Quote:
Originally Posted by wayneking7
Do u guys realize that ur calculating EV for SNGs slightly wrong?
Yes, we knew that "ties are missing" from nearly all ICM Simulation software because we were the ones who alerted others in the community that virtually all of the competitive applications were doing the same thing (without naming names). We have been working on an alternative solution since the early fall IIRC. We just finished coding an update that better incorporates ties for two way all-ins that we are currently testing in alpha for Holdem games, we expect it to be available next week. This is the first stage in our equity improvements that accompanies many new changes in tournament calculation options that you will see in the next version - and there is more to come. This is not easy to code, there are a surprising amount of tricky situations that we need to be able to account for in our database, its a lot more complicated than just simple math, but you will soon see some of the improvements we have made.
Did you discover this on your own, or did you read about it in a thread? We have found that most threads discussing this have misrepresented the issue, if so please share the URL so we can check it out and provide an explanation why virtually all developers were doing the same thing, and how we have been working to create an improved equity calculation solution that incorporates the potential for ties.
Quote:
Originally Posted by wayneking7
Ur using the total hand equity instead of breaking it down into win/lose/draw %'s.
Thats a simplified explanation, its not really accurate, but its close enough for this thread. I believe the potential for ties were overlooked for so long because it's the equivalent of a rounding error and in turn requires longer processing time to address that rounding error, but we have already addressed it and you can expect it in our next release.
Quote:
Originally Posted by wayneking7
Obv this only makes a very slight difference in 1 hand, but can build up to be very significant over a large sample.
We have found that our improved algorithm does not create any significant changes in our test databases that will be visibly noticeable in your graph in very large samples, as stated above the problem is mostly a rounding error issue. With that said, the improvement in accuracy is a worth while investment in our development time, hence ties are now properly addressed.
Quote:
Originally Posted by wayneking7
Plus, i havnt checked yet but i was told PT4 is treating same hand situations (AKo v AKo) as 50% and 50%, which is lol. Has this been fixed yet?
In the PT4 equity simulator we only display the hand equity, we don't display the Win, Tie, or Loss percentages - but these are factored in behind the scenes. Win Lose & Tie percentages are more helpful when sharing results, PT4's equity calc was not designed to share results like Equilator is, This was a choice we made for development resources, it allowed us to focus more time on the features we added that others don't have such as range modeling, and the incredibly powerful range creation tools for Omaha. Maybe someday we will revisit to consider adding a more in-depth display, but since it has ben requested very few times its not on our priority list at the moment (priorities include OSX, TagTracker, and other yet to be announced features - such as tie compensation in the next release).
PokerStrategy's Equilator is a great example of a free application that shows Win-Tie Loss in an overview display in a method that allows you to copy the results. PT4's equity calculator was not designed to replace Equilator or other similar tools, it was designed to augment the replayer's street by street equity display by allowing you to create ranges for your opponents. In the next version of PokerTracker 4 we include an option to show equity ties, click the save menu and then select Display Ties,
Last edited by PokerTracker; 03-10-2013 at 03:11 PM.
Reason: Better explanation of improvements in pending PT4 release