Jeff Hwang Discusses All-In EV Software Statistics

Recently, Jeff Hwang, a semi-professional poker player and renowned poker author, wrote an article for CardPlayer Magazine explaining the issues related to the "All-In EV" or "All-In Equity" statistics calculated by Holdem Manager and PokerTracker. The article was republished in its entirety on PokerTracker’s website. Hwang 's article provides precise details on the calculation formulas behind the well-used statistic along with a clear explanation on why the stat is unreliable with the exception of cases in which all hole cards are revealed during heads-up play.

The "All-In EV" statistic, which is labeled "All-In Equity" on PokerTracker 4, got its start from the overbearing desire of online poker players to know exactly how lucky (or unlucky) they are running in all-in situations. The concept is that a player can be acutely aware of he or she should have profited or lost over the long-term by calculating the corresponding odds at the point in which the money got in during an all-in situation.

Armed with this information, a poker player would in theory be able to tell with more accuracy whether his or her total profit or loss is a true reflection of actual game play at the virtual tables.

However, as Hwang pointed out in his article, the two issues that make accurate calculation of this statistic impossible are (1) calculating expectation in multi-way pots and (2) the fact that many online poker sites do not reveal opponents' hole cards unless the hero loses the hand. This is true for Bodog, Boss, Yatahay, and Everleaf, among others.

Hwang makes it clear in his article that depending on the software you are using, the "All-In EV" stat could be skewed long-term to show that a player is running anywhere from very unlucky to slightly lucky, which makes the stat meaningless except for heads-up play.

The issues regarding multi-way pots revolve around unexposed hole cards. Hwang writes that in order for the "results-oriented" calculation to provide accurate results, all opponent hole cards at the time you went all-in must be shown and included in the calculation. However, this is not what happens in many cases. The following is the example used in his article.

For the sake of argument, let's suppose that the game is No Limit Hold'em, there are three players who see the flop, and our hero has less money behind him than the other two opponents:

Hero ($25) – Kh-Ks
Opponent #1 ($100) – Qh-Qs
Opponent #2 ($100) – Ah-Kd
Flop: 4d-4h-4s

Let's suppose that our hero shoves his remaining stack into the pot and the other two opponents call.

The turn comes a blank card that (let's say the 5d) and Opponent #1 subsequently bets the remainder of his stack, resulting in a fold from Opponent #2, who has not hit either of his overcards. This causes an issue with the statistic displayed in the programs because Opponent #2's hole cards will not be included in the calculation even though they were part of the equation when our hero got his money into play. This specifically causes the “All-In EV” stat to be inaccurate because the software would need to incorporate Opponent #2's A-K in order to make a complete calculation.

In summary, although Holdem Manager and PokerTracker are marquee programs that provide a sure-fire way to improve one's long-term expectation through data storage and analysis, the "All-In Equity" statistic should not be considered 100% accurate unless a player only competes in heads-up poker on networks that reveal your opponents’ hole cards at all times.

Read the entire article by visiting PokerTracker’s website.