What methods does PE use for collecting data?

[dmpokerplayer]

Do the PE servers collect data on some/most players automatically?
-or-
Do PE members have to have tables open in order for PE to collect data on those opponents?
-or-
Does the PE team ever purchase boatloads of very recent hand histories and update the PE database?

All three methods are used... PE uses their servers to collect data on "some" tables on some poker-sites... but the majority of data comes from the many PE users having tables open while they play and/or datamine.

Poker-Edge has a large farm of servers that track some cash tables directly. We're told that they can't track every table on every poker-site they support... cuz that would be thousands upon thousands of tables.

PE users track every table that they have open. If you play a set level or range of levels then it is in your best interest (mine to) to open as many of those tables as possible, as often as possible, for as long as possible. The tables can be minimized or displayed, mined or actively played... doesn't matter. PE will collect data from them with every lookup window refresh. The more tables at your level being collected on the more opponents you will have data on and the more informed your decisions will be in your future games. That's why we "datamine" (open tables we're not playing). Even if you're only playing 1 or 2 tables you can still have others open and minimized for the purpose of collecting data on more opponents. It doesn't take long to collect a lot of data on a lot of opponents when you're mining 8-16 tables at a time, for several hours each day, for a week or two straight. It would make a huge difference for you, on your poker-site, at your level, in the number of opponents you'd come across that already have data collected on them.

With PEv5 the PE team have been purchasing boatloads of very recent hand histories and updating the shared database... "bigtime". These days it's common to see 10's of thousands of hands collected on most opponents in cash games. MTT & SNG coverage is less as those types of hands are harder to come by... but still seeing a big improvement in PEv5 over PEv4.

I know one player (the g-man) that starts a 45 man (5 table) tourney on Full Tilt. Then he opens all five tables right from the start. As the tables collapse down to four, three, two, one... the new players he sees already have PE data collected on them. This type of approach sets you up for more informed decision making against opponents you definitely will see later. And you'll probably play against many of these same players in other like tourneys since you play tourneys like these... and apparently they do too.

And hey... it's a centralized database. So if we all use this approach... we'd all benefit from our own data-collecting... and from each others.

UPDATE: I did an experiment once (long term members will remember) where I datamined 16 SNG tables at once for 6-7 hours a day for six days on FullTilt. The buy-in levels from $10 - $100. By the 6th day I had substantially improved on the number of opponents I came across that already had data collected on them... and also in the amount of data on average per opponent.
- At the $10 - $20 stake levels I had data on like 65-90% of the players, in amounts from 150 - 500 hands played. These levels get a lot of new players so 65-90% was a big improvement over when I started.
- At the $30 - $100 stake levels I had data on like 80-90% of the players in amounts ranging from 200 - 2000 hands played. There are less new players at these levels so having data on 80-90% of the players was very nice... to put it mildly.

The actual experiment... Pokeredge -> Data Mining - does it help?