This article presents the Excitability-Margin Model (EMM) – a quantitative, falsifiable framework that unifies schizophrenia, major depression, and PTSD under a common biophysical bottleneck: the narrowing of the excitability margin in ventral CA1 pyramidal neurons (defined as spike threshold minus resting potential).Through re-analysis of published datasets and calibrated modeling, we show that chronic stress alone reduces this margin by ~11 mV (via ↓KCC2, ↑IL-6/NKCC1, ↓GIRK/TASK, ↓Na⁺/K⁺-ATPase α1). Repeated activation of the same memory trace (CREB-high “hot spot”) further lowers the buffer, and a single common risk allele (e.g., CACNA1C, SCN2A) drives the effective excitability margin to 1–3 mV. At this point, ordinary network events — dendritic plateaus (~5.9 mV), K⁺ bursts (~3 mV), theta peaks, or ripples — carry enough energy to force involuntary replay of emotional engrams.All known risk factors, however diverse — infections/IL-6, high-glucose diet, microplastics, reactive oxygen species (ROS), psychoactive substances, and genetic variants — converge on the same endpoint: narrowing of the vCA1 excitability margin. Once the buffer collapses, the valence of the replay dictates neurochemical cascades and clinical phenotype:• Fear replay → dopamine and cortisol surges → Glu↑ / GABA↓ imbalance maintains the critically low margin and enables further activations. ROS accumulation depletes glutathione; once reserves are exhausted, PV-interneurons deteriorate, leading to gamma desynchronization, loss of inhibition, and disease progression. This self-reinforcing loop aligns with hallmark findings in schizophrenia.• Sadness replay → reduced monoamines, CRF upregulation, δ–θ dominance → depression, consistent with the serotonergic/dopaminergic deficit, HPA dysregulation, and DMN overactivation.• Trauma replay → excessive noradrenaline, β–γ instability → PTSD, consistent with locus coeruleus hyperactivity, flashback-like replays, and exaggerated startle responses.Importantly, all three phenotypes share one striking commonality: ventral CA1 hyperactivity. According to EMM, this is not a superficial correlation but the direct consequence of margin collapse. This explains why EEG and fMRI consistently show ventral hippocampal overactivation across schizophrenia, depression, and PTSD.Each replay not only triggers acute symptoms but biologically deepens the margin collapse, explaining clinical progression. The model also clarifies why depression can evolve into schizophrenia: once sadness clusters reach a critically low buffer, they can recruit fear engrams into the same unstable state.Beyond symptom-specific cascades, the model accounts for transdiagnostic features — impaired episodic memory replay, anhedonia via dopaminergic blunting, and neuroinflammatory signatures (IL-6, TNF-α). Genomic studies report substantial shared heritability among psychiatric disorders (e.g., MDD and SCZ genetic correlation ~0.43), and epidemiological surveys show 35–68% comorbidity between PTSD and depression.Finally, we include an optional, falsifiable ELF phase-bias module: weak 7–30 Hz magnetic fields, potentially transduced by magnetite chains, may bias hippocampal θ activity. This branch is modular and detachable, with preregistered falsification criteria (β-loop: environmental ELF variability should precede psychiatric admissions). Negative results retire the ELF branch without affecting the core model.Together, EMM provides a compact, quantitative, and testable scaffold that unifies risk, replay, and phenotype. It also delivers concrete, reproducible predictions for translational studies, with pre-specified biomarkers (on-scalp MEG γ-burst reduction, HRV rMSSD increase) and a therapeutic blueprint (Four-Axis Reset).