Skip to content

README.md

Latest commit

 

History

History
45 lines (33 loc) · 1.87 KB

README.md

File metadata and controls

45 lines (33 loc) · 1.87 KB
Raw

Monte Carlo simulator (standalone)

simulator.py is a tool which is used to explore how a given R-multiple distribution plays out over many simulated trade sequences. To achieve this it resamples with replacement from a loaded R-multple distribution (which is considered as a simple "bag of marbles").

The R-multiple distribution is read from a .csv file with a single Rmul column (e.g. out/tables/Rmul_trades.csv, produced by the main pipeline), and passed via --rmul-dist.

Configuration

Configuration is set from the file tst/config/simulator_conf.json:

  • risk_percent — fraction of balance risked per trade.
  • balance — starting account balance.
  • sim_len_max — maximum number of trades per simulated run.
  • iterations — number of simulation runs.
  • plot_frac — fraction of the simulated runs that are drawn on the plot. Added to keep dense plots readable.
  • outlier — used to cap the plot's y-axis (median + outlier × stdev of final balances).

Commandline parameters

From the repository root:

python tst/simulator.py --rmul-dist <path> [--basedir <path>] [--outdir <path>] [--loglevel <level>]

Options

  • --rmul-dist is the path to a CSV file containing the R-multiple values to resample, in a column named Rmul.
  • --basedir defaults to the current working directory. It does not affect where the configuration file is loaded from.
  • --outdir sets the output directory. Relative paths are resolved against basedir; absolute paths are used as-is. Defaults to out.
  • --loglevel accepts DEBUG, INFO (default), WARNING, ERROR or CRITICAL.

Output

The output diagram is written to <outdir>/images/monte_carlo_plot_rmul.png, showing a plot_frac sample of the simulated balance trajectories plus summary statistics (median, stdev, min/max, loss streaks, max drawdown, SQN) computed over all runs.