EventTime Quantifies Cybersecurity Impact on Financial Time Series

External shocks, such as cybersecurity breach disclosures, can cause significant and abrupt disruptions in financial time series, leading to substantial abnormal losses. While these events are reported discretely, their financial consequences unfold through continuous market dynamics. The challenge lies in predicting short-term post-disclosure abnormal loss given pre-event market history and limited event metadata, rather than forecasting the entire post-event trajectory. Traditional time-series models often struggle with these rare, heterogeneous external events due due to their focus on endogenous regularities. To address this, researchers introduce EventTime, a multi-resolution framework designed to quantify event impacts. EventTime integrates long-horizon market context, short-horizon pre-event dynamics, and event-specific metadata. A key component is its event fusion module, which effectively couples temporal representations with event attributes to identify relevant market patterns. To overcome the issue of sparse supervision, EventTime employs a dynamic contrastive objective during training, constructing event- and time-series-aware positive and negative pairs. The study also presents SECURE, a new real-world dataset linking cybersecurity incidents with stock market data and semantic features derived from LLMs. Experiments demonstrate that EventTime consistently outperforms state-of-the-art time-series and event-aware baselines in estimating financial losses following cybersecurity disclosures. Further analysis highlights EventTime's ability to create more event-sensitive representations, its robustness to incomplete metadata, and its capacity for more interpretable estimates of short-term market impact.