Financial commentary has a structural problem: it is produced on a schedule regardless of whether anything happened, and it is rewarded for confidence rather than for being right. Both pressures push in the same direction, and resisting them is mostly a matter of rules.
Say where it came from
Every figure should be attributable. Not a vague gesture at data, but a specific enough source that a reader could go and check. This single rule eliminates most of the bad content in market writing, because a claim that cannot be sourced usually cannot be made.
It also constrains the writer usefully. Having to name a source for a number makes you notice when you do not have one.
Say when it was measured
A number without a time is not a fact. Market readings age, sometimes in minutes, and prose that implies currency for a figure measured this morning is misleading even when every digit is correct.
Distinguish what happened from what it means
The report and the interpretation should be separable by a reader who wants only the first. Blending them produces writing where disagreeing with the analysis requires disentangling it from the facts, which is work the reader should not have to do.
Correct in public
Errors happen. The obligation is that a correction is at least as visible as the original and that the original is not quietly edited to match. An archive that always agrees with what the desk currently believes is not a record.
And do not predict
A desk that reports structure, history and what is being said has plenty to write about without forecasting. The moment it starts predicting, every other claim it makes inherits the credibility of its predictions, which over a long enough run is not a good trade.
None of it is financial advice, and a desk that needs that line to be load-bearing has already gone wrong.