A new poker analysis site that i've been working on

[pokergrader]

The X-axis is just hands played, from 1-> however many hands you put in. The reason a winning player should have a graph with a positive slope is that if you are playing well, IE getting the money in when you are expected to be ahead, then your potential profit should be positive in most hands you play, so with enough data the graph should be somewhat of a straightish line upwards.

Bluffing is just calculated if you are betting or raising with a less than 25% expected chance of winning the hand. Currently it is counted as a success if you win the pot at any time, but I'm probably going to tweak that to only be successful if you take the pot down right there (It wont be that big a difference in the numbers).

[Bubu]

I'm still confused. If each x-coordiante represents a hand where you willfully played then if you play +ev hands each of these datapoints should lie above the x-axis (ie be profitable). However, it would seem to me that each hand's ev should be independent from the others - if hand #352 is AA, why would hand #353 have higher ev ?

It seems to me that a correct preflop choices should have a graph which is mostly positive - with the slope carrying no information.

(Actually, there is a weaker correlation between hand number and ev in that position will increas the ev of any hand - however, this is repeats every ten hands and so at most should be a sawtooth pattern)

Does the algortihm take into account what other people were holding if there is a showdown ?

The profit part of the program is what I like best. I think its sweet to have a means of seeing id one was making +ev decisions preflop. Helps one keep sane.

[pokergrader]

[ QUOTE ]
I'm still confused. If each x-coordiante represents a hand where you willfully played then if you play +ev hands each of these datapoints should lie above the x-axis (ie be profitable). However, it would seem to me that each hand's ev should be independent from the others - if hand #352 is AA, why would hand #353 have higher ev ?

It seems to me that a correct preflop choices should have a graph which is mostly positive - with the slope carrying no information.

(Actually, there is a weaker correlation between hand number and ev in that position will increas the ev of any hand - however, this is repeats every ten hands and so at most should be a sawtooth pattern)

Does the algortihm take into account what other people were holding if there is a showdown ?

The profit part of the program is what I like best. I think its sweet to have a means of seeing id one was making +ev decisions preflop. Helps one keep sane.

[/ QUOTE ]

A little clarification.

The X-Axis is hands played, and the Y-Axis is Potential Profit. It is not just a hand that you played PF, it is all hands. So if you fold PF not in a blind, your potential profit for that hand will be $0, because you didn't put any money into the pot. Since the graph is showing the Potential Profit of your total session, and not each hand, each data point naturally builds off the previous one, so if you have an overall positive slope for the graph, then you are average + potential profit on each hand, which is good.

Also, the Potential Profit is not only pre-flop, but every aspect of the hand where money went into the middle (flop, turn, river,

[Bubu]

Why plot it as cumulative ev ? I understand that at the end of the day you want to give the "total profit". But I think it might be more useful to plot the derivative. I think this would make moments when you lose control (tilt) more evident. Specially if they happen towards the end of a long session, when the overall scale of the graph might blur it out.

I guess, the derivative plots too erratically to convey any info... so by integrating you are smotthing the function. Maybe it would be nice to plot the unintegrated fcn for only the hands that you play.

B

[pokergrader]

[ QUOTE ]
Why plot it as cumulative ev ? I understand that at the end of the day you want to give the "total profit". But I think it might be more useful to plot the derivative. I think this would make moments when you lose control (tilt) more evident. Specially if they happen towards the end of a long session, when the overall scale of the graph might blur it out.

I guess, the derivative plots too erratically to convey any info... so by integrating you are smotthing the function. Maybe it would be nice to plot the unintegrated fcn for only the hands that you play.

B

[/ QUOTE ]

I think it would be far too erratic to be relevant. The problem is that the potential profit is very closely related to the pot size, and the pot size of hands you are involved in has some pretty large variance.

Also, since potential profit isn't exactly perfect. For example it is hard to get a positive potential profit if you aren't getting any good cards, by smoothing it out you can see the overall picture a little better, so only long downturns will be noticable, which are more easily categorized as bad play.

[pokergrader]

A nice update today, which among other things:

- Auto-detects usernames, no need to type them in anymore
- Added VP$IP and Aggression stats for your session
- Superimposed both graphs on the same grid which gives it more resolution and makes it easier to compare the two
- Made all the luck statistics vary per $, resulting in accuracy that not only measures how luck or unlucky each situation was, but also whether they came during big pots or small pots

[Mr Mojo Risin]

I wish I was able to do something like you. You have created something amazing. And you keep improving it. I can not thank you enough.

[pokergrader]

New update:

- Added a few more stats including your rakeback statistics for your session (using your share of the overall rake instead of your individual rake).
- Added a leaderboard to see exactly who has the best/worst luck

[Mr Mojo Risin]

Does this work with SNGs?

[pokergrader]

[ QUOTE ]
Does this work with SNGs?

[/ QUOTE ]

SNGs work for PokerStars, but still does not do Party+Skins SNGs correctly. There is a blinds issue which always screws up the numbers, and I am working on it.