event_beta() — realized-loading regression of asset returns on event-probability news, with a base-R Newey-West (Bartlett/HAC) covariance, model-implied outcome levels under the risk-neutral adding-up constraint, and — given an externally measured exposure via deta — the loading test of beta = 1.
q_from_ffutures() — meeting-implied probabilities from 30-day fed funds futures on the standard month-average extraction, with the new fomc_meetings calendar dataset (2021-2027).
q_from_deal_spread() — completion probabilities from merger-arb spreads: deal clocks feed the same q -> logit -> A pipeline.
New dataset djt2024 (daily DJT prices, June-November 2024) powering the event-beta example against polymarket2024.
eventclock 0.2.0
Inference & validation
event_clock(se = TRUE) — standard errors and confidence intervals for the realized-variation estimate, via the quarticity analogue (log-based intervals) or a moment-matched wild bootstrap (se_method = "bootstrap").
ec_signature() and plot_signature() — the sampling-frequency signature plot of the clock (subsample-averaged), the standard microstructure-noise diagnostic.
ec_validate() — data-quality report for event-probability series: coverage, staleness, tick size, gaps, and outliers, in the flag-don’t-drop spirit.
ec_simulate_path() gains jump_share/n_jumps for lumpy information arrival (Bayes-consistent per step), the testbed for the jump-robust estimator variants.
New vignette “Validating the event-clock estimator” (Monte Carlo: consistency, CI coverage, outcome-independence/measure robustness, bipower under jumps, microstructure and the snapshot rule).
CI now runs the full R CMD check including vignettes on a three-OS matrix in a non-UTC timezone, plus a coverage job.
eventclock 0.1.0 (initial version)
New features
Initial version of the package.
as_event_prices() turns any data.frame/tibble of traded event probabilities into a standardized event_prices object with flag-don’t-drop cleaning (q_from_price() handles discount and overround normalization).
event_clock() estimates event-clock (information) time A as the realized variation of log-odds, with truncation, bipower, and largest-move robustness variants; event_clock_path() returns the cumulative clock; event_clock_forecast() implements the trailing-window real-time benchmark.
