SQL Server 2012 的平均缩减和中位数

Question

我有一个 table 'TEMP' 列 'Date'、'Ticker' 和 'Price' :

Date       Ticker   price
01/01/13    ABC     100.00 
01/02/13    ABC     101.50 
01/03/13    ABC      99.80 
01/04/13    ABC      95.50 
01/05/13    ABC      78.00 
01/01/13    JKL      34.57 
01/02/13    JKL      33.99 
01/03/13    JKL      31.85 
01/04/13    JKL      30.11 
01/05/13    JKL      35.00 
01/01/13    XYZ      11.50 
01/02/13    XYZ      12.10 
01/03/13    XYZ      17.15 
01/04/13    XYZ      14.10 
01/05/13    XYZ      15.55

我计算了每日最高价格和回撤：

   select t.Date,
          t.Ticker,
          t.price, 
       max(t.price) over (partition by ticker order by date) as max_price,
       (t.price / max(t.price) over (partition by ticker order by date)) - 1 as Drawdown
  FROM [dbo].[temp] t;

Date    Ticker   price  max_price   Drawdown
01/01/13    ABC  100.00  100.00     0.000
01/02/13    ABC  101.50  101.50     0.000
01/03/13    ABC   99.80  101.50    -0.017
01/04/13    ABC   95.50  101.50    -0.059
01/05/13    ABC   78.00  101.50    -0.232
01/01/13    JKL   34.57   34.57     0.000
01/02/13    JKL   33.99   34.57    -0.017
01/03/13    JKL   31.85   34.57    -0.079
01/04/13    JKL   30.11   34.57    -0.129
01/05/13    JKL   35.00   35.00     0.000
01/01/13    XYZ   11.50   11.50     0.000
01/02/13    XYZ   12.10   12.10     0.000
01/03/13    XYZ   17.15   17.15     0.000
01/04/13    XYZ   14.10   17.15    -0.178
01/05/13    XYZ   15.55   17.15    -0.093

如何计算每日平均回撤和中值回撤？

我可以在 Excel 中计算它，但很难使用 SQL 代码来生成此输出：

Date       Average  Median
01/01/13     0.000   0.000
01/02/13    -0.006   0.000
01/03/13    -0.032  -0.017
01/04/13    -0.122  -0.129
01/05/13    -0.108  -0.093

Answer 1

您可以使用 PERCENTILE_CONT 函数轻松找到中位数。这是一个示例查询，我已将您的查询用作子查询

declare @t table (
    Date date
    , Ticker varchar(10)
    , price decimal(10, 2)
)

insert into @t values
('20130101', 'ABC', '100.00')
,('20130102', 'ABC', '101.50')
,('20130103', 'ABC', '99.80')
,('20130104', 'ABC', '95.50')
,('20130105', 'ABC', '78.00')
,('20130101', 'JKL', '34.57')
,('20130102', 'JKL', '33.99')
,('20130103', 'JKL', '31.85')
,('20130104', 'JKL', '30.11')
,('20130105', 'JKL', '35.00')
,('20130101', 'XYZ', '11.50')
,('20130102', 'XYZ', '12.10')
,('20130103', 'XYZ', '17.15')
,('20130104', 'XYZ', '14.10')
,('20130105', 'XYZ', '15.55')

select
    *, Average = cast(avg(Drawdown) over (partition by date) as decimal(10, 3))
    , Median = cast(percentile_cont(.5) within group (order by Drawdown) over (partition by date) as decimal(10, 3))
from (
    select 
        t.Date,
        t.Ticker,
        t.price, 
        max(t.price) over (partition by ticker order by date) as max_price,
        (t.price / max(t.price) over (partition by ticker order by date)) - 1 as Drawdown
    FROM @t t
) t

SQL Server 2012 的平均缩减和中位数

Drawdown AVERAGE and MEDIAN in SQL Server 2012

average

median

sql-server-2012