Your Edge

[Lyric]

Has anyone calculated their edge in poker?

To do this, one would need to calculate total amount won and divide by total amount VPIP. PT does not include total dollars VPIP.

Has anyone determined a method for calculating their edge using PT?

[N 82 50 24]

It seems to me like it should be just $PIP, not $VPIP. The blinds are part of your investment to play poker... right? It would seem like your edge is total $ returned over total money invested, including forced bets. Or am I thinking about this wrong?

[Lyric]

[ QUOTE ]
It seems to me like it should be just $PIP, not $VPIP. The blinds are part of your investment to play poker... right? It would seem like your edge is total $ returned over total money invested, including forced bets. Or am I thinking about this wrong?

[/ QUOTE ]

It would be nice to see numbers for overall edge AND for edge on every dollar voluntarily contributed.

[schnoodleC]

To get a very approximate ROI take hands at a given game * ave pot for that game / number of player / total winning = ROI. Or something like that, you get the picture.

[Lyric]

[ QUOTE ]
To get a very approximate ROI take hands at a given game * ave pot for that game / number of player / total winning = ROI. Or something like that, you get the picture.

[/ QUOTE ]

This assumes that you are contributing an avg amount to each pot, but yeah -- a nice little rough estimate.

[_dave_]

I think this does it - not quite sure though, not tested thorougly. Execute this SQL against your PT database:

------------------------ code ---------------------------
SELECT
screen_name, COUNT(game_id) AS HANDS,
SUM(total_bet) AS BET, SUM(total_won) AS WON,
ROUND(((SUM(total_won)/SUM(total_bet))-1)*100,<font color="red">3</font>) as EDGE
FROM players, game_players
WHERE players.player_id = game_players.player_id
AND screen_name='<font color="blue">your_screenname_here</font>'
GROUP BY screen_name;
--------------------------------------------------------

Alter '<font color="blue">your_screenname_here</font>' to suit.

This should give you the "edge" of the specified screenname, rounded to 3 decimal places (change the <font color="red">3</font> for more/less precision), across all levels in the PT database you run this on, for the screen name (and aliases to it) you specify.

My first HSNL post [img]/images/graemlins/smile.gif[/img] Hopefully useful to someone... or maybe I misunderstood what you are asking for?

FWIW in the quick test I ran on an old Access DB, I had a 2.77% edge over $25nl last January, which sounds within reason I think.


dave.

Edited to add: "and aliases"

[greg nice]

if you are using amount won, then how is this any different than measuring your edge with your winrate?

[_dave_]

[ QUOTE ]
if you are using amount won, then how is this any different than measuring your edge with your winrate?

[/ QUOTE ]

You mean (pt)BB/100, right?

Initially, I just read this:
[ QUOTE ]
It seems to me like it should be just $PIP, not $VPIP. The blinds are part of your investment to play poker... right? It would seem like your edge is total $ returned over total money invested, including forced bets. Or am I thinking about this wrong?

[/ QUOTE ]

And thought - PT can do this, let's try some SQL. I came up with the above. Then I saw your post, and thought - Have I just done something completely pointless? Is this not just winrate? I am still not sure....

I added some code to my original statement to filter levels, and ran more tests on my more recent PostgreSQL database, incedentally, it seems PGSQL doesn't like the ROUND() function, so I removed it.

I ended up with this:

------------------------- code ----------------

SELECT
screen_name, COUNT(game_players.game_id) AS HANDS,
SUM(total_bet) AS BET, SUM(total_won) AS WON,
((SUM(total_won)/SUM(total_bet))-1)*100 as EDGE,
SUM(total_won)-SUM(total_bet) as PROFIT

FROM players, game_players, game, game_level
WHERE players.player_id = game_players.player_id
AND game_players.game_id = game.game_id
AND game.game_level_id = game_level.game_level_id
AND game_level.game_level_desc = '<font color="red">NL $50</font>'
AND screen_name='<font color="blue">your_screenname_here</font>'
GROUP BY screen_name;

--------------------------------------------------


I added PROFIT, which matches up with PT stats, so I guess it works OK, but EDGE seems to bear little relation to winrate (ptBB/100) as displayed by Poker Tracker.

Example:

NL $50 - 8965 hands, 3.82ptBB/100, EDGE 1.345%
PL $50 6m - 46 hands, 3.15ptBB/100, EDGE 1.857%

Very small sample size for the PL $50 I know, but these two are the two levels that have a similar ptBB/100 in my database. The thing is here - NL50 better winrate but a higher EDGE.

I am unfortunately none the wiser as to what this newfound stat describes, but I know it isn't the same as ptBB/100.

I am thinking it has something to do with the amount of varience in a player's game, so it will take many hands to become at all valuable? or maybe it is greatly affected by uncalled overbets?

Perhaps this is a totally useless stat? Maybe someone who knows more about numbers can enlighten us as to the value of this calculation.


dave.

[cero_z]

Hi Lyric,

I just use the following simple formula to determine my edge:

Edge = avg. edge of an expert + a bajillion%

[Lyric]

[ QUOTE ]
if you are using amount won, then how is this any different than measuring your edge with your winrate?

[/ QUOTE ]

Winrate and edge are different.

BB/100 (winrate)caclulates how much money you're dragging in without concern for the amount of money you put at risk.

Players can have the same winrate but have different edges.

We'll use a one hand example to illustrate. Imagine you move all in for $1,000 and win the $15 blinds. Your winrate is 75PTBB/100((15x100)/20). Your edge is 1.5% (15/1000)

Another player raises to $35 and wins the same blinds. His winrate is still 75PTBB/100, but his edge is 43%(15/35).