Data and Research
The data explorer and the alpha factor analysis tool.
Data Explorer
datadata, developed by FMZ Quant, is a quantitative financial data platform. The Data Explorer module of FMZ integrates its services and features, and FMZ users can use it without registering a separate datadata account. Analyze large amounts of data with SQL, build various charts through a visual interface and share them with your team. For examples, see the Data Explorer articles.
- Data sources: the data sources provided by datadata are updated continuously in real time and cover many kinds of data; you can also upload CSV files as private data sources and preview them on the Data Explorer page.
- Queries: query data with SQL, with query parameters; results can be downloaded as CSV or JSON files.
- Saving research: click "Save" at the top right to save the current SQL query to the resource list of your Data Explorer (the resource list button is to the left of "Save").
- Visualization: query results can be shown as tables or with various visualization components.
- Sharing research: as a public link, embed code (e.g. in a community post), an embedded web page, a data link or a preview image link. A "data link" also feeds the data straight to strategies, in both backtests and live robots.
Alpha Factor Analysis Tool
The analysis formulas reference the market calculation methods from worldquant's publicly available 101%20Formulaic%20Alphas.pdfalpha101, with basic compatibility for its syntax (unimplemented features are noted), and have been enhanced. This tool is used for rapid time series computation and validation of trading ideas. Alpha Factor Analysis Tool Page.
Functions and Operators
The {} below represents placeholders, all expressions are case-insensitive, x represents data time series
abs(x), log(x), sign(x): absolute value, logarithm and sign function respectively.
The following operators +, -, *, /, >, < also conform to their standard meanings, ==: equality check, ||: logical OR, x ? y : z: ternary conditional operator.
rank(x): Cross-sectional ranking, returns the percentile position. Requires a candidate pool of several instruments; with a single instrument nothing can be ranked and the raw value is returned.delay(x, d): Returns the value of series x from d periods ago.sma(x, d): Calculates the simple moving average of series x over d periods.correlation(x, y, d): Calculates the correlation coefficient between time series x and y over the past d periods.covariance(x, y, d): Calculates the covariance between time series x and y over the past d periods.scale(x, a): Normalizes data such thatsum(abs(x))=a(a defaults to 1).delta(x, d): Calculates the current value of time series x minus the value from d periods ago.signedpower(x, a):x^a.decay_linear(x, d): Calculates the d-period weighted moving average of time series x, with weights d,d-1,d-2....1 (normalized).indneutralize(x, g): Industry neutralization based on industry classification g, currently not supported.ts_{O}(x, d): Performs operation O on the past d periods of time series x (O can specifically represent min, max, etc., see below), d will be converted to integer.ts_min(x, d): Minimum value over the past d periods.ts_max(x, d): Maximum value over the past d periods.ts_argmax(x, d): Position ofts_max(x, d).ts_argmin(x, d): Position ofts_min(x, d).ts_rank(x, d): Ranking of time series x over the past d periods (percentile ranking).min(x, d):ts_min(x, d).max(x, d):ts_max(x, d).sum(x, d): Cumulative sum over the past d periods.product(x, d): Cumulative product over the past d periods.stddev(x, d): Standard deviation over the past d periods.
Input Data
Input data is case-insensitive. Default data is the instrument selected on the webpage, but can also be specified directly, for example: binance.ada_bnb
returns: Close price returns.open, close, high, low, volume: Open price, close price, high price, low price and volume within the period.vwap: Volume-weighted average price (not yet implemented, currently using close price).cap: Total market capitalization (not yet implemented).IndClass: Industry classification (not yet implemented).