strange behaviour on minutes

Fathom this out if you can.

Here is the average minute by minute volatility of EU in the past 6 months, using the A0 feed from Esginal. A0 is the name for their aggregate feed - a combination of around 100 or so brokers and MMs.

Why is there a markedly different volatility between odd and even minutes across the hour? Is this:
a) poor data from Esignal
b) some kind of fallout from HFT
c) because people prefer to trade on even minutes rather than odd ones (ok... maybe not)

My 1 min data is from forexite (it's bid data Jan 2001 - 01 Jun 2009), below are statistics from 2001 to today from EURUSD grouped by year.

Odd column tells if the minute was odd or even

+------+------+---------------+
| year | odd | avg(high-low) |
+------+------+---------------+
| 2001 | 0 | 0.000121998 |
| 2001 | 1 | 0.000122638 |
| 2002 | 0 | 0.000104367 |
| 2002 | 1 | 0.000105268 |
| 2003 | 0 | 0.000144948 |
| 2003 | 1 | 0.000145752 |
| 2004 | 0 | 0.000130768 |
| 2004 | 1 | 0.000130797 |
| 2005 | 0 | 0.000129588 |
| 2005 | 1 | 0.000130216 |
| 2006 | 0 | 0.000110939 |
| 2006 | 1 | 0.000112221 |
| 2007 | 0 | 0.000106843 |
| 2007 | 1 | 0.000107600 |
| 2008 | 0 | 0.000264162 |
| 2008 | 1 | 0.000265770 |
| 2009 | 0 | 0.000319259 |
| 2009 | 1 | 0.000320654 |
+------+------+---------------+

And here is chart from 2009 data - it's intresting to note that the middle peak is at the same minute (32) so the data is probably about the same but there seems to be something different in esignal data.

Solid research there Mikkom. If I could produce data like that in the speed you could...

I've had my suspicions about the quality of A0 (if you look at some of the individual contributors, like Barclays, there are gaps all over the place). I suspect that they're only sampling one of the bigger contributors every 2 minutes; without going through every one it's impossible to say.

Thanks for posting