DUCKQUANT / THE QUANT & AI DIRECTORY
1,650 resources, ranked by in-directory PageRank
Closes a bar when a run of same-signed trades exceeds what the recent buy probability makes plausible.
The imbalance rule applied to signed *volume* rather than signed tick count, so one large order weighs more than many small ones pointing the…
Closes a bar when signed tick flow becomes unusually one-sided relative to what recent history led you to expect.
Closes a bar once a target traded *value* is reached. Unlike volume bars this stays comparable as the price level changes — 1,000 shares…
Open-source MCP server for tennis scores, fixtures, rankings, and win probabilities for prediction-market research. API plan limits apply
Chinese open-source Python quantitative trading system covering stocks, options, futures and Bitcoin with ML.
Validates that a latency measurement is meaningful before any latency is computed from it.
Decides why a bar is missing: the session was closed, the instrument was halted, the feed dropped, or a sequence number was skipped.
Compares two return intervals and reports exactly how they overlap.
Builds a common clock for several instruments by advancing only when every one of them has refreshed. One established construction, not the standard remedy.
Interpolates between the quotes either side of a target time. Not causal, and no accuracy advantage over carrying the previous quote forward is claimed.
Samples a series onto a grid by carrying the last value known *at that moment* forward.
Classifies a quote as normal, locked (bid equals ask) or crossed (bid above ask).
Applies a stock dividend, where shareholders receive additional shares instead of cash.
Reconstructs what was actually knowable at a given moment, using each record's available_at rather than its observation_time.
Compares two versions of a feed schema and classifies every change as compatible or breaking.
Reconciles quotes from several providers into one defensible price, or refuses to.
Weight by full market capitalisation. The textbook construction, and the starting point every float and capping rule modifies.
Follows a symbol through renames in either direction. FB became META, and any store keyed on the symbol now holds two disconnected halves of…
Calculate advancing issues per declining issue without hiding zero denominators, incomplete coverage, or correction timing.
Build a revision-aware issue-count breadth snapshot that reconciles stable listings and refuses unsupported results.
Measure participation beneath headline index movement. Representative subjects: Advance-decline line, McClellan Oscillator, McClellan Summation Index, ratio-adjusted breadth, TRIN, breadth thrust.
Neutralises currency exposure with rolling forwards. Hedged and converted variants of the same index can diverge by double digits over a year — they…
Applies an exponent to market value before weighting, compressing the gap between the largest and smallest members.
Float weighting with a ceiling on any single constituent. Regulation and mandate limits require it, and the redistribution is subtler than it looks: capping…
Weight by the shares that can actually be bought. Excluding strategic, government and insider holdings is what makes an index replicable — a fund…
Classifies a candle body as long, short or doji **relative to recent bodies** rather than against a fixed threshold.
Decomposes one candle into body, upper and lower shadow, and direction.
Convert OHLC structure into testable events. Representative subjects: Candle anatomy, single- and multi-candle patterns, context filters, scale-aware thresholds, pivot confirmation, gap logic.
The difference between a 19-day and a 39-day EMA of net advances.
The absolute difference between advances and declines, in issues — direction discarded on purpose, and no interpretation of the magnitude claimed.
Divides net advances by the number of issues traded before accumulating, which keeps the line comparable across decades as listing counts change.
Build a seeded, calendar-complete, revision-aware cumulative breadth line that refuses to publish incomplete evidence.
Describe systematic return drivers. Representative subjects: CAPM beta, Fama-French factors, momentum, quality, value, size.
Measure ordinary and extreme portfolio loss. Representative subjects: Historical/parametric/Monte Carlo VaR, expected shortfall, drawdown, marginal/component VaR, stress tests, scenario analysis.
Convert forecasts and constraints into holdings. Representative subjects: Mean-variance, minimum variance, maximum Sharpe, risk parity, HRP, Black-Litterman.
