http://www.google.com/mail/help/figh...explained.html

Thanks, I'll give it a try. FWIW, yahoo used to be fine for the last few years, but just this last years it's turned to crap and lets lots of obvious spam through for some reason.

Juk

Yep, -X% is below expectation and +X% is above expectation.

Use the original version of my app and graph your entire playing history. If you look at the "EV luck" line only, then the biggest "swing" in this is what I'm talking about. For example: if you played 11s only, then +3000% would mean you had won $3000 more than you should have over a certain set of SNGs. If you played 33s only, then -2000% would mean you had won $6000 less than you should have, ect.

Juk

So if a graph has:
+$200 adjusted
-$250 actual
at the 6.50s

You would take the money difference first, 200 - (-250) = 450

Then 450/6 and multiply that by 10 which is 750, which makes the luck factor 750%?

Ok switching gears, I have a general question to ask. Generally speaking, if I'm playing good for say 200+ games, should my adjusted allin line keep going up? If it doesn't go up or goes down does it mean I'm probably playing bad? So then assuming this is true, the bigger the sample you had the more certain you could be that you're playing good/bad based on what the adjusted allin line is doing?

What would a winning player's lifetime graph look like, would the lines be close to even with each other? How about a losing player's graph?

Here's me being up about $19.5k total. I think it's pretty safe to say your overall line can, and should, have a negative slope. It's even more telling to filter by # of players, and it would be ideal to filter by blind level. At the low blind levels, you shouldn't be getting all-in unless you're way ahead, and they're aren't a lot of good big-pot bluff opportunities, so the only preflop all-ins you should be seeing are those where you're pretty far ahead. That makes the slope positive. Somewhere in teh mid-levels it turns around, and by the higher levels, it should be negative. You should be getting all-in preflop a lot and when you see your opponent's cards, you should be behind if you're shoving correctly.

There are other factors that influence your luck other than all-ins. Also the way that I create the adjusted line is to take the sum of all luck for a SNG and add it to the prize you won for the SNG which isn't ideal (compared to how you can do it on a "per-hand" basis for cash games), but I can't see a better way of doing it.

Overall your adjusted line should keep going up and approximate to a y=ax graph (where a is your winrate/lossrate), but there will be times when it fluctuates because of the other factors that influence variance. Also, if you are very unlucky (or lucky) it's possible that the steepness of the line gets slightly scaled because you will be not getting to apply your skill-edge in the later stage of a lot of the SNGs you are getting knocked out of (or you are getting to more bubbles in the case that you are running very lucky)

If your luck adjusted line is going up steadily over 500+ SNGs then I'd say you can be fairly sure you are a winning player, but over just 200 SNGs then I don't think you can be so confident (I often play 150 SNGs in a day and have seen it do strange stuff for a couple of days in a row...). After you have 2000+ samples then I think you can start to get a good idea of what your true ROI is. All that it's doing is telling you these things quicker than you would get them by looking at non-adjusted results, but how much quicker is not obvious.

For me personally, then after about 2000-3000 luck-adjusted SNG results I'm happy to start drawing conclusions now such as "at level X I can make ~1.x times more than at level Y", whereas before this I still wasn't happy to make decisions like this even after 10000-15000 SNGs and always had niggling doubts that I could be out by several % points of ROI.

Juk

Thanks juk that's what I was hoping it would do. I understand it's a really broad view of things, but hey it's better than nothing. Now if we can only get something for ftp! =)

The link doesn't work anymore ...

I didn't realize you were doing ti that way. Why not just spit out the per-hand result like the original program? The "how much money you should have won" is a good result for people's personal satisfaction, but it doesn't really say anything about your game.

I'm not 100% sure I know what you mean?

I do still output both "per-hand" and "per-sng" files in cvs format, but the luck-adjusted "per-hand" version is no better really as then I have to just keep your current cumulative amount won constant throughout all of the hands of each SNG (you can see the "steps" if you run it on a small number of SNGs).

The reason I think the "per-sng" version is better to plot is because you don't get artificial "scaling" based on the length of a SNG (ie: the x-axis will get condensed for periods where you bust out of lots of SNGs early and magnified when you run really well and play lots of hands in each...) and also it's alot easier to get graphing apps to open a file with several thousand data-points compared to several hundred thousand.

The results printed at the end aren't effected as they are just one huge cumulative sum over all the hands/games.

Juk

juk, I'm just trying to understand the difference between your program and the original one. I think I've got it now.

For data visualization, you could probably do some useful stuff with the per-hand data by plotting it as a surface, maybe making your mesh be "hands" by "prize pool equity" and plotting the occurrences in each?

I don't really understand. For example, concerning 11s only.
+3000% of 10$ is 300$, isn't it?

I'm trying to calculate my biggest swing.
If for example it is -150$ on 6.5$ SnG, what is the swing?
(-150 / 6 * 100) About -2500%?

Thanks.

The % figure is in % of total prize pool afaik. So -150 / (6*9) * 100 = -278%

Hi,

I tried installing version 0.1.6, but when I try to run it
I get the error message: This Version has timed out.

Can someone please give me hand installing this thing?

Thanks,
Jonathan

It doesn't work anymore. You can try jukofyork's program if you play 9-max on Party or Stars (with P-->S hand converter). (Right?)

So Bodypull no longer releases any beta version?
Nor a commercial release?

What happened? Too much of headache?

Yep, it's "% of total prize pool" so for example if you playing 10-man $11 SNGs: 1%=1$, 10-man $22s: 1%=$2, etc.

Juk

Yep - you can also run it on 6-man (5-man?) and HU if you pass in the prize-share %s on the command line (upto 3 prizes paid) - see the readme.txt file and the example batch files for an idea of how to do this.

Juk

Is there any chance that the software works in a soon future please ?

jukofyork is it possible for you to make your program work also for Full Tilt or it takes a lot of time?

Gnumsploitz and Excedrin came up with a ftp converter, but I'm having problems with some hhs. I think it only works with 6.50s and 12s right now.

When I try to run it, it says could not find hour of day a lot and then it just skips the hh. I still was able to do about 1264 hhs out of 1600 total, but I still have problems with my latest sets. If someone wants to take a look and try to fix it that'd be cool.

Here's a rar of my hhs from my last set where I get the error: http://www.megaupload.com/?d=R15LV92I

Here's a link to the ftp converter: http://rapidshare.com/files/11168013...nvert.rar.html

Just extract those files to juk's directory and I'm sure you can figure out the rest.

Frozinite edited the ahk converter file and now it works on more hhs without as many errors. Here is a rar with the updated file.

http://rs168l3.rapidshare.com/cgi-bi...15356515744210

I can't seem to get hold of this, but could somebody send me a copy (or post the whole script here). Rapidshare rarely works for me because I think my ISP uses a transparent proxy server and it doesn't like it.

If it works like dave_w11's Stars one then I should be able to compile it and create some batch files like the Stars version (does it take command line arguments?).

Juk

Sorry, rapidshare sucks.

http://jimmylikesmen.com/ahk/juk.ftp.converter.zip

Some batch files are already in there. (Sorry, I only created some for a few buyins to test it out.) You shouldn't need to compile it to make it work.

Fair warning: you're going to get some errors. But, it works well enough in its present form.

Btw, I don't see why it wouldn't work with any buyin. However, you will get errors in situations where the sb/bb is eliminated. Excedrin was taking a look at that, but I'm pretty sure those spots don't convert over to the Party format properly yet. (Read: juk's program will return errors thinking that players are acting out of turn, etc.)

Most credit is due to dave_w11 still. I just got this started by using exe2ahk and starting to modify his starsconverter code to work with ftp2party instead. I'm not too great with AHK, so when I got stuck, Excedrin pretty much bailed me out entirely. I wouldn't have come even remotely close to making this work as well as it does without his help.