(Please refer to SQLFiddle for a working example of this post)
I have a table with stock information, as follows:
sp100_id _date bullishness agreement
----------------------------------------------
1 2011-03-16 1.01 0.33
1 2011-03-17 0.85 1.28
1 2011-03-18 0.89 1.25
1 2011-03-21 1.46 1.21
1 2011-03-22 0.39 2.53
2 2011-03-16 3.07 1.27
2 2011-03-17 2.09 0.80
2 2011-03-18 0.91 0.12
2 2011-03-21 1.50 0.00
2 2011-03-22 2.62 1.10
3 2011-03-16 0.73 1.13
3 2011-03-17 1.13 1.21
3 2011-03-18 1.12 0.45
3 2011-03-21 1.00 1.01
3 2011-03-22 1.00 0.53
4 2011-03-16 0.40 1.10
4 2011-03-17 2.40 0.03
4 2011-03-18 3.16 0.10
4 2011-03-21 0.86 0.50
4 2011-03-22 1.00 0.10
I need to order the companies (sp100_id) by their averge bullishness over time into a top-3:
SELECT
sp100_id,
AVG(bullishness) as bullishness,
AVG(agreement) AS agreement
FROM stocks
WHERE _date BETWEEN '2011-03-16' AND '2011-03-22'
GROUP BY sp100_id LIMIT 3
This works fine, as the result is
SP100_ID BULLISHNESS AGREEMENT
2 2.038 0.658
4 1.564 0.366
3 0.996 0.866
Now that I have the top-3, I need the top-3 to be re-ordered by AGREEMENT, ascending:
SP100_ID BULLISHNESS AGREEMENT
4 1.564 0.366
2 2.038 0.658
3 0.996 0.866
Is this possible to do with one query? I tried the following but it didn’t work. It still only orders by bullishness
SELECT
sp100_id,
AVG(bullishness) as bullishness,
AVG(agreement) AS agreement
FROM stocks
WHERE _date BETWEEN '2011-03-16' AND '2011-03-22'
GROUP BY sp100_id
ORDER BY bullishness DESC, agreement ASC LIMIT 3
So to be clear: (1) I need to find the top-3 companies with highest average bullsihness (2) this top-3 then needs to be ordered from lowest to highest agreement. Preferably with one query. Do you know how?
It’s called structured query language because you can build structures in which queries (aka virtual tables) are nested inside other queries.
Take your first query, which is correct except it needs its own ORDER BY clause, and nest it in another, like so.
Go fiddle: http://sqlfiddle.com/#!2/c9ff0/7/0