Elephant Grows Up: Planned Features And Support – Interview Part II

"Elephant should become THE poker software per se."In yesterday's interview, our I.T. guys talked about the growth spurts of Elephant. Today, you'll hear from galam, plexiq, eniki and khalthag about features which are currently being developed, where to get support and how to participate in the Elephant project.

PokerStrategy: Hello dear friends! In yesterday's interview, we questioned 6Wishmaster6, Arngrim, Sape and Iarwe about Elephant. Please introduce yourselves and your area of responsibility!

galam: Hello! I am Sebastian, 24 years old and a working student at PokerStrategy.com. Apart from this, I study business informatics. In general and especially for Elephant, I'm responsible for the English and German poker software. Apart from the forum work, I also coordinate the questions of foreign-language communities via the corresponding supporters. If anybody has a question, he'll probably get a reply from me.
I am also responsible for the software tests and the bug tracking in new versions. So if a bug made it into a new version, it simply means that I didn't dig deep enough. 😉
Furthermore, I also discuss new features with the Elephant team, write task booklets and pass on questions to the team.

plexiq: Hi, my name is Helmuth Melcher, I am 27 years old and have been working full-time for PokerStrategy's education department for about half a year. In the forum, I use the nickname "plexiq". The calculation of the Nash Ranges in the ICM Trainer is mainly on my head.
I've only worked directly on Elephant for a few weeks when I wrote the hand history parser for Mansion Poker and Everest Poker. Before that, I developed modules which are supposed to be integrated into Elephant later in time.

eniki: Hello! My name is Holger Tag, and I am 36 years old. At Pokerstrategy, I work as a software engineer with my primary emphasis being on database design and -application, momentarily, of course, mainly for Elephant. This also includes the parser for the hand histories which becomes better with each reported bug detection. Many thanks at this point to galam, who always provides me with information from the community, and especially to all the users who have helped with their posts!

khalthag:
Hello, I am Paul Tolksdorf, am 29 years old and currently work on the HUD (live
overlay of statistics at the poker table) and the Optical Character
Recognition (OCR). Generally speaking, you could say that I am responsible for the direct readout of player data. I have also worked on other areas of Elephant though, e.g. on different analysis - and enumeration routines for Hold'em games.

 

PokerStrategy: How do I get support from the Elephant team as a member?

galam: The best way to receive support is via the forum. In our preferred communication tool, the Elephant area, you have the option to voice questions, requests or criticism. Many questions are already answered via the Search feature and the FAQ. For serious problems you can of course get in touch with us immediately via Skype/ICQ. The corresponding data can, of course, also be found in the forum.

PokerStrategy: How do you like working in the forum? How is the atmosphere there?

galam: Basically, I have a lot of fun helping others. Usually, the atmosphere in the forum is also very productive. Unfortunately, there are also always people who occasionally cross the line. That makes it very difficult for me to help them, as they mostly move away from objectivity.

 

PokerStrategy: What is currently being criticised the most, and what do you want to do about it?

galam: With a product with such a big target group, it's not possible to make every one happy. Elephant is still in its Beta phase and only has had a relatively short development time. Most people thus miss the features they are familiar with which haven't been included yet. As long as the installations aren't too big, we are trying to realize everything in a relatively timely manner.
Big projects like the SnG support require elaborate planning and extensive testing though. Thus they take longer. That we get criticized for this is understandable from the member's point of view. At times it's also advantageous to make us work faster as we always get our attention drawn to the priorities.

PokerStrategy: So what are the members most frequent problems?

galam: Many problems actually don't appear with Elephant, but with the included database management system PostgreSQL. Unfortunately, the installation of a database system is not really easy. So a lot of problems could show up whose origin lies elsewhere, but in the end fall back onto Elephant. In the future, we are definitely going to automatize this even more though to enable the users to have a problem-free installation.

 

PokerStrategy: What possibilities do the PokerStrategists have to participate in the development?

galam: Currently we are still integrating basic features for which we have concrete perceptions. When this section has been completed, we want to give every PokerStrategist the possibility to include her-/himself in the further development. Respective suggestions will be discussed with the members before the actual execution. This will probably be done via the forum and eventually also via special developer chats. More details about this in due time.

 

PokerStrategy: Helmuth,
you mentioned modules you've developed which are supposed to be integrated into  Elephant at a later point in time, and are already more than basic features. What kind of modules are these?

plexiq: For example one involves the "Future Game Simulation" (FGS). This is an extended equity evaluation for SnG's. Like the ICM Trainer it evaluates a player's share of the tournament prize pool.
ICM only uses the payout structure and the stack sizes for the evaluation of the equity though. Factors like the blind level, the absolute position and relative position to other players are not taken into account. This results in some clear mistakes in some situations.
A known example is e.g. a short stack in early position. The short stack should push further in almost every situation than suggested by the ICM, as it does not consider that the player is "forced to do something" before the blinds hit him.
The FGS includes factors like blind level and positions into the evaluation and is thus able to deliver better results in many of these situations than the ICM.

 

PokerStrategy: How will SnG players profit more from this?

plexiq: For SnG players, the FGS is able to significantly reduce the learning effort. Players invest a lot of time on training and review to develop a "feeling" for the correct ranges under the ICM.
Only when the player has already developed a good understanding of the ICM, is he also able to understand the most important limitations of the model, and thus deviate his playing style from the "ICM-optimal" way in the necessary situations. This is an elaborate and difficult process, especially when the player has no pointers regarding the size of the ICM deviations.
The FGS now offers the option to immediately learn many of these ICM deviations. The player has the possibility to directly practice with a "corrected" version of the ICM. Of course the FGS evaluations aren't perfect either. A player who strictly plays according to the FGS evaluations will definitely have a considerable edge over players who play strictly according to the ICM though.

PokerStrategy: Are there further interesting projects which will find their way into Elephant?

plexiq: The next project will revolve around the evaluation of values like VP$IP, PFR
etc., like e.g. how often they will be used in HUD's. This idea hasn't been implemented in a piece of commercial poker software yet even though suggestions for an optimized evaluation of these parameters have been around for some time, like e.g. in Chen's and Ankeman's Mathematics of Poker.
To clarify the basic idea: Let's assume we are sitting with an unknown player on a mid stakes NL 6-max table who plays six of his hands in the first two orbits. Conventional HUD's would now show us the player's VPIP to be 50% - so that money was invested in six of twelve hands.
As a matter of fact, we know though that such high VPIP's are rather rare in this kind of game. Of course, the six-in-twelve hands is an indication that the player is rather loose. It is rather improbable though, that he will really play 50% VPIP on a bigger sample size. It is much more probable that the player also received above-average hands in the two orbits.
The point is that we don't want to play against this player as if he really had a VPIP of 50%. Nonetheless, we want to slip in the watched hands into our game as information. And with this task we want to help the players. We now try to calculate the "most probable" VPIP value of the player with the aid of the VPIP spread in the player population and the watched hands of a player.

PokerStrategy: How will the PokerStrategists profit from this in a concrete way?

plexiq: This is mainly for beginners and players who haven't dealt with this problem yet in order to help them avoid making a "wrong read". The goal is that the shown VPIP value, in the absence of other reads, also really means that you should play against this player as if his true VPIP would also be this value. We want to "alleviate" the problem of small sample sizes, so to speak.

PokerStrategy: Paul, you've mentioned the Optical Character Recognition. What exactly is that and what is it used for?

khalthag: With the OCR you want to recognize letters on a picture which is displaying text according to optical characters.
We
need the OCR to read out the table and player data from the poker software. Of course, there are other ways to get this data, but they are heavily tailored to the respective poker software. Especially when the software protects the data from being read out, the recognition via the OCR is often the last option.

PokerStrategy: Does the OCR work the same everywhere?

khalthag: In the end it should, but during the development it's easier to get closer to some poker clients than to others. It's especially difficult to read characters correctly when the poker software uses specific antialiasing methods. This way, a character's size is often only a few pixels and thus doesn't offer enough data to be discerned from another similar character.

PokerStrategy: What is the additional value for the players when the OCR works?

khalthag: The main advantage is of course the real-time usage. Usually, the hand histories are processed to get the table and player data. The drawback of the hand histories is that you can only process them when they are available. With most poker clients they are only created after the hand has finished. With the OCR you would have the option though to analyze always current game data and react accordingly.
So it would be possible for beginners to read out the starting hand and then overlay a game suggestion immediately on the table. Furthermore, it would be possible to include situational equity analyzes, as well as display odds and outs for draws depending on the game situation and the pot size.

PokerStrategy: Holger, what kind of challenges does Elephant contain for you?

eniki: There is a lot, starting at the further development of the parser and the inclusion of further platforms right up to the new additional statistics. I am looking forward to the constant improving of the stability, the accuracy and the speed of  Elephant.

PokerStrategy:
So there is a lot to look forward to, including the OCR, FGS, an increased accuracy of VPIP values via mathematical evaluations. And the PokerStrategists will be able to participate in the project in the future!
Thanks a lot for taking some time out for this interview and updating us. We're excited!