diff --git a/datasets/wca/parse_wca.py b/datasets/wca/parse_wca.py
new file mode 100755
index 0000000..7810214
--- /dev/null
+++ b/datasets/wca/parse_wca.py
@@ -0,0 +1,107 @@
+#!/usr/bin/env python2
+
+
+#
+# NOTE: Due to some competitions without results, this script will still
+# generate a tensor with some gaps in the competition mode.
+#
+
+import csv
+import urllib
+import zipfile
+import datetime
+import numpy
+
+urllib.urlretrieve('https://www.worldcubeassociation.org/results/misc/WCA_export.tsv.zip', 'WCA_export.tsv.zip')
+zip_ref = zipfile.ZipFile('WCA_export.tsv.zip', 'r')
+zip_ref.extract('WCA_export_Results.tsv')
+zip_ref.extract('WCA_export_Competitions.tsv')
+zip_ref.close()
+
+
+# dicts
+Competitions = dict()
+Events = dict()
+Rounds = dict()
+Persons = dict()
+
+
+# Parse competition dates and assign contiguous IDs
+Competition_dates = dict()
+with open('WCA_export_Competitions.tsv', 'r') as f:
+    reader = csv.reader(f, delimiter="\t")
+    reader.next() # skip header
+    for line in reader:
+        comp_id = line[0]
+        year  = int(line[5])
+        month = int(line[6])
+        day   = int(line[7])
+        Competition_dates[comp_id] = datetime.date(year, month, day)
+
+# Assign contiguous competition IDs, sorted by date
+for comp in sorted(Competition_dates, key=Competition_dates.get) :
+    Competitions[comp] = len(Competitions) + 1
+
+# Read solve data
+with open('WCA_export_Results.tsv', 'r') as f:
+    reader = csv.reader(f, delimiter="\t")
+    data = list(reader)
+
+data = data[1:] # First line is a header
+
+# Tensor
+Results = []
+
+def getOrSetDictVal( _key, _dict ):
+    if _key in _dict :
+        return _dict[_key]
+    else:
+        _dict[_key] = len(_dict) + 1
+        return _dict[_key]
+
+for entry in data :
+    _time = 0
+
+    if int(entry[5]) > 0 :  # use average if average > 0
+        _time = int(entry[5])
+    elif int(entry[4]) > 0 : # use best if best > 0
+        _time = int(entry[4])
+
+    # FMC and MBLD have different result formats
+    if (entry[1] == '333fm') or (entry[1] == '333mbf'):
+      continue
+
+    # XXX
+    # only do 333 for now
+    if entry[1] != '333':
+      continue
+
+    if _time > 0:
+        _time = float(_time) / 100. # convert back to seconds
+
+        # A few WCA IDs have no times associated with them, so make sure that
+        # they are not added to the tensor.
+        _competition = getOrSetDictVal(entry[0], Competitions)
+        _event = getOrSetDictVal(entry[1], Events)
+        _round = getOrSetDictVal(entry[2], Rounds)
+        _person = getOrSetDictVal(entry[7], Persons)
+
+        #_result = (_competition, _event, _round, _person, _time);
+        _result = (_competition, _round, _person, _time);
+        Results.append(_result)
+
+# Save tensor file
+#numpy.savetxt("WCA_Results.tns", Results, fmt='%u %u %u %u %0.2f')
+numpy.savetxt("WCA_Results.tns", Results, fmt='%u %u %u %0.2f')
+
+def writeMap(_dict, _file) :
+    with open(_file, 'w') as f:
+        for _key in sorted(_dict, key=_dict.get) :
+            f.write('%s\n' % (_key))
+
+writeMap(Competitions, 'mode-1-competitions.map')
+writeMap(Events,       'mode-2-events.map')
+writeMap(Rounds,       'mode-3-rounds.map')
+writeMap(Persons,      'mode-4-persons.map')
+
+