games.hold_em.aug_llms

Texas Hold’em specialized augmented LLM configurations.

This module provides specialized augmented LLM configurations for Texas Hold’em poker, with customized prompts, output schemas, and model configurations for different aspects of poker gameplay:

• Hand evaluation and analysis

• Opponent modeling and profiling

• Betting strategy and decision-making

• Position-based play adaptation

• Pot odds and equity calculations

These specialized configurations build on the base engines in engines.py but provide more targeted capabilities for specific poker reasoning tasks.

Examples

>>> from haive.games.hold_em.aug_llms import get_hand_analyzer, get_bluff_detector
>>> from haive.core.engine.aug_llm import AugLLMConfig
>>>
>>> # Get a specialized hand analyzer
>>> hand_analyzer = get_hand_analyzer("advanced")
>>> result = hand_analyzer.invoke({
>>>     "hole_cards": ["Ah", "Kh"],
>>>     "community_cards": ["Qh", "Jh", "2s", "7c", "9d"],
>>> })

Functions

get_betting_strategist([style])	Get a specialized betting strategy configuration.
get_bluff_detector([sensitivity])	Get a specialized bluff detection configuration.
get_complete_llm_suite([player_style])	Get a complete suite of specialized LLMs for a poker player.
get_hand_analyzer([level])	Get a specialized hand analyzer configuration.
get_opponent_profiler([tracking_depth])	Get a specialized opponent profiling configuration.
get_situation_analyzer([focus])	Get a specialized situation analyzer configuration.
get_table_dynamics_analyzer()	Get a specialized table dynamics analyzer configuration.

Module Contents

games.hold_em.aug_llms.get_betting_strategist(style='gto')

Get a specialized betting strategy configuration.

This function returns an augmented LLM configuration specialized for making betting decisions with different strategic approaches: - gto: Game Theory Optimal balanced approach - exploitative: Adjusts to exploit opponent tendencies - aggressive: Higher variance, aggressive betting strategy - conservative: Lower variance, tighter betting strategy

Parameters:

style (str) – Strategic style (“gto”, “exploitative”, “aggressive”, or “conservative”)

Returns:

Configured betting strategist

Return type:

AugLLMConfig

games.hold_em.aug_llms.get_bluff_detector(sensitivity='balanced')

Get a specialized bluff detection configuration.

This function returns an augmented LLM configuration specialized for detecting opponent bluffs with different sensitivity levels: - conservative: Lower false positive rate, only identifies clear bluffs - balanced: Moderate sensitivity to bluffing signals - aggressive: Higher sensitivity, may have more false positives

Parameters:

sensitivity (str) – Bluff detection sensitivity (“conservative”, “balanced”, or “aggressive”)

Returns:

Configured bluff detector

Return type:

AugLLMConfig

games.hold_em.aug_llms.get_complete_llm_suite(player_style='balanced')

Get a complete suite of specialized LLMs for a poker player.

This function creates a coordinated set of specialized LLM configurations that work well together based on a player’s overall style.

Parameters:

player_style (str) – Overall player style (“tight”, “loose”, “aggressive”, “passive”, “balanced”, or “tricky”)

Returns:

Dictionary of specialized LLM configurations

Return type:

Dict[str, AugLLMConfig]

games.hold_em.aug_llms.get_hand_analyzer(level='standard')

Get a specialized hand analyzer configuration.

This function returns an augmented LLM configuration specialized for analyzing poker hands with different levels of sophistication: - basic: Simple hand strength evaluation - standard: Balanced analysis considering draws and relative strength - advanced: Sophisticated analysis with equity calculations and range analysis

Parameters:

level (str) – Complexity level of the analyzer (“basic”, “standard”, or “advanced”)

Returns:

Configured hand analyzer

Return type:

AugLLMConfig

games.hold_em.aug_llms.get_opponent_profiler(tracking_depth='standard')

Get a specialized opponent profiling configuration.

This function returns an augmented LLM configuration specialized for building opponent models with different levels of detail: - basic: Simple tracking of betting patterns - standard: Balanced profiling of play style and tendencies - deep: Sophisticated profiling with psychological modeling

Parameters:

tracking_depth (str) – Depth of opponent tracking (“basic”, “standard”, or “deep”)

Returns:

Configured opponent profiler

Return type:

AugLLMConfig

games.hold_em.aug_llms.get_situation_analyzer(focus='general')

Get a specialized situation analyzer configuration.

This function returns an augmented LLM configuration specialized for analyzing poker game situations with different focus areas: - general: Balanced analysis of the overall situation - positional: Focus on positional dynamics and advantages - tournament: Specialized for tournament situations with ICM considerations - cash_game: Specialized for cash game dynamics

Parameters:

focus (str) – Analysis focus (“general”, “positional”, “tournament”, or “cash_game”)

Returns:

Configured situation analyzer

Return type:

AugLLMConfig

games.hold_em.aug_llms.get_table_dynamics_analyzer()

Get a specialized table dynamics analyzer configuration.

This function returns an augmented LLM configuration specialized for analyzing overall poker table dynamics, player interactions, and meta-game considerations.

Returns:

Configured table dynamics analyzer

Return type:

AugLLMConfig