Stats tracking software wishlist?

[ugomonster]

Imagine you have a stats tracking software with backtracking. So every time players shows his own cards the stats tracker looks at his actions through the hand and logs how the opponent played in this specific conditions. Basically the same stuff that all of us do when we are not multitabling .

So, what stats are the most important things we need to know about opponent at the microlimits? For simplicity we can assume that there is not strong player-to-player history at microlimits, which means player will play the same game against opponent1 and opponent2 in the same situation if they have the same stats.

Here is a list/tree I came up with so far.

Position Aware (some abstract value that indicates "position awareness" of the player)

Good Player (some abstract value, based on PositionAware and other PF/AF stats). Instead of tracking "opponent awareness" for the player we can calculate how good he is based on other stats, and assume that good players are opponent-aware and bad players are not.


All other stats are percent-based and have multiple layers that we can use to "slice" the stat. For example for a stat "Checked when had lead" we would like to get a sliced view by betting round, by number of opponents and by position.

BPS is short for "board progress state (betting round)" (Flop, Turn, River)
In parenthesis for each stat I put the "slice layers" that could be interesting for this stat

<font class="small">Code:</font><hr /><pre>


Aggressive stats % (BPS)
Check (BPS)
After Having Lead (Position, BPS, Opponents)
Does Not Have Lead And Did Not Improve
Slow played very strong hand (BPS)
Bet (BPS)
After Having Lead (Position, BPS, Opponents)
Bluff (Position, BPS, Opponents)
Scare Board
Everyone checked previous round
Into PF Raiser (BPS, Position)
Ragged Flop
AK High Flop
Top Pair Less Then Q
Donk Bet (Does Not Have Lead And Did Not Improve this round)
Value bet with weak hands (BPS, Position, Opponents)
Raise (BPS)
After Having Lead (Position, BPS, Opponents)
Bluff (Position, BPS, Opponents)
PF Raiser with TopPair less then Q
Strong draw (Position, BPS, Opponents)
Check-Raise (BPS)
Check-Call (BPS)
Check-Fold (BPS)
Bet-Fold (BPS)
Bet-Call (BPS)
Raise-Fold (BPS)

Passive stats %
Called (BPS)
With a draw when did not have odds (BPS)
With a weak pair against PF raiser (BPS)
AK board
Ragged board
Slow played very strong hand (BPS)
Slow played strong hand (BPS) (raise on the next round)
Scare-play strong hand (BPS) (call on the next round)

Went To Showdown %
Won At Showdown %

</pre><hr />

Another question - are there any know correlations between some stats? For example if players after flop aggression factor is 2.5 and his average bet bluffing is 10% and his average raise bluffing is 1% - what is the % that he would raise PF raiser on the ragged board with nothing?


There are so many projects around that gather hand history from 100s of players. This kind of info should exist somewhere... The question is if anyone published it yet or kept for personal use

[jukofyork]

This is quite an interesting idea, but I think you would need a heck of a lot of data from each player to get such detailed stats to be meaningful.

If you could find some way to use virtually all of your mined data as a whole, and then cluster player types and their associated "typical" plays it might be more feasible.

Juk [img]/images/graemlins/smile.gif[/img]

[i]EDIT: "Donk Bet (Does Not Have Lead And Did Not Improve this round)" - "WinSD% when donk [by stage]" would be pretty cool stat to have, as some fish tend to only donk their monsters and other fish tend to do totally irrational donks - knowing what type of donk a fish like to do could be pretty helpful IMO. I guess the same could be said for limp-reraises; some fish only LRR their monsters, yet others LRR total junk - again this would be helpful to know.