|
|
我在2+2的原帖里提过，少数场景是可以使用gto帮助决策的，当时我举的例子是豪客赛：
C) There are some game situations GTO can be useful
Super high roller tournaments are one of the few game conditions that these quasi-GTO software can actually help. If you look at these players, they are in general very sharp, spot opponent weakness quick. They are very close in skill level. The game mostly played at short stack and heads up. All these combined, plus the stake they play, make it worth trying to play the software by trial and error, or just trying to get some suggestions they normally would pass.
But even for SHR, the usefulness of GTO solver are exaggerated in some articles. These players get there because of their talent and effort. A brutal truth is that poker is a game heavily relies on player’s raw talent. Even for some GTO promoters, they benefited from their ability to extract useful information from a rough tools. This does not apply to the mass they promote the software to. The value part of GTO training is almost always from self-taught, not from the trainer/instructors.
你说的GG极速100BB是另外一种可用的场景，这里首先是极速，导致牌手历史数据的缺失，其次是这种游戏环境快速消耗掉很差的牌手，导致整体水平接近，然后大家逐渐适应这个环境而导致的下注/加注筹码逐渐逼近，使得gto软件对牌手本人和对手的下注/加注尺度假设误差缩小，再固定到大家有效筹码100bb等很多条件的符合。你觉得这个条件下计算的“GTO”结果记下来对现场适用吗？不用说上面说的条件符合问题，就是现场抽水跟网络抽水不同这一点，你得到的gto解在理论意义上就已经不是gto了。牌局抽水本身相当于这个零和游戏的第7位牌手。它变了，牌局规则/条件就变了，原来的gto就不再是gto了。
事实上，网络游戏出现过很多特定牌局的优化解，比如早期的SSS，DON，这些都是在gto软件出现之前就有的。即便没有纳什均衡为基础的gto出现，以今天的算力，也同样得到这些所谓的GTO解。事实上，现今市场上的各种gto软件，没有一款是使用的gto（纳什均衡）的算法，都是使用计算蛮力在那里做树搜索穷举，这些软件使用的算法根本无法证明它的计算结果是收敛的，是无限逼进正解的，得出的计算结果也是伪gto，这是为什么我在上面英文第一句里面称这些软件为quasi-GTO software.
回到我们简单的现实需求，这些特定场合的计算对我们现场实战能力的提升有多大帮助？是怎样的帮助？很多人正是对这类gto结果的迷信导致导致他们大骂对手是鱼却又赢不了，以为自己的是正解，却不知道自己成了软件鱼。
好的牌手在桌上做决策，不以对手“应该”怎么做为基础，而是考虑对手“会”怎么做，围绕这一点最大限度剥削。这个不是gto能帮你的
|
|
楼主
