Merge two queries that use different timeframes for min/max/avg - mysql

Question

I am using 10.1.39-MariaDB - mariadb.org binary and I have the following table:
| id | date | ticker | close |
|-------|---------------------|--------|-----------|
| 39869 | 2019-09-18 00:00:00 | AAPL | 221.96160 |
| 39870 | 2019-09-17 00:00:00 | AAPL | 220.70000 |
| 39871 | 2019-09-16 00:00:00 | AAPL | 219.90000 |
| 39872 | 2019-09-13 00:00:00 | AAPL | 218.75000 |
| 39873 | 2019-09-12 00:00:00 | AAPL | 223.09000 |
| 39874 | 2019-09-11 00:00:00 | AAPL | 223.59000 |
| 39875 | 2019-09-10 00:00:00 | AAPL | 216.70000 |
I have two queries where I am calculating metrics for 52-weeks and the second query calculates aggregation metrics for 20-days:
52-weeks:
SELECT
Y.*
FROM
(
SELECT
MAX(CLOSE) AS week_52_High,
DATE AS week_52_High_date,
MIN(CLOSE) AS week_52_Low,
DATE AS week_52_Low_date,
AVG(CLOSE) AS week_52_Avg
FROM
`prices`
WHERE
DATE >= DATE(NOW()) - INTERVAL 52 WEEK AND ticker = "AAPL") Y
LEFT JOIN prices tmax52 ON
tmax52.date = Y.week_52_High_date AND tmax52.close = week_52_High
LEFT JOIN prices tmin52 ON
tmin52.date = Y.week_52_Low_date AND tmin52.close = week_52_Low
LEFT JOIN prices tavg52 ON
tavg52.close = week_52_Avg
20-days
SELECT
Y.*
FROM
(
SELECT
MAX(CLOSE) AS day_20_High,
DATE AS day_20_High_date,
MIN(CLOSE) AS day_20_Low,
DATE AS day_20_Low_date,
AVG(CLOSE) AS day_20_Avg
FROM
`prices`
WHERE
DATE >= DATE(NOW()) - INTERVAL 20 DAY AND ticker = "AAPL") Y
LEFT JOIN prices tmax20 ON
tmax20.date = Y.day_20_High_date AND tmax20.close = day_20_High
LEFT JOIN prices tmin20 ON
tmin20.date = Y.day_20_Low_date AND tmin20.close = day_20_Low
LEFT JOIN prices tavg20 ON
tavg20.close = day_20_Avg
Both queries calculate the min/max/avg for each close price and attach the days, when this happened.
Any suggestions how to merge these two queries to get everything in 1 output?
I appreciate your replies!

Answer 1

Your first problem is that your query won't actually work. The correct way to get the dates for the high and low values is as below. Note that it is meaningless to try and get a date for the average close as it's highly unlikely that the stock will have closed at that price.
SELECT
Y.day_20_High,
tmax20.DATE AS day_20_High_date,
Y.day_20_Low,
tmin20.DATE AS day_20_Low_date,
Y.day_20_Avg
FROM
(
SELECT
MAX(CLOSE) AS day_20_High,
MIN(CLOSE) AS day_20_Low,
ROUND(AVG(CLOSE),2) AS day_20_Avg
FROM
`prices`
WHERE
DATE >= CURDATE() - INTERVAL 20 DAY AND ticker = "AAPL") Y
LEFT JOIN prices tmax20 ON tmax20.close = Y.day_20_High
LEFT JOIN prices tmin20 ON tmin20.close = Y.day_20_Low
Output (for my sample data)
day_20_High day_20_High_date day_20_Low day_20_Low_date day_20_Avg
107.50 2019-09-20 101.10 2019-09-10 104.05
Demo on dbfiddle
Having corrected the query, you can now just JOIN to the same query for 52-week data:
SELECT
Y20.day_20_High,
tmax20.DATE AS day_20_High_date,
Y20.day_20_Low,
tmin20.DATE AS day_20_Low_date,
Y20.day_20_Avg,
Y52.week_52_High,
tmax52.DATE AS week_52_High_date,
Y52.week_52_Low,
tmin52.DATE AS week_52_Low_date,
Y52.week_52_Avg
FROM ((
SELECT
MAX(CLOSE) AS day_20_High,
MIN(CLOSE) AS day_20_Low,
ROUND(AVG(CLOSE),2) AS day_20_Avg
FROM
`prices`
WHERE
DATE >= CURDATE() - INTERVAL 20 DAY AND ticker = "AAPL") Y20
LEFT JOIN prices tmax20 ON tmax20.close = Y20.day_20_High
LEFT JOIN prices tmin20 ON tmin20.close = Y20.day_20_Low)
JOIN ((
SELECT
MAX(CLOSE) AS week_52_High,
MIN(CLOSE) AS week_52_Low,
ROUND(AVG(CLOSE),2) AS week_52_Avg
FROM
`prices`
WHERE
DATE >= DATE(NOW()) - INTERVAL 52 WEEK AND ticker = "AAPL") Y52
LEFT JOIN prices tmax52 ON tmax52.close = Y52.week_52_High
LEFT JOIN prices tmin52 ON tmin52.close = Y52.week_52_Low)
Output (for my sample data)
day_20_High day_20_High_date day_20_Low day_20_Low_date day_20_Avg week_52_High week_52_High_date week_52_Low week_52_Low_date week_52_Avg
107.50 2019-09-20 101.10 2019-09-10 104.05 109.70 2019-08-24 100.00 2019-08-21 104.19
Demo on dbfiddle

Answer 2

If you want to merge your queries one after another:
with t as (
select 0 as id, 1 as v0, 2 as v1
union all select 1 as id, 1 as v0, 2 as v1
)
select *
from (
select t.*, 'weekly_report' as typ from t
union all
select t.*, 'monthly_report' as typ from t
) as t1
id | v0 | v1 | typ
-: | -: | -: | :-------------
0 | 1 | 2 | weekly_report
1 | 1 | 2 | weekly_report
0 | 1 | 2 | monthly_report
1 | 1 | 2 | monthly_report
db<>fiddle here
And to have your results column by column, do a join:
with t as (
select 0 as id, 1 as v0, 2 as v1
union all select 1 as id, 1 as v0, 2 as v1
)
select t0.id, t0.v0, t0.v1, t1.v0, t1.v1
from (
select * from t
) as t0
full join (
select * from t
) as t1 on t0.id = t1.id
id | v0 | v1 | v0 | v1
-: | -: | -: | -: | -:
0 | 1 | 2 | 1 | 2
1 | 1 | 2 | 1 | 2
db<>fiddle here