I've written this solution as an aggregation because aggregations are likely to be what you want to do more often, so will show you how to set it up.
disclaimer: I don't have your data so I am doing this without testing.
tradesDB = MongoGetCollection[db, "trades"]
create an aggregation in which you project the two fields that you are wanting
aggregate = MongoCollectionAggregate[tradesDB, {
<|"$project" -> <|"_id" -> 0, "t" -> 1, "size" -> 1|>|>}]
This will return a cursor object (note that MongoCollectionFind[tradesDB,<||>,
<|"_id" -> 0, "t" -> 1, "size" -> 1|>]
will do this also but I wanted to illustrate aggregations)
To return some data to check what you have you can do something like this:
cursor = {};
Do[cursor = Flatten[{cursor, MongoCursorNext[aggregate]}], 5]
cursor
this should return a list of associations. You can retrieve all of the data (from the cursor point) straight into a Dataset
Dataset[aggregate]
If the above works for you perhaps you can post some screen grabs so that future readers can see the output. If it doesn't work ping me and maybe email your data for me to try.
Edit
To show this working I've implemented the above with a dummy set of data.

and here it is as a Dataset

In the next line I've done an aggregation to find the average population of the cities in the database with populations greater than 25000 from CA or NY to give an idea of building an aggregation using Mathematica and MongoLink. Hopefully it illustrates the power of using mongo from within Mathematica.

Edit 2
By the way if you wanted to use MongoCollectionFind
the syntax is MongoCollectionFind[collection,query,projection]
. If you simply want to project and not query then by analogy with mongo shell where you enter {} for the query, enter an empty association as above.