Speaker impedance patterns - how to read them?


I know this has probably been discussed ad nauseam, but here's my question. I'm grateful for any sensible opinion.

Premise.
A speaker system may have a complex impedance pattern that can vary from 3-ohms to 30-ohms depending on frequency. Some speaker systems with similar impedance patterns may be stated as being 8-ohm loads because the speaker is 8 ohm in an important part of its frequency response, but only there.
A tube amp, as a voltage amplifier, likes a regular impedance pattern. Almost no speaker system has a constant, regular impedance pattern. I am aware that solid state, powerful amps are built to compensate for modern speakers' complex impedance patterns, but there are tube amps around and people who love them. It's difficult to understand what average load a given speaker system will present to the amp.

Question.
When looking for suitable speakers for a tube amp - mine is a PrimaLuna EVO 100, 40w x 2 from EL34s - what must be looked at? How to best interpret the impedance pattern, whenever available? Assuming that one listens at normal levels in a medium-sized room, what are the criteria for matching speaker and amp?

Thanks for your views. M.
martinguitars
clearthinker,

thanks, those are the criteria I am currently considering of importance; as for the Asperger affair, I am always amazed at how people feel entitled to throw judgments and epithets into the discussion because they are safe this side of the Internet... But I am no newbie to forums.

rudyb,

if the only determining factor were one's ears, audiophiles wouldn't spend a lot of time, money and mental energy changing and listening.. I find it funny that most of them buy a car, a fridge, a pair of shoes with the outmost care then rely on absolute lack of science and on hearsay to assemble a music system that is supposed to rely on science to do its job properly. Nothing wrong with some objectivity I think. Thanks for your opinion.

Max 
@Max Yes, doing our research, our reading, and maybe even some measurements, is big part of the fun of the hobby, as is exchanging knowledge, ideas and opinions on a forum like this. But while I might buy a new photo camera solely based on reviews ... amps, speakers and headphones are the kind of products I would never buy without having heard them myself first, no matter what raving professional- or user reviews or semi-scientific analysis may say.
@timlub You are mixing up the impedance of a driver for the impedance of a completed speaker, meaning 1 driver on your desktop vs. multiple drivers attached to a crossover.

Of course anything with a coil will eventually behave like an inductor, and woofer’s are the easiest of these to use to make an example. However to tell an end-user that has much to do with the end result is really not a good characterization.

Here’s a typical impedance curve for a simple 2 way:

https://speakermakersjourney.blogspot.com/2016/12/crossover-basics-impedance.html

It’s very very different than the impedance curve for the woofer it used, which can be seen in this impedance chart:

https://www.blogger.com/blog/post/edit/8271724650517523711/3931230143413830127

To say all speakers rise with frequency ignores everything that happens in between, and ignores both frequency and impedance equalization, not to mention the basic high and low pass filters as well.

@erik_squires 
Actually Erik,  no, I am not mixing up a driver vs a finished system.  I didn't say ALL speakers rise with frequency, Many drivers have a fairly smooth curve.   Here is a simple 2 way,  they show driver impedance as well as system impedance and add zobel.  Impedance compensation is the only thing that fixed these curves. You can see the curves.  
I'm sorry, but I've read through that blog that you reference, it give very basic info and doesn't show detail.  What you are saying about a driver vs that same driver within a system has minor merit, good compensation will affect the curve, but to say that a drivers individual impedance curve does not effect or control the system in the frequencies that it operates in wrong. This is not opinion, but fact.  I've done this hundreds of times. Too bad that we live so far apart,  I have an entire side of my basement shelved with drivers. We could do measurements together as well as do some system designs.  Its easy enough to show. I think that the miscommunication or (mixing up) is what a drivers curve looks like in free air vs what it looks like in its enclosure.  The curve can change quite a bit in an enclosure.  Again,  my goal here is not to argue, but to give the op some information on his inquirey.  I believe that I know of a good example.  I'll look around and if I find it,  I'll post. 
BR-1 manual.indd (daytonaudio.com)
Here you go op,  this should help.  Normally when I am building a speaker.  In design phase, I keep the crossovers external with the drivers in the box, this helps give me very accurate impedance and phase all along the design process.  
This person has done a pretty good job explaining as well as several examples that will help you get a good handle understanding. 
Speaker impedance curve explained with examples - Audio Judgement