Selecting max price from every 5 rows - mysql

Question

I've been trying and trying and haven't been able to figure this out.
In stock charts that have Open, High, Low, Close, you can always chart every minute, 5 min, 10 min, hour, etc. I have data for every minute and I'm trying to select out the Open, High, Low, Close from that minute-by-minute data, but for every 5 minutes.
I have data similar to this:
__________________________________________________
| Date | TIME | TICKER | Open | High | Low | Close |
| 20121203 | 090000 | QQQQ | 23.54 | 24.12 | 23.01 | 23.24 |
| 20121203 | 090100 | QQQQ | 23.24 | 24.14 | 22.98 | 24.13 |
| 20121203 | 090200 | QQQQ | 24.13 | 25.88 | 23.75 | 25.81 |
| 20121203 | 090300 | QQQQ | 25.81 | 25.83 | 24.63 | 24.99 |
| 20121203 | 090400 | QQQQ | 24.99 | 25.21 | 23.89 | 24.12 |
| 20121203 | 090500 | QQQQ | 24.12 | 24.19 | 21.94 | 22.03 |
| 20121203 | 090600 | QQQQ | 22.03 | 22.97 | 20.68 | 21.44 |
| 20121203 | 090700 | QQQQ | 21.44 | 24.06 | 19.32 | 23.56 |
| 20121203 | 090800 | QQQQ | 23.56 | 25.48 | 23.07 | 25.01 |
| 20121203 | 090900 | QQQQ | 25.01 | 28.00 | 24.18 | 27.21 |
| 20121203 | 091000 | QQQQ | 27.21 | 27.55 | 24.31 | 24.31 |
I need to grab the max(high) for the rows that have a time >= 090000 (that's 9 a.m.) 09 hours 00 minutes 00 seconds.
Similar to this, I need min(low), and then I'll grab the close price when time is 090400 because I'm getting every 5 minutes. I could also use the open of the next 5 minute increment, so that's flexible.
I've used nested SELECT statements, multiple joins, etc. The problem is that the MySQL execution time is about 1 second per row returned. That's crazy slow when you figure there are 12 rows per hour (60 minutes / 5 minutes = 12), and then because I'm actually doing FOREX, the trading is around the clock, so 24 hours. That gives me 288 per day, or just under 5 minutes per day. To do 1 year of data (~ 250 trading days) would be about 20 hours. There has to be a faster way.
I've got some solutions for this with the ID being continuous, and though that might be easiest, I'm not 100% sure my data would be correct in doing that. The reason why is that on Fridays the trading day ends at normal business hours in NY and opens up with the first trading in Tokyo (about mid-afternoon in the United States).
I've looked at GROUP BY, but I'm not sure how I can group the data to get a group of 5 where the time is within 5 minutes of each 5 minute group.
Thanks for your thoughts and discussion.
Jarod

Answer 1

This should show max(high) and min(low) for every 5 minutes
SELECT Max(high),
Min(low)
FROM tbl
GROUP BY ROUND(Unix_timestamp(Date(Concat(`date`, `time`))) / ( 5 * 60 ))
In the Group by clause we concat your date time column. So it forms something like 20121203090000. This is one the format that is recognized as date in mysql. So we pass it to date() function. Then its converted to UNIX_TIMESTAMP. Its divided by 5 mins timespan. The result will be a float value. But we require same value for a specific time span. Hence the ROUND(). It makes the floating value to the nearest integer. To understand how its working run this query.
SELECT high,
low
Unix_timestamp(Date(Concat(`date`, `time`))) / ( 5 * 60 ) `5-min span`
ROUND(Unix_timestamp(Date(Concat(`date`, `time`))) / ( 5 * 60 )) `5-min span rounded`
FROM tbl

Answer 2

try this:
SELECT CONCAT(DATE,SUBSTRING(Time,1,2),"["
,IF(SUBSTRING(Time,4,1)<5,CONCAT(SUBSTRING(Time,3,1),"0"),CONCAT(SUBSTRING(Time,3,1),"5"))
,"-"
,IF(SUBSTRING(Time,4,1)<5,CONCAT(SUBSTRING(Time,3,1),"5"),CONCAT(SUBSTRING(Time,3,1)+1,"0"))
,"]") AS timeStr ,MAX(High) ,MIN(LOW) FROM tb1 GROUP BY timeStr;
http://sqlfiddle.com/#!2/6b748/1