#! /usr/bin/gawk -f
# Last edited on 2015-02-23 21:49:04 by stolfilocal

# Creates a fake trade data file with random volumes, where the log of the price is a 
# brownian variable.

# Reads from standard input a real trade data file, with some headers and then 
# date and time  as the first two fields.  Writes to standard output 
# some appropriate headers, and then one line for each input data line, 
# with the same date and time, with random prices and volumes.

# Users must define (with "-v") the random generator's {seed}, the initial price {P0},
# and the deviation of the increment in the decimal log of the price.

BEGIN \
  {
    if (seed == "") { arg_error("must define {seed}"); }
    if (P0 == "") { arg_error("must define {P0}"); }
    if (dev == "") { arg_error("must define {dev}"); }
    
    srand(seed);
    z = log(P0/1000.0)/log(10.0); # Decimal log of (price/1000).
    
    printf "# Generated by {fake_brownian_data.gawk}. Do not edit.\n";
    printf "# The decimal log of the price is a Brownian variable whose increments\n";
    printf "# are normally distributed with mean 0 and deviation %9.7f.\n", dev;
    # Print {txtable-reformat}'s header:
    printf "Timestamp          ";
    printf " ! %12s ! %12s ! %12s ! %12s", "Open", "High", "Low", "Close";
    printf " ! %16s ! %16s ! %12s", "V.BTC", "V.USD", "WTPrice";
    printf "\n";
  }
  
/^20[0-9][0-9][-]/ \
  { if (NF != 16) { data_error(("bad field count " NF)); }
    
    # Copy date and time from the input file:
    dt = $1;
    tm = $2;
    
    # Brownian integration:
    dz = dev * nrandom();
    z = z + dz;
    
    # Convert to linear scale:
    P = exp(log(10)*z)*1000; # Price in USD.
    VB = 2000*rand(); # Volume in BTC.
    VU = P*VB;
    printf "%s %s", dt, tm;
    printf " | %12.5f | %12.5f | %12.5f | %12.5f", P, P, P, P;
    printf " | %16.4f | %16.4f | %12.5f", VB, VU, P;
    printf "\n";
  }

function nrandom(   n,i,sum,avg,dev) 
  {
    # Normal random variable with zero mean and unit deviation.
    n = 20; # Number of uniform variables to add. 
    sum = 0;
    for (i = 0; i < n; i++) { sum = sum + rand(); }
    avg = 0.5*n; # Expcted value of {sum}.
    dev = sqrt(n/12.0);
    return (sum - avg)/dev;
  }
          
function data_error(msg)
  { printf "%s:%s: ** %s\n", FILENAME, FNR, msg > "/dev/stderr"; 
    printf "  «%s»\n", $0 > "/dev/stderr"; 
    abort = 1;
    exit(abort);
  } 
          
function arg_error(msg)
  { printf "** %s\n", msg > "/dev/stderr"; 
    abort = 1;
    exit(abort);
  }