\section{Research design}
\label{sec:method}

\subsection{Clocks, cutoffs and the information set}

Figure~\ref{fig:timeline} summarises the event-time design. For firm $i$ and fiscal quarter $q$ we observe the period end
$E_{i,q}$, the earnings-release date $A_{i,q}$ (the reveal date, with its before-open/after-close timing when known), the
10-Q/10-K filing date $F_{i,q}$ and the vendor availability instant $V_{i,q}=F_{i,q}+1$ day. The prediction cutoff is
$c=A_{i,q}-h$ trading days for $h\in\{40,20,10,5,1\}$, with $h=20$ as the preregistered primary horizon. A cutoff is
admissible only if $c<\min(A_{i,q},F_{i,q})$ and $c>A_{i,q-1}$, so the feature window never contains the release of
quarter $q$ and never straddles two reveals. The information set $\mathcal{I}(c)$ contains regular-session bars with
session date $\le c$ and accounting facts with $V_{i,j}\le c$ only; the latest such quarter is $j^*(c)=q-1$ in the
retained sample (we require that the forecaster predicts the quarter immediately following the latest available one).
Twelve leakage guards (future facts, fiscal-end joins, post-cutoff bars, test-period transforms, daylight-saving errors,
identifier drift, split artefacts, release-day bars, overlapping windows across splits, survivorship, ingestion dates,
randomised placebos) are implemented as unit tests in the replication package; the local as-of join was verified live
against the FinData API's \texttt{as\_of} semantics on random entities and instants.

A fixed-calendar design (month-end cutoffs, target = the next release within 65 trading days) is used for the pricing
tests and as a robustness check because it does not condition on knowledge of the release date.

\subsection{Targets}

The primary outcome is the scaled operating-income surprise $S_{i,q}$ of equation~\eqref{eq:surprise}, with
$\mathbb{E}[X_q]=X_{q-4}+\tfrac14\sum_{k=0}^{3}(X_{q-1-k}-X_{q-5-k})$ computed from quarters available at $c$ and scaled by
the latest available total assets. The binary adverse outcome is $D_{i,q}=\mathbb{1}[S_{i,q}\le-0.5\%]$, i.e., a
shortfall of at least half a percent of assets (roughly a fifth of a typical quarterly return on assets). Secondary
outcomes are the net-income analogue, a year-on-year ROA decline of at least one point, a negative operating-cash-flow
innovation of at least one percent of assets, a liquidity/coverage deterioration (current ratio down 10\% or interest
coverage falling below one), a symmetric placebo $|S_{i,q}|\ge0.5\%$ that a pure volatility channel would predict, and an
exploratory EPS miss against the vendor's estimate (available for \epsEstimateShare\% of releases; vintage unknown).
The continuous surprise, winsorised at the training-period 1st/99th percentiles, is modelled by ridge regression. No
target uses prices.

\subsection{Features by sampling frequency}

All market features are derived from the same raw 1-minute bars of the regular session (09:30 to the closing print),
aggregated to 5-minute and 30-minute buckets and to daily bars. Each intraday \emph{family} is computed per session
on bars of the given length and averaged (or summed) over the 20-session window ending at $c$: realized variance,
bipower variation, relative jump, negative-semivariance share, signed-jump variation, realized skewness and kurtosis,
concentration of variance in the largest bar, first-order autocorrelation, intraday drift, opening and closing volume
shares, volume dispersion (Gini), down-bar volume share, share of expected bars present, share of variance on
market-down bars, realized beta to SPY (ratio of summed cross-products), variance ratio against the daily return variance,
and volatility-of-volatility. The 30-minute, 5-minute and 1-minute sets therefore have the \emph{same} \kIntra{} columns.
Daily features (\kDaily{} columns) are 20- and 60-day returns, volatilities, Parkinson range, skewness, kurtosis,
overnight and intraday return components, volume ratios, Amihud illiquidity, zero-return share, dollar volume, maximum
drawdown and bar activity. Point-in-time accounting features (\kFund{} columns) are margins, ROA, accruals, cash,
leverage, current ratio, coverage, capex, R\&D, issuance, revenue growth, size, the last available surprise, the trailing
mean surprise, the trailing adverse rate and the age of the information. Bar statistics are descriptive; nothing is
interpreted as order flow, spread or informed trading.

\subsection{Baseline ladder and equal budgets}

Table~\ref{tab:infosets} nests the information sets. B0 is the training base rate (and a shrunk firm-history rate);
B1--B5 are $\ell_2$-penalised logistic regressions whose penalty is tuned on the validation period only; LB1--LB5 are
gradient-boosted trees \citep{ke2017lightgbm} with an identical random-search budget (24 trials, early stopping on the
validation period) for every information set, so that the frequency comparison is a comparison of information sets,
not of capacity. Winsorisation bounds, imputation medians and standardisation constants are estimated on training rows
only. A supplementary set (B5all) concatenates all three intraday resolutions. Neural and text models are not used.

\subsection{Validation and inference}

The chronological split trains on cutoffs in 2012--2018, tunes on 2019--2020 and evaluates on 2021--2026, with purging
of training outcomes revealed after the next split's start minus a one-quarter embargo \citep{lopezdeprado2018}.
Expanding-window annual refits from 2016 provide out-of-fold predictions for the panel regressions and the timing and
pricing exercises; a 20\% firm hold-out evaluates transfer to unseen firms. The primary metric is log loss; Brier score,
AUC, PR-AUC and calibration are reported for every model. Paired differences (B5$-$B2, B5$-$B4, B5$-$B3) are assessed
with a block bootstrap over calendar quarters of the reveal date \citep{kunsch1989, politis1994} and with the
Giacomini--White test on quarterly mean loss differentials with Newey--West variance \citep{diebold1995,
giacomini2006}; Holm's correction is applied within each hypothesis family \citep{holm1979}. Panel regressions
\begin{equation}
  S_{i,q}=\alpha_i+\delta_{s(i)\times t(q)}+\beta\,\mathrm{LFDS}_{i,c}+\gamma'X_{i,c}+\varepsilon_{i,q}
  \label{eq:panel}
\end{equation}
use the out-of-fold B5 probability as the Latent Fundamental Deterioration Signal (LFDS), firm and sector-by-quarter
fixed effects, and two-way clustering by firm and calendar quarter \citep{petersen2009, cameron2011}. Mechanism tests
(H3) interact the \emph{incremental} signal $\mathrm{LFDS}^{\mathrm{B5}}-\mathrm{LFDS}^{\mathrm{B2}}$ with ex-ante
terciles of size, illiquidity and leverage formed within each calendar quarter.