Question

I have a table with the following fields: id, opp1, opp2, opp1_votes, opp2_votes.

If one of the opps have more votes than another so we consider that he won. How can I get opp with most wins from this table?

Answer 1

This query will find the number of wins for each user, and order them in a descending order (so the one with most wins will be the first row):

SELECT winner, COUNT(*)
FROM
    (
    SELECT opp1 AS winner FROM table
    WHERE opp1_votes > opp2_votes
    UNION ALL
    SELECT opp2 AS winner FROM table
    WHERE opp2_votes > opp1_votes
    )
GROUP BY winner
ORDER BY COUNT(*) DESC

Explanation of the query:

1. First, we find all the rows for which opp1 was the winner. We generate a table where the opp1 winners are listed under the field name “winner”.
2. We do the same as in (1), but now for cases where opp2 was the winner.
3. We UNION the two table and receive a list of all winners in all games.
4. Now, we group the resulting table by winner ID, and select the count out from this grouping, resulting in a list of winners, and for each one the number of times he/she won.
5. The last step is ordering by by the number of wins, descending.

Notes:

• According to the query above, if there’s a tie – there’s no winner. If you want to treat ties as wins for one of the users, you’ll need to define a tie-breaking heuristic.
• If you really only want a single result, you can start with SELECT TOP 1 winner, ... or finish with LIMIT 1, but as far as I know it will not save computation time in the general case, so I’ll leave it.