graphing winrate (not net win/loss)

#1 10-22-2005, 12:52 PM

I'm looking for a program that will plot first and second derivatives of my net win/loss. Poker Grapher and PokerPatterns both only plot the net win/loss and offer no functionality for more advanced analysis. Any suggestions?

olavfo · #2 10-22-2005, 02:05 PM

A profit graph is not continuous (it consists of discrete points where profit is plotted vs number of hands or hours). Thus, calculating its derivative(s) at different points does not make sense mathematically.

#3 10-22-2005, 02:41 PM

There are numerical ways. It's pretty damn close to continuous.

It is certainly possible to graph a very good approximation of the first derivitave (the winrate) by hand, it would just be very tedious. Hence the need for software.

#4 10-22-2005, 02:59 PM

Here's an even better question. Does anyone know how I could export my net winnings as a discrete time function in a text format? This is from either PT or any other software package that interfaces with the PT Access databases. If I had this data in a readable format I could do all the analysis myself with Matlab.

TheIrishThug · #5 10-22-2005, 04:37 PM

poker patterns can graph BB/100. u have to go a few thousand hands before it stops fluctuating too much. not sure if this is really what u want.

#6 10-23-2005, 12:05 AM

PP apears to graph average BB/100, not instantaneous BB/100. I *want* fluctuation. I want to see when my BB/100 goes negative for a specific session (running bad) and when it skyrockets (running good) etc. I don't really care what my average BB/100 is (PT tells me this), I just want to know what it is over the last few sessions, and compare that to my history. Ideally I'd like to see a smooth increase in my BB/100 which would become apparent after many many hands (enough to "zoom out" on the daily variance)

Nomad84 · #7 10-23-2005, 04:39 AM

Search for an old post by Vern from before PokerGrapher. He listed the query to use in Access to produce the data used to generate the graphs. You can then copy this into Excel and numerically differentiate there. That would probably be easier than using Matlab, but then again, I've only used Matlab a few times and hated it [img]/images/graemlins/mad.gif[/img] For the projects I do, I much prefer Mathcad. It's much more user friendly, IMO, but I can see how that would depend on your application. FWIW, I would probably try numerically differentiating over a range of quite a few hands rather than trying to get an instantaneous rate, since the instantaneous rate really won't produce any meaningful information (I think). Something like 50-100 hands would probably be a decent number. Something along the lines of:

B51=(A101-A1)/10000

That gets your average BB/100 over a 100 hand range centered at A51. I know that if you use Matlab, you can surely understand what I mean, but I thought a brief description might be useful for some.