Trouble using group by to get a max value across two tables - mysql

Question

I have been trying to solve a problem for a very long time- days- and I am not making any progress. Basically, I have two tables, players and matches. Each player in players has a unique player_id, as well as a group_id that identifies which group he/she belongs to. Each match in matches has the player_ids of two players in it, first_player and second_player, who are always from the same group. first_score corresponds to the score that first_player scores and second_score corresponds to the score that second_player scores. A match is won by who ever scores more. Here are the two tables:
create table players (
player_id integer not null unique,
group_id integer not null
);
create table matches (
match_id integer not null unique,
first_player integer not null,
second_player integer not null,
first_score integer not null,
second_score integer not null
);
Now what I am trying to do is to get the players with the most wins from each group, their group ID as well as the number of wins. So, for example, if there are three groups, the result would be something like:
Group Player Wins
1 24 23
2 13 25
3 34 20
Here's what I have right now
SELECT p1.group_id AS Group, p1.player_id AS Player, COUNT(*) AS Wins
FROM players p1, matches m1
WHERE (m1.first_player = p1.player_id AND m1.first_score > m1.second_score)
OR (m1.second_player = p1.player_id AND m1.second_score > m1.first_score)
GROUP BY p1.group_id
HAVING COUNT(*) >= (
SELECT COUNT(*)
FROM players p2, matches m2
WHERE p2.group_id = p1.group_id AND
((m2.first_player = p2.player_id AND m2.first_score > m2.second_score)
OR (m2.second_player = p2.player_id AND m2.second_score > m2.first_score))
)
My idea is to only select players whose wins are greater than, or equal to, the wins of all other players in his group. There is some syntactic problem with my query. I think I am using GROUP BY incorrectly as well.
There is also the issue of a tie in the number of wins, where I should just get the player with the least player_id. But I haven't even gotten to that point yet. I would really appreciate your help, thanks!
EDIT 1
I have a few sample data that I am running my query against.
SELECT * FROM players gives me this:
Player_ID Group_ID
100 1
200 1
300 1
400 2
500 2
600 3
700 3
SELECT * FROM matches gives me this:
match_id first_player second_player first_score second_score
1 100 200 10 20
2 200 300 30 20
3 400 500 30 10
4 500 400 20 20
5 600 700 20 10
So, the query should return:
Group Player Wins
1 200 2
2 400 1
3 600 1
Running the query as is returns the following error:
ERROR: column "p1.player_id" must appear in the GROUP BY clause or be used in an aggregate function
Now I understand that I have to specify player_id in the GROUP BY clause if I want to use it in the SELECT (or HAVING) statement, but I do not wish to group by player ID, only by the group ID.
Even if I do add p1.player_id to GROUP BY in my outer query, I get...the correct answer actually. But I am a bit confused. Doesn't Group By aggregate the table according to that column? Logically speaking, I only want to group by p1.group_id.
Also, if I were to have multiple players in a group with the highest number of wins, how can I just keep the one with the lowest player_id?
Edit 2
If I change the matches table to such that for Group 1, there are two players with 1 win each, the query result omits Group 1 from the result altogether.
So, if my matches table is:
match_id first_player second_player first_score second_score
1 100 200 10 20
2 200 300 10* 20
3 400 500 30 10
4 500 400 20 20
5 600 700 20 10
I would expect the result to be
Group Player Wins
1 200 1
1 300 1
2 400 1
3 600 1
However, I get the following:
Group Player Wins
2 400 1
3 600 1
Note that the desired result is
Group Player Wins
1 200 1
2 400 1
3 600 1
Since I wish to only take the player with the least player_id in the case of a draw.

Answer 1

WITH first_players AS (
SELECT group_id,player_id,SUM(first_score) AS scores FROM players p LEFT JOIN matches m ON p.player_id=m.first_player GROUP BY group_id,player_id
),
second_players AS (
SELECT group_id,player_id,SUM(second_score) AS scores FROM players p LEFT JOIN matches m ON p.player_id=m.second_player GROUP BY group_id,player_id
),
all_players AS (
WITH al AS (
SELECT group_id, player_id, scores FROM first_players
UNION ALL
SELECT group_id, player_id, scores FROM second_players
)
SELECT group_id, player_id,COALESCE(SUM(scores),0) AS scores FROM al GROUP BY group_id, player_id
),
players_rank AS (
SELECT *,
ROW_NUMBER() OVER(PARTITION BY group_id ORDER BY scores DESC, player_id ASC) AS score_rank,
ROW_NUMBER() OVER(PARTITION BY scores ORDER BY player_id ASC) AS id_rank FROM all_players ORDER BY group_id
)
SELECT group_id, player_id AS winner_id FROM players_rank WHERE score_rank=1 AND id_rank=1
Results
group_id winner_id
1 45
2 20
3 40
Try it Out

Answer 2

try like below
with cte as
(
select p.Group_ID,t1.winplayer,t1.numberofwin
row_number()over(partition by p.Group_ID order by t1.numberofwin desc,t1.winplayer) rn from players p join
(
SELECT count(*) as numberofwin,
case when first_score >second_score then first_player
else second_player end as winplayer
FROM matches group by case when first_score >second_score then first_player
else second_player end
) t1 on p.Player_ID =t1.winplayer
) select * from cte where rn=1

Answer 3

It works when you add the player_id in the GROUP BY because you know each player plays only in one group. So you group by the player in a certain group. That is why, logically, you can add the player_id to the GROUP BY.