Please bore me with the math
It is part of my overall magic item economy fix, but it started with spellbooks so it works as a stand-alone. I have different rules for high-magic (which I consider 5e magic item
pricing), medium-magic, and low-magic (I consider 5e magic item assumed
frequency to be medium-low) parts of my multiverse. In general I assume something akin to the default 'magic item trade like rare art trade', but there are places where that isn't true (my party just bought 4
bags of holding in Sigil--from two different shops, and each bag was unique, but it was easily doable). However, the rules for spellbooks are the same everywhere, because it works better that way.
In 5e it costs half as much (50 gp/spell level) to copy a spell into your spellbook as it did in 3e (100 gp/spell level).
3e has a suggested price for permission to copy a spell from someone's spellbook (you'd still have to pay to write it into yours) of 50 gp/spell level. So I decided I'd also half that value and say the standard fee is 25 gp/spell level for a spell merchant to let you copy a spell from one of their spellbooks (in their well guarded copying room under the watchful eyes of their golem or such).
So now I need a price scheme for buying a spellbook as an alternative to copying the spells a la carte, that makes sense for both the buyer and the seller.
From the PHB we know the cost to write a spell you already know into a spellbook: 10 gp/spell level. A blank spellbook costs 50 gp (which the spell merchant passes on to the buyer, of course). So we can see what it would cost for the spell merchant to create any particular spellbook (assuming there is no limit to how many spells you can fit in it, not a good assumption, but multiple books wouldn't be that big of a deal).
The spell merchant isn't likely to to sell spells in a spellbook for much less than he'd sell permission to copy them ala carte. So let's set the lower price for a spellbook at the same 25 gp/spell level (+50 gp for the book). For the high end of the pricing, I decided a simple doubling would be a good idea. The 50 gp/per spell level is the same as what it costs to learn/scribe a spell, and the overall symmetry is appealing, so I went with that. The value changes based on how valuable that particular book is to the buyer and seller. I average those numbers for 37.5 gp/spell level as the average--then apply a random modifier to account for specific value to the individual. Also, even treating spellbooks as quasi-commodities, the spell merchant is going to buy low and sell high. Note that you are likely buying some spells you already know in the book, which is part of the reason you aren't paying 50 gp/level for it. A specifically commissioned book (or one that the seller knows the buyer can get full use out of) probably does cost 50 gp/level.
This pricing scheme also allows both buyer and seller to examine the relative pros and cons of buying a book versus permission to copy a spell. I specifically designed these prices so that there is no uniform overwhelming advantage to either method. The level of the spells, the books available, the inconvenience of the copying to the seller, the value retention of the book even after copied--all of these can go into the decision of whether to buy ala carte or to buy a book with some spells you don't need, as well as which the merchant prefer to sell you.
So, since the wizard can get full use out of this book I made a mistake in my previous price by assuming the average "blind sale" value. Assuming the rogue knows the standard pricing (which the OP isn't assuming), since he knows the wizard has full use for this precise book, the price should be 5,500 gp. (Actually, 5,350 gp, but why not round up?)
Of course, if the wizard can bluff that he has a backup copy, he can bluff that it wouldn't possibly be worth much more than 1,200 gp, since he could make a copy for less than that.
But the real question is why is a wizard able to cast 7th level spells
negotiating with a magically ignorant thief?