Algo.md

What is an algo? A namespace with this symbols defined:

(algo-calc ds algo-opts) (algo-opts-default) (chart-spec) (table-spec)

Result of algo-calc is a dataset with algo-specific columns added. The dataset can be used to create a chart. The last row is a map of the current data which is use to display a grid of columns for multiple assets. In backtesting the window will be really wide (limited only by RAM). In realtime the window will be as short as possible; so lookback needs to be defined depending on the algo and its options. If trades should be generated the column :signal is added which can be #{:buy :sell :hold :no}. Chart spec is used to create custom charts for different algos. Table-spec is used to customize the format the table for the algo.

Realtime views:

asset list: assets that use the same algo and the same options. UI is used to switch between different assets. chart for a single algo. Selected via asset-list-select. Streamed to live clients. Table: formatted via table-spec. Each bar changes the table in clj and sent via websocket to live clients (webbrowser) Backtest views:

same as realtime but calculated only once. additionally backtest statistics.

Screener

run algo on multiple assets.

Multi-timeframe algo (algo-calc ds algo-opts) :daily bars (algo-calc ds algo-opts) :5 min bars

Into :5 min bars series a column is added with row-id of the most-recent :daily bar. Therefore on each :5min row it is not only possible to access any :daily column of the most-recent :daily row but also any days prior to that.

Backtest Env gets ds-1min ds-daily (let [d (algo-calc ds-daily algo-opts) 1m (algo-calc ds-1min algo-opts) links (create-links ds-1min ds-daily) 1m-d (add-cols 1m links d)] (algo-calc 1m-d algo-opts))

Realtime Same as backtest but on-daily and on-1m and on-daily events trigger such calculations.