MySQL Query:

    (select w.id 
     from
        wands w
        inner join
        wands_property wt
        on w.code = wt.code
     where w.power = wa.power 
        and wt.age = wp.age 
        and w.coins_needed = min(wa.coins_needed)
     limit 1
    ) wand_id,
    wp.age wand_age,
    min(wa.coins_needed) wand_price,
    wa.power wand_power
from
    wands wa
    inner join
    wands_property wp
    on wa.code = wp.code
where wp.is_evil = 0
group by wa.power, wp.age
order by wa.power desc, wp.age desc;

Exaplanation:

This query worked in MySQL because MySQL is lenient and allows the use of aggregate functions like MIN() inside the WHERE clause of a subquery, even though this is not strictly valid in standard SQL. In proper ANSI SQL, aggregate functions can only appear in the SELECT or HAVING clauses or within subqueries that include a GROUP BY. MySQL implicitly treats the MIN() as an aggregation over the entire subquery, which is why it doesn't throw an error, but this behavior is not portable to stricter SQL databases like PostgreSQL, SQL Server, or Oracle, where such a query would fail. To ensure strict correctness and portability, the subquery should either use GROUP BY to compute the minimum or, more efficiently, apply window functions like ROW_NUMBER() to select the minimum value per group. So while your logic and result were correct for the immediate context, the technical application of MIN() in the WHERE clause is not universally valid SQL.