def _get_candidate_bets(self, match_header):
bets_match = get_bets_match(match_header['uuid'])
bets = bets_match['bets']
candidate_bets = filter_bets(self._candidate_bet_patterns, bets)
return candidate_bets
def _update_bets_whoscored_match(match_uuid, whoscored_match):
if 'matchCentreData' not in whoscored_match:
return
bets_match = get_bets_match(match_uuid)
for i in range(len(bets_match['bets'])):
bets_match['bets'][i]['ground_truth'] = check_bet(bets_match['bets'][i], whoscored_match=whoscored_match)
bets_collection = db['bets']
bets_collection.update_one({ 'match_uuid': match_uuid }, { '$set': { 'bets': bets_match['bets'] } })
def handle(self, match_header, prediction_info, **kwargs):
prediction = prediction_info['prediction']
if prediction is None:
return
bets_match = get_bets_match(match_header['uuid'])
if bets_match is None:
return
bets = bets_match['bets']
self._handle_bets(bets, match_header, prediction_info, **kwargs)
def _update_with_betarch_match(match_uuid, betarch_match):
bets_match = get_bets_match(match_uuid)
whoscored_match = get_extended_info(match_uuid)['whoscored']
bets_dict = { tuple(bet['pattern']): bet for bet in bets_match['bets'] }
transformed_betarch_bets = transform_betarch_bets(betarch_match)
for bet in transformed_betarch_bets:
bet_pattern_tuple = tuple(bet['pattern'])
if bet_pattern_tuple not in bets_dict:
bets_dict[bet_pattern_tuple] = bet
else:
bets_dict[bet_pattern_tuple]['value'] = bet['value']
bets_dict[bet_pattern_tuple]['match_uuid'] = match_uuid
if bets_dict[bet_pattern_tuple]['ground_truth'] is None:
bets_dict[bet_pattern_tuple]['ground_truth'] = check_bet(bet, whoscored_match=whoscored_match)
bets = list(bets_dict.values())
bets_collection = db['bets']
bets_collection.update_one({ 'match_uuid': match_uuid }, { '$set': { 'bets': bets } }, upsert=True)
def _predict(self, fitteds, match_header, **kwargs):
[ corners_first_period_while_home_winning_statistic_fitted, tournament_first_period_while_home_winning_counts_means_fitted, corners_first_period_while_away_winning_statistic_fitted, tournament_first_period_while_away_winning_counts_means_fitted, corners_first_period_while_draw_statistic_fitted, tournament_first_period_while_draw_counts_means_fitted, goals_game_situation_durations_statistic_fitted, goals_frequencies_by_tournament_and_minute_data_fitted ] = fitteds
match_uuid = match_header['uuid']
bets = get_bets_match(match_uuid)['bets']
first_period_result_1_pattern = (None, 'Исход', '1-й тайм', '1')
first_period_result_1_bets = filter_bets([ first_period_result_1_pattern ], bets)
if len(first_period_result_1_bets) == 0:
return None
first_period_result_1_bet = first_period_result_1_bets[0]
first_period_result_1_value = first_period_result_1_bet['value']
first_period_result_2_pattern = (None, 'Исход', '1-й тайм', '2')
first_period_result_2_bets = filter_bets([ first_period_result_2_pattern ], bets)
if len(first_period_result_2_bets) == 0:
return None
first_period_result_2_bet = first_period_result_2_bets[0]
first_period_result_2_value = first_period_result_2_bet['value']
first_period_result_X_pattern = (None, 'Исход', '1-й тайм', 'X')
first_period_result_X_bets = filter_bets([ first_period_result_X_pattern ], bets)
if len(first_period_result_X_bets) == 0:
return None
first_period_result_X_bet = first_period_result_X_bets[0]
first_period_result_X_value = first_period_result_X_bet['value']
probabilities_sum = 1 / first_period_result_1_value + 1 / first_period_result_2_value + 1 / first_period_result_X_value
normed_first_period_result_1_probability = 1 / first_period_result_1_value / probabilities_sum
normed_first_period_result_2_probability = 1 / first_period_result_2_value / probabilities_sum
normed_first_period_result_X_probability = 1 / first_period_result_X_value / probabilities_sum
result_probabilities = np.array([ normed_first_period_result_1_probability, normed_first_period_result_2_probability, normed_first_period_result_X_probability ])
corners_first_period_while_home_winning_prediction = self._corners_first_period_while_home_winning_results_result_predictor.predict([ corners_first_period_while_home_winning_statistic_fitted, tournament_first_period_while_home_winning_counts_means_fitted ], match_header, **kwargs)
if corners_first_period_while_home_winning_prediction is None:
return None
corners_first_period_while_away_winning_prediction = self._corners_first_period_while_away_winning_results_result_predictor.predict([ corners_first_period_while_away_winning_statistic_fitted, tournament_first_period_while_away_winning_counts_means_fitted ], match_header, **kwargs)
if corners_first_period_while_away_winning_prediction is None:
return None
corners_first_period_while_draw_prediction = self._corners_first_period_while_draw_results_result_predictor.predict([ corners_first_period_while_draw_statistic_fitted, tournament_first_period_while_draw_counts_means_fitted ], match_header, **kwargs)
if corners_first_period_while_draw_prediction is None:
return None
tournament_id = match_header['tournamentId']
home_corners_counts = np.zeros(100)
away_corners_counts = np.zeros(100)
for i in range(100):
game_situation = 2
for minute in range(45):
if game_situation == 2 and (tournament_id, minute) in goals_frequencies_by_tournament_and_minute_data_fitted.data.index.tolist():
minute_probabilities = np.array([
goals_frequencies_by_tournament_and_minute_data_fitted.data.at[(tournament_id, minute), 'home_frequency'],
goals_frequencies_by_tournament_and_minute_data_fitted.data.at[(tournament_id, minute), 'away_frequency'],
1 - (goals_frequencies_by_tournament_and_minute_data_fitted.data.at[(tournament_id, minute), 'home_frequency'] + goals_frequencies_by_tournament_and_minute_data_fitted.data.at[(tournament_id, minute), 'away_frequency'])
])
probabilities = result_probabilities * minute_probabilities
probabilities /= probabilities.sum()
game_situation = np.random.choice((0, 1, 2), p=probabilities)
if game_situation == 0:
home_corners_counts[i] += corners_first_period_while_home_winning_prediction[0] / goals_game_situation_durations_statistic_fitted.statistic['home_winning_duration'].mean()
away_corners_counts[i] += corners_first_period_while_home_winning_prediction[1] / goals_game_situation_durations_statistic_fitted.statistic['home_winning_duration'].mean()
elif game_situation == 1:
home_corners_counts[i] += corners_first_period_while_away_winning_prediction[0] / goals_game_situation_durations_statistic_fitted.statistic['away_winning_duration'].mean()
away_corners_counts[i] += corners_first_period_while_away_winning_prediction[1] / goals_game_situation_durations_statistic_fitted.statistic['away_winning_duration'].mean()
elif game_situation == 2:
home_corners_counts[i] += corners_first_period_while_draw_prediction[0] / goals_game_situation_durations_statistic_fitted.statistic['draw_duration'].mean()
away_corners_counts[i] += corners_first_period_while_draw_prediction[1] / goals_game_situation_durations_statistic_fitted.statistic['draw_duration'].mean()
home_corners_prediction = home_corners_counts.mean()
away_corners_prediction = away_corners_counts.mean()
prediction = (home_corners_prediction, away_corners_prediction)
return prediction