Speakers (including subwoofers) make more difference for the sound than anything else. My advice is to spend as little as possible on the receiver, buying a good brand (e.g., Yamaha, Pioneer, Marantz, Denon), and spend all the rest on speakers.
Of course, you will want to make sure you have all the features you need in the receiver, but features do not generally give you better sound.
I was running a $600 receiver with speakers that retailed for over $6k, and it sounded great. I replaced the receiver with one that retails for about $1700, and unless I engage a feature that affects the sound, it sounds the same as the $600 receiver.
On the other hand, with lesser speakers, I would hear a difference, so putting money into speakers makes a lot of sense if you care about actual sound quality.
Now, if my speakers were inefficient or a difficult load, then it might have made a difference in sound, as the new receiver can put out about twice as much power as the old one, but the old one could drive my speakers to levels that I found painful with crystal clarity, so I have no need for the extra power.
In general, if you end up selecting difficult to drive speakers, you will want to go with a separate power amp and a receiver with preamp outputs instead of trying to get a powerful enough receiver to drive them. But if you don't get difficult to drive speakers, paying for extra power is a waste of money.
My advice is to go out into the world and listen to as many brands of speakers, and to as many types of speakers (e.g., ribbons, domes, horns, whatever) as you can stand to listen to. Different people value different things in sound, and so they have wildly different preferences from each other. Some people love horns, and others hate them. I personally tend to like ribbon speakers, but that is irrelevant to you, as what matters for you is whatever produces sound that you like, and you won't know what that is without going out into the world and auditioning a bunch of speakers.
For a subwoofer, there are a number of good brands, but I would go with
SVS (and have done so).