works for one of the major electricity generators, modelling the potential revenue streams arising from investments in new generating capacity. These days most of the action is in small-scale plants which can be switched on and off rapidly to plug the gaps when the sun goes in or the wind stops blowing, and as you can probably imagine the modelling is incredibly complex. It has to take account of supply scenarios, demand scenarios, and National Grid's contractual mechanisms relating to availability and utilisation. Anyway she recently inherited responsibility for a big asset valuation model which has been in use for some years, but unlike some of her predecessors she's not comfortable using a model unless she's been able to verify that it's working correctly. So she dug around its innards and after a couple of weeks of work she had a long long list of outdated assumptions, incorrect references, redundant formulae, and so on. She reckons that if she were to rebuild it from scratch, it would be 90% smaller and run 100 times faster. All because the original builders seem to have got carried away with building a complex model and don't seem to have designed it to be maintainable.
In a similar vein, In a former life I used to do modelling and analysis in the rail industry, and I once discovered a spreadsheet error with 9-figure price tag (>£100,000,000) on it. Seriously. I worked on railway franchise bids, and it's well publicised that the Department for Transport got its fingers badly burned by faulty analysis which they couldn't spot was faulty. They eventually solved it by dictating that any spreadsheet model used in a bid had to have a very rigidly enforced structure: strict separation of inputs, calculations and outputs; every column in a spreadsheet having exactly the same formulae in each row; indirect and referencing functions banned; VBA banned; linking across sheets banned; that sort of thing. It forced us to plan the structure of a spreadsheet model carefully before starting to build it, but that's not a bad thing. Advanced spreadsheet modelling is really no different from software engineering, but the vast majority of spreadsheet users don't treat it like that.
