-- https://stackoverflow.com/questions/53416531/clickhouse-moving-average

DROP TABLE IF EXISTS bm;

CREATE TABLE bm (amount float, business_dttm DateTime) engine Log;

INSERT INTO bm VALUES (0.3,'2018-11-19 13:00:00'), (0.3,'2018-11-19 13:05:00'), (0.4,'2018-11-19 13:10:00'), (0.5,'2018-11-19 13:15:00'), (0.6,'2018-11-19 13:20:00'), (0.7,'2018-11-19 13:25:00'), (0.8,'2018-11-19 13:30:00'), (0.9,'2018-11-19 13:45:00'), (0.5,'2018-11-19 13:50:00');

WITH
    (
        SELECT arrayCumSum(groupArray(amount))
        FROM
        (
            SELECT
                amount
            FROM bm
            ORDER BY business_dttm
        )
    ) AS arr,
    1 + rowNumberInAllBlocks() AS id,
    3 AS window_size
SELECT
    amount,
    business_dttm,
    if(id < window_size, NULL, round(arr[id] - arr[id - window_size], 4)) AS moving_sum
FROM
(
    SELECT
        amount,
        business_dttm
    FROM bm
    ORDER BY business_dttm
) ORDER BY business_dttm;

DROP TABLE bm;