2

假设您有一些这样的数据:

AC-057|Ethanol CBOT (Pit) Liq Cont|20050329|0.121|0.123|0.121|0.123|47|233|32|219
AC-057|Ethanol CBOT (Pit) Liq Cont|20050330|0.124|0.124|0.122|0.122|68|233|0|219
AC-057|Ethanol CBOT (Pit) Liq Cont|20050331|0.123|0.123|0.123|0.123|68|246|57|226
AC-057|Ethanol CBOT (Pit) Liq Cont|20050401|0.122|0.122|0.122|0.122|5|241|5|221
AC-057|Ethanol CBOT (Pit) Liq Cont|20050404|0.12|0.12|0.12|0.12|1|240|0|220
AC-057|Ethanol CBOT (Pit) Liq Cont|20050405|0.12|0.12|0.12|0.12|5|241|0|220
AC-057|Ethanol CBOT (Pit) Liq Cont|20050406|0.12|0.12|0.12|0.12|4|241|2|220
AC-057|Ethanol CBOT (Pit) Liq Cont|20050407|0.119|0.119|0.116|0.116|30|233|23|209
AC-057|Ethanol CBOT (Pit) Liq Cont|20050408|0.115|0.115|0.115|0.115|35|217|34|194
AC-057|Ethanol CBOT (Pit) Liq Cont|20050411|0.117|0.117|0.117|0.117|5|217|0|194
AC-057|Ethanol CBOT (Pit) Liq Cont|20050412|0.117|0.117|0.117|0.117|5|217|2|194
AC-057|Ethanol CBOT (Pit) Liq Cont|20050413|0.117|0.117|0.117|0.117|9|217|0|194
AC-057|Ethanol CBOT (Pit) Liq Cont|20050414|0.117|0.117|0.117|0.117|9|217|0|194
AC-057|Ethanol CBOT (Pit) Liq Cont|20050415|0.117|0.117|0.117|0.117|9|218|4|190
AC-057|Ethanol CBOT (Pit) Liq Cont|20050418|0.117|0.117|0.117|0.117|5|218|0|190
AC-057|Ethanol CBOT (Pit) Liq Cont|20050419|0.119|0.119|0.119|0.119|5|218|5|190
AC-057|Ethanol CBOT (Pit) Liq Cont|20050420|0.119|0.119|0.119|0.119|0|218|0|190
AC-057|Ethanol CBOT (Pit) Liq Cont|20050421|0.119|0.119|0.119|0.119|5|218|0|190
AC-057|Ethanol CBOT (Pit) Liq Cont|20050422|0.119|0.119|0.119|0.119|5|223|0|190
AC-057|Ethanol CBOT (Pit) Liq Cont|20050425|0.119|0.119|0.119|0.119|0|223|0|190
AC-057|Ethanol CBOT (Pit) Liq Cont|20050426|0.119|0.119|0.119|0.119|0|223|0|190
SYMBOL|DESCRIPTION                |yyyymmdd|OPEN |HIGH |LOW  |CLOSE|.|.  |.|...

...具有各种不同的符号。

和这样的架构:

CREATE TABLE IF NOT EXISTS ma (
    id INTEGER PRIMARY KEY AUTOINCREMENT,
    symbol TEXT,
    description TEXT,
    year INTEGER,
    month INTEGER,
    day INTEGER,

    open REAL,
    high REAL,
    low  REAL,
    close REAL
);

CREATE INDEX ma_id_idx  ON ma(id);
CREATE INDEX ma_sym_idx ON ma(symbol);
CREATE INDEX ma_yea_idx ON ma(year);
CREATE INDEX ma_mon_idx ON ma(month);
CREATE INDEX ma_day_idx ON ma(day);

CREATE INDEX ma_open_idx  ON ma(open);
CREATE INDEX ma_high_idx  ON ma(high);
CREATE INDEX ma_low_idx   ON ma(low);
CREATE INDEX ma_close_idx ON ma(close);

和一个将数据导入数据库的python脚本,如下所示:

import csv
import sqlite3 as lite

__infile__  = 'ma.csv'
__outfile__ = 'ma3.db'
input = csv.reader(open(__infile__, 'rb'), delimiter='|')
conn  = lite.connect(__outfile__)

ssql = """
    PRAGMA JOURNAL_MODE = MEMORY;

"""

isql = """
    INSERT INTO ma (
        symbol,
        description,
        year,
        month,
        day,
        open,
        high,
        low,
        close
    ) VALUES (
        ?, ?, ?, ?, ?, ?, ?, ?, ?
    )
"""

conn.executescript(ssql)

for row in input:
    year  = row[2][0:4]
    month = row[2][4:6]
    day   = row[2][6:8]
    tup   = (row[0], row[1], year, month, day, row[3], row[4], row[5], row[6])
    conn.execute(isql, tup)

conn.commit()

您将如何收集一组记录来生成此架构:

CREATE TABLE trends (
    id INTEGER PRIMARY KEY AUTOINCREMENT,
    symbol TEXT,
    date DATE,
    p1 REAL,
    p20 REAL,
    p50 REAL,
    p100 REAL,
    p200 REAL
);

在该特定符号的每个日期点。

我已经尝试了很多东西。最后一个,特别是要花很长时间,所以我不知道它是否会起作用。(嗯,它不会工作,因为它需要一周的计算时间)。原始的 csv 数据现在是 250 兆,但将来它会增长到 2.5 或更多,我可能不得不使用更大的数据库。

这是我尝试过(或正在尝试)的其他东西:

ma.sql
__________________________

    CREATE TABLE symbols (
        id INTEGER PRIMARY KEY AUTOINCREMENT,
        symbol TEXT,
        UNIQUE(symbol) ON CONFLICT IGNORE
    );

    CREATE TABLE descriptions (
        id INTEGER PRIMARY KEY AUTOINCREMENT,
        description TEXT,
        UNIQUE(description) ON CONFLICT IGNORE
    );

    CREATE TABLE dates (
        id INTEGER PRIMARY KEY AUTOINCREMENT,
        entry DATE,
        UNIQUE(entry) ON CONFLICT IGNORE
    );

    CREATE VIEW trend_dates AS
        SELECT
            id AS id,
            entry AS p1,
            date(entry, '-7 day') AS p7,
            date(entry, '-14 day') AS p14,
            date(entry, '-20 day') AS p20,
            date(entry, '-50 day') AS p50,
            date(entry, '-100 day') AS p100,
            date(entry, '-200 day') AS p200, -- LEFT OFF HERE



    CREATE TRIGGER update_entry_format AFTER INSERT ON dates
    BEGIN
        UPDATE dates SET entry =
            (SELECT
                substr(entry, 1, 4) || '-' ||
                substr(entry, 5, 2) || '-' ||
                substr(entry, 7, 2)
            )
            WHERE rowid = new.rowid;
    END;

    CREATE TABLE trends (
        id INTEGER PRIMARY KEY AUTOINCREMENT,
        symbol INTEGER,
        entry INTEGER,
        p1 REAL,
        p20 REAL,
        p50 REAL,
        p100 REAL,
        p200 REAL
    );

    CREATE TABLE master (
        id INTEGER PRIMARY KEY AUTOINCREMENT,
        SYMBOL INTEGER,
        DESCRIPTION INTEGER,
        ENTRY INTEGER,
        OPEN REAL,
        HIGH REAL,
        LOW  REAL,
        CLOSE REAL,
        VOLUME INTEGER,
        OPEN_INTEREST INTEGER,
        CONTRACT_VOLUME INTEGER, 
        CONTRACT_OPEN_INTEREST INTEGER
    );

    CREATE INDEX symbols_index ON symbols(symbol);
    CREATE INDEX descriptions_index ON descriptions(description);
    CREATE INDEX dates_index ON dates(entry);

    CREATE INDEX symbols_index2 ON symbols(id, symbol);
    CREATE INDEX descriptions_index2 ON descriptions(id, description);
    CREATE INDEX dates_index2 ON dates(id, entry);

    CREATE INDEX symbols_index3 ON symbols(id);
    CREATE INDEX descriptions_index3 ON descriptions(id);
    CREATE INDEX dates_index3 ON dates(id);

    CREATE INDEX master_index ON master(
        id,
        SYMBOL,
        DESCRIPTION,
        ENTRY,
        OPEN,
        HIGH,
        LOW,
        CLOSE,
        VOLUME
    );
    CREATE INDEX master_index2 ON master(id);
    CREATE INDEX master_index3 ON master(symbol);
    CREATE INDEX master_index4 ON master(entry);
    CREATE INDEX master_index5 ON master(close);


    CREATE VIEW ma AS SELECT
        master.id,
        symbols.symbol,
        descriptions.description,
        dates.entry,
        master.OPEN,
        master.HIGH,
        master.LOW,
        master.CLOSE,
        master.VOLUME,
        master.OPEN_INTEREST,
        master.CONTRACT_VOLUME,
        master.CONTRACT_OPEN_INTEREST
    FROM master
        INNER JOIN symbols
        INNER JOIN descriptions
        INNER JOIN dates
    WHERE 
        master.SYMBOL = symbols.id AND
        master.DESCRIPTION = descriptions.id AND
        master.entry = dates.id
    ;



    CREATE TRIGGER update_master INSTEAD OF INSERT ON ma
    BEGIN
        INSERT INTO symbols(symbol) VALUES (new.SYMBOL);
        INSERT INTO descriptions(description) VALUES (new.DESCRIPTION);
        INSERT INTO dates(entry) VALUES (new.ENTRY);

        INSERT OR REPLACE INTO MASTER(
            SYMBOL,
            DESCRIPTION,
            ENTRY,
            OPEN,
            HIGH,
            LOW,
            CLOSE,
            VOLUME,
            OPEN_INTEREST,
            CONTRACT_VOLUME,
            CONTRACT_OPEN_INTEREST
        )
        VALUES(
            coalesce(
                (   SELECT id FROM symbols
                    WHERE symbol = new.SYMBOL
                ),
                    new.SYMBOL
                ),

            coalesce(
                (   SELECT id FROM descriptions
                    WHERE description = new.DESCRIPTION
                ),
                    new.DESCRIPTION
                ),

            coalesce(
                (   SELECT id FROM dates
                    WHERE entry = new.ENTRY
                ),
                    new.ENTRY
                ),

            new.OPEN,
            new.HIGH,
            new.LOW,
            new.CLOSE,
            new.VOLUME,
            new.OPEN_INTEREST,
            new.CONTRACT_VOLUME,
            new.CONTRACT_OPEN_INTEREST
        );
    END;

CREATE VIEW sma
    AS SELECT
        a.ENTRY,
        a.CLOSE,
        AVG(b.close)
    FROM
        ma AS a
        JOIN ma AS b
            ON a.ENTRY >= b.ENTRY
            AND b.ENTRY >= date(a.CLOSE, '-20 day')
        GROUP BY a.ENTRY, a.CLOSE
        ORDER BY 1
    ;

ma.py
----------------------  
import sqlite3 as lite
import csv
import glob;
print 'connecting...'
conn = lite.connect('MA.db')
infile = csv.reader(open('MA.CSV', 'rb'), delimiter='|', quotechar=r'"')
conn.execute('BEGIN TRANSACTION')
conn.execute('PRAGMA JOURNAL_MODE = MEMORY')


isql = 'insert into ma(SYMBOL, DESCRIPTION, ENTRY, OPEN, HIGH, LOW, CLOSE, VOLUME, OPEN_INTEREST, CONTRACT_VOLUME, CONTRACT_OPEN_INTEREST) values (?,?,?,?,?,?,?,?,?,?,?)'

print 'inserting data...'
for row in infile:
    conn.execute(isql, row)

conn.commit()

conn.close()

import sqlite3 as lite
conn = lite.connect('ma.db')

tsql = 'SELECT close FROM master WHERE symbol = ? AND entry = ?'
cur1 = conn.cursor()
cur2 = conn.cursor()
cur3 = conn.cursor()
cur4 = conn.cursor()
cur5 = conn.cursor()
dcur = conn.cursor()
scur = conn.cursor()

dcur.execute('SELECT id FROM dates ORDER BY entry DESC')
scur.execute('SELECT id FROM symbols ORDER BY symbol ASC')

dates = dcur.fetchall()
symbols = scur.fetchall()

print 'building trends...'
conn.execute('PRAGMA synchronous=OFF')
conn.execute('PRAGMA journal_mode=MEMORY')
conn.execute('BEGIN TRANSACTION')

while len(dates) > 0:
    for symbol in symbols:
        try:
            cur1.execute(tsql, (symbol[0], dates[0][0]))
            cur2.execute(tsql, (symbol[0], dates[20][0]))
            cur3.execute(tsql, (symbol[0], dates[50][0]))
            cur4.execute(tsql, (symbol[0], dates[100][0]))
            cur5.execute(tsql, (symbol[0], dates[200][0]))
        except Exception, e:
            print repr(e)
            pass

        try:
            p1 = cur1.fetchone()[0]
            p2 = cur2.fetchone()[0]
            p3 = cur3.fetchone()[0]
            p4 = cur4.fetchone()[0]
            p5 = cur5.fetchone()[0]
            conn.execute('INSERT INTO trends(symbol, entry, p1, p20, p50, p100, p200) VALUES(?, ?, ?, ?, ?, ?, ?)', (symbol[0], dates[0][0], p1, p2, p3, p4, p5))   
            #print "(" + repr(dates[0][0]) + ", " + repr(symbol[0]) + "): " + repr(p1) + " " + repr(p2) + " " + repr(p3) + " " + repr(p4) + " " + repr(p5)
        except Exception, e:
            #print repr(e)
            pass

    print "done: " + repr(dates[0][0])  
    dates.remove(dates[0])

conn.commit()
conn.close()

谢谢!


tl;dr:对于原始列表中的每个条目,我想使用收盘价获取每个交易品种在每个日期的 7、14、20、50、100、200 天价格。并把它放在一张桌子上。我更喜欢用纯 SQL 来做,但是 python 也可以。

4

1 回答 1

3

知道已经有一个针对滚动财务计算进行了优化的 python 库,您可能会松一口气……它被称为pandas.

我不认为pandas 会从 SQL 中读取;但是,pandas将从 csv 读取...我冒昧地使用 csv 数据(您似乎已存储在其中ma.csv)...一旦您这样做了,获得 CLOSE 的 7 天滚动平均值就像...

>>> import pandas as pn
>>> from datetime import date
>>> df = pn.read_csv('fut.csv', index_col=2, parse_dates=[2])
>>> pn.rolling_mean(df['CLOSE'], window=7)
yyyymmdd
2005-03-29         NaN
2005-03-30         NaN
2005-03-31         NaN
2005-04-01         NaN
2005-04-04         NaN
2005-04-05         NaN
2005-04-06    0.121429
2005-04-07    0.120429
2005-04-08    0.119429
2005-04-11    0.118571
2005-04-12    0.117857
2005-04-13    0.117429
2005-04-14    0.117000
2005-04-15    0.116571
2005-04-18    0.116714
2005-04-19    0.117286
2005-04-20    0.117571
2005-04-21    0.117857
2005-04-22    0.118143
2005-04-25    0.118429
2005-04-26    0.118714
>>>
>>> pn.rolling_mean(df['CLOSE'], window=7)[date(2005,4,26)]
0.11871428571428572
>>>

df上面是一个,它是一个专门的结构,用于保存与对象关联的时间索引值表...在这种情况下,保存您的 HIGH、LOW、CLOSE 等...pandas DataFrameDataFrame

除了让您的工作更轻松之外,pandas还将大部分繁重的工作卸载到 Cython,这使得运行数千次此类计算变得相当快。


fut.csv

SYMBOL,DESCRIPTION,yyyymmdd,OPEN,HIGH,LOW,CLOSE,tmp1,tmp2,tmp3,tmp4
AC-057,Ethanol CBOT (Pit) Liq Cont,20050329,0.121,0.123,0.121,0.123,47,233,32,219
AC-057,Ethanol CBOT (Pit) Liq Cont,20050330,0.124,0.124,0.122,0.122,68,233,0,219
AC-057,Ethanol CBOT (Pit) Liq Cont,20050331,0.123,0.123,0.123,0.123,68,246,57,226
AC-057,Ethanol CBOT (Pit) Liq Cont,20050401,0.122,0.122,0.122,0.122,5,241,5,221
AC-057,Ethanol CBOT (Pit) Liq Cont,20050404,0.12,0.12,0.12,0.12,1,240,0,220
AC-057,Ethanol CBOT (Pit) Liq Cont,20050405,0.12,0.12,0.12,0.12,5,241,0,220
AC-057,Ethanol CBOT (Pit) Liq Cont,20050406,0.12,0.12,0.12,0.12,4,241,2,220
AC-057,Ethanol CBOT (Pit) Liq Cont,20050407,0.119,0.119,0.116,0.116,30,233,23,209
AC-057,Ethanol CBOT (Pit) Liq Cont,20050408,0.115,0.115,0.115,0.115,35,217,34,194
AC-057,Ethanol CBOT (Pit) Liq Cont,20050411,0.117,0.117,0.117,0.117,5,217,0,194
AC-057,Ethanol CBOT (Pit) Liq Cont,20050412,0.117,0.117,0.117,0.117,5,217,2,194
AC-057,Ethanol CBOT (Pit) Liq Cont,20050413,0.117,0.117,0.117,0.117,9,217,0,194
AC-057,Ethanol CBOT (Pit) Liq Cont,20050414,0.117,0.117,0.117,0.117,9,217,0,194
AC-057,Ethanol CBOT (Pit) Liq Cont,20050415,0.117,0.117,0.117,0.117,9,218,4,190
AC-057,Ethanol CBOT (Pit) Liq Cont,20050418,0.117,0.117,0.117,0.117,5,218,0,190
AC-057,Ethanol CBOT (Pit) Liq Cont,20050419,0.119,0.119,0.119,0.119,5,218,5,190
AC-057,Ethanol CBOT (Pit) Liq Cont,20050420,0.119,0.119,0.119,0.119,0,218,0,190
AC-057,Ethanol CBOT (Pit) Liq Cont,20050421,0.119,0.119,0.119,0.119,5,218,0,190
AC-057,Ethanol CBOT (Pit) Liq Cont,20050422,0.119,0.119,0.119,0.119,5,223,0,190
AC-057,Ethanol CBOT (Pit) Liq Cont,20050425,0.119,0.119,0.119,0.119,0,223,0,190
AC-057,Ethanol CBOT (Pit) Liq Cont,20050426,0.119,0.119,0.119,0.119,0,223,0,190
于 2012-11-24T12:56:53.020 回答