How to calculate running sum based on conditions in Microsoft SQL server - sql-server

Question

I have the data as below.
stock
datetime
AVG_Price
MF
NetMF
ABCD
2022-12-06 09:15:00
234.50
237864
237864
ABCD
2022-12-06 09:16:00
236.55
357784
595648
ABCD
2022-12-06 09:17:00
233.23
334959
260689
ABCD
2022-12-06 09:18:00
233.23
498756
759445
ABCD
2022-12-06 09:19:00
225.23
456789
302656
ABCD
2022-12-06 09:20:00
222.23
678546
-375890
ABCD
2022-12-06 09:21:00
222.23
567483
-943373
ABCD
2022-12-06 09:22:00
220.23
67483
-1010856
The desired output in NetMF
I want to calculate Net MF using below conditions.
At datetime 9:15, I will return MF value into Net MF
From 9:16 onwards below is the logic i want to follow
If AVG_Price at 9:16 > AVG_Price at 9:15 then Net MF at 9:16 = Net MF at 9:15 + MF at 9:16
If AVG_Price at 9:16 < AVG_Price at 9:15 then Net MF at 9:16 = Net MF at 9:15 - MF at 9:16
If AVG_Price at 9:16 = AVG_Price at 9:15 and Net MF at 9:15 < 0 then Net MF at 9:16 = Net MF at 9:15 - MF at 9:16
If AVG_Price at 9:16 = AVG_Price at 9:15 and Net MF at 9:15 >= 0 then Net MF at 9:16 = Net MF at 9:15 + MF at 9:16
This logic continues till the end of the time period.
I need this logic to work for Multiple stock codes and multiple datetimes.
Currently I have below code in place. This code returns only values for first 1 or 2 rows
WITH v_table_name AS
(
SELECT stock_code,[datetime], [AVG_Price],[MF], CASE WHEN cast(datetime as time) = '09:15:00.000' and close_price >= open_price THEN [MF]
WHEN cast(datetime as time) = '09:15:00.000' and close_price < open_price THEN -1*[MF]
WHEN Avg_Price > LAG(Avg_Price,1,-1) over (Partition by stock_code order by datetime asc) THEN (LAG([Net MF],1,-1) over (Partition by stock_code order by datetime asc)) + [MF]
WHEN Avg_Price < LAG(Avg_Price,1,-1) over (Partition by stock_code order by datetime asc) THEN (LAG([Net MF],1,-1) over (Partition by stock_code order by datetime asc)) - [MF]
WHEN Avg_Price = LAG(Avg_Price,1,-1) over (Partition by stock_code order by datetime asc) and (LAG([Net MF],1,-1) over (Partition by stock_code order by datetime asc)) < 0 THEN (LAG([Net MF],1,-1) over (Partition by stock_code order by datetime asc)) - [MF]
WHEN Avg_Price = LAG(Avg_Price,1,-1) over (Partition by stock_code order by datetime asc) and (LAG([Net MF],1,-1) over (Partition by stock_code order by datetime asc)) > 0 THEN (LAG([Net MF],1,-1) over (Partition by stock_code order by datetime asc)) + [MF]
END as [Net MF] from Equity
)
UPDATE [Equity] set Equity.[Net MF] = v_table_name.[Net MF]
FROM v_table_name
WHERE Equity.stock_code = v_table_name.stock_code
and Equity.datetime = v_table_name.datetime

Answer 1

As the Net MF depends on previous row value, one simple solution is to use a recursive CTE and calculate the Net MF in the recursive part of the CTE based on your required condition
with
cte as
(
select rn = row_number() over (partition by stock_code order by [datetime]),
stock_code, [datetime], AVG_Price, [MF]
from Equity
),
rcte as
(
-- anchor
select c.rn, c.stock_code, c.[datetime], c.AVG_Price, c.[MF], [Net MF] = c.[MF]
from cte c
where c.rn = 1
union all
-- recursive
select c.rn, c.stock_code, c.[datetime], c.AVG_Price, c.[MF],
[Net MF] = case when c.AVG_Price > r.AVG_Price
then r.[Net MF] + c.[MF]
when c.AVG_Price < r.AVG_Price
then r.[Net MF] - c.[MF]
when c.AVG_Price = r.AVG_Price
and r.[Net MF] < 0
then r.[Net MF] - c.[MF]
when c.AVG_Price = r.AVG_Price
and r.[Net MF] >= 0
then r.[Net MF] + c.[MF]
end
from cte c
inner join rcte r on c.stock_code = r.stock_code
and c.rn = r.rn + 1
)
update e
set [Net MF] = r.[Net MF]
from rcte r
inner join Equity e on r.stock_code = e.stock_code
and r.[datetime] = e.[datetime]
OPTION (MAXRECURSION 0);
Note : open/close price logic is not in as it is not clear for me. You may add it in yourself
db<>fiddle demo