Obviously there's a way to do it and a way not to do it - see the "How much does a slave cost" question. There's one answer that is based entirely on speculation and historical antecedent ("I'm gonna guess from Roman prices..."), and then there's one that is based on not perfectly canonical but much closer sources (3pp for the same game, etc.). The first one is +8/-11 and the second is +17/-0.
There's not a binary answer here. To the degree to which your answer is not backed up in a compelling way, it'll draw downvotes. To the degree to which it is, it'll get upvotes.
This means that when you answer a question, you should stop and think - how far are you stretching? Are you ignoring closer data? (On the Tressym question, if there was an answer in 5e material and someone answered for Pathfinder, I'd expect downvotes and even delete votes. Since there appears to be not, a Pathfinder analogue is moderately well received at +2/-1.) If you're drawing analogies from really far away ("historical Rome!") you should expect deserved downvotes, while if you're super close you shouldn't.
Back It Up! isn't binary. Answers that completely don't back it up may get deleted, post-noticed, etc. But ones that back it up poorly will get more down than up votes, and those that do it well will get more up than down. The System Works (tm).
Remember - "there is no answer" is OK if there is no answer. Asking for a rule for example, "There's no rule but here's the house rule I've used and here's how it worked out" is the limit as far afield as we want to get - "here's what I'd make up if faced with this" is over the line.
So if you want to answer the Tressym question with reasoning from Roman slavery or from exotic pet stores in your neighborhood assuming a Tressym is like a wombat and a gold piece is like $1 - keep it zipped. If you want to reason it as "exotic familiars in related systems cost about 200 gp" - go for it. Between those, you pays your money, you takes your chances.