Reasoning Patterns, Critically

Chain-of-thought asks a model to reason step by step before answering; self-consistency samples several chains and takes a majority; ReAct interleaves reasoning with tool calls; tree-of-thought and search methods explore branching reasoning paths with pruning. Each buys gains on puzzles and multi-step tasks, and each has a price in tokens, latency, and cost. The critical lesson for finance is that fluent reasoning is not arithmetic reliability or factual grounding. Models that narrate confident derivations still mis-carry digits, misremember filings, and compound errors across steps. Treat these patterns as useful priors for decomposing work, never as verification, and pair them with the tools and evaluations later in this track.