diff --git a/bdr.py b/bdr.py
index 8d57e0b..b01cbef 100644
--- a/bdr.py
+++ b/bdr.py
@@ -13,7 +13,7 @@ def analyze_game(instance: HanabLiveInstance, actions: List[Action]) -> Tuple[Li
             discard = instance.deck[action.target]
             if not game.is_trash(discard):
                 if game.is_critical(discard):
-                    termination = 'Discard crit'
+                    termination = 'Discard crit' if action.type == ActionType.Discard else 'Bomb crit'
                     break
                 if discard.rank != 1:
                     if discard in game.deck[game.progress:]:
diff --git a/bots.py b/bots.py
new file mode 100644
index 0000000..430d8db
--- /dev/null
+++ b/bots.py
@@ -0,0 +1,19 @@
+import subprocess
+
+from games import get_game_json, GAMES_PATH
+
+
+def run_strategy(game_id: int, strategy: str):
+    if strategy not in ['cheat', 'info', 'random']:
+        print('Strategy has to be one of cheat, info or random')
+        return
+    get_game_json(game_id)
+    result = subprocess.run(['./rust_hanabi', '--file', str((GAMES_PATH / str(game_id))), '--json-output', '%-' + strategy, '--strategy', strategy])
+    if result.returncode != 0:
+        print('Failed to run game {} with cheating stratgey'.format(game_id))
+
+
+def ensure_bot_game_exists(game_id: int, strategy: str):
+    filename = GAMES_PATH / '{}-{}'.format(game_id, strategy)
+    if not filename.exists():
+        run_strategy(game_id, strategy)
diff --git a/endgames.py b/endgames.py
new file mode 100644
index 0000000..db49f7d
--- /dev/null
+++ b/endgames.py
@@ -0,0 +1,106 @@
+import json
+import re
+import subprocess
+from typing import Dict, Optional
+
+from games import GAMES_PATH
+from bots import ensure_bot_game_exists
+
+from pathlib import Path
+DATA_FILE = Path('endgame-data.json')
+
+if not DATA_FILE.exists():
+    DATA_FILE.write_text('{}')
+
+
+with open(DATA_FILE, 'r') as f:
+    DATA: Dict = json.loads(f.read())
+
+
+def analyze_game(filename: str):
+    max_draw_pile_size = 15
+    try:
+        result = subprocess.run(['./endgame-analyzer', '-f', filename, '-d', str(max_draw_pile_size), '--interactive', '0', '--quiet', '-r'], stdout=subprocess.PIPE, timeout=30)
+        raw_output = result.stdout
+    except subprocess.TimeoutExpired as time_err:
+        raw_output = time_err.stdout
+    output = raw_output.decode('utf-8')
+
+    # Now, parse all results that we obtained (unclear how many depending on whether we ran into the timeout)
+    probabilities = {}
+
+    m = re.search(r"Specified draw pile size of ([0-9]+) cannot be reached with specified replay\.\nReplay ends at turn [0-9]+ with score of ([0-9]+)\.", output, re.M)
+    if m:
+        won = '100' if m.group(2) == '25' else 0
+        for draw_pile_size in range(1, int(m.group(1)) + 1):
+            probabilities[str(draw_pile_size)] = won
+
+    for m in re.finditer('Probability with ([0-9]+) cards left in deck: .*/.* ~ ([0-9.]+)', output):
+        probabilities[str(m.group(1))] = m.group(2)
+
+    return probabilities
+
+
+def full_analyze_game(filename: str):
+    max_draw_pile_size = 10
+    try:
+        result = subprocess.run(['./endgame-analyzer', '-f', filename, '-d', str(max_draw_pile_size), '-i', '0', '--all-clues', '-r', '--quiet'], stdout=subprocess.PIPE, timeout=30)
+        raw_output = result.stdout
+    except subprocess.TimeoutExpired as time_err:
+        raw_output = time_err.stdout
+    output = raw_output.decode('utf-8')
+
+    probabilities = {}
+    zero_dict = {
+        (('+' if clue_modifier >= 0 else '') + str(clue_modifier)): 0 for clue_modifier in range(-8, 9)
+    }
+    hundred_dict = {
+        (('+' if clue_modifier >= 0 else '') + str(clue_modifier)): 100 for clue_modifier in range(-8, 9)
+    }
+
+    m = re.search(r"Specified draw pile size of ([0-9]+) cannot be reached with specified replay\.\nReplay ends at turn [0-9]+ with score of ([0-9]+)\.", output, re.M)
+    if m:
+        won = hundred_dict if m.group(2) == '25' else zero_dict
+        for draw_pile_size in range(1, int(m.group(1)) + 1):
+            probabilities[str(draw_pile_size)] = won
+
+    for m in re.finditer('Probability with ([0-9]+) cards left in deck and [0-8] clues \((.[0-8])\).*: .*/.* ~ ([0-9.]*)', output):
+        if m.group(1) not in probabilities.keys():
+            probabilities[m.group(1)] = {}
+        probabilities[m.group(1)][m.group(2)] = m.group(3)
+    return probabilities
+
+
+def full_analyze_game_cached(game_id: int, strategy: Optional[str] = None):
+    key = 'all' if strategy is None else 'all-{}'.format(strategy)
+    if strategy is not None:
+        ensure_bot_game_exists(game_id, strategy)
+    cached = DATA.get(key, {}).get(str(game_id), None)
+    if cached is not None:
+        return cached
+    result = full_analyze_game(str(GAMES_PATH / str(game_id)) + ('-{}'.format(strategy) if strategy is not None else ''))
+    if key not in DATA.keys():
+        DATA[key] = {}
+    DATA[key][game_id] = result
+    save_cache()
+    return result
+
+
+def analyze_game_cached(game_id: int, strategy: Optional[str] = None):
+    key = 'normal' if strategy is None else 'normal-{}'.format(strategy)
+    if strategy is not None:
+        ensure_bot_game_exists(game_id, strategy)
+    cached = DATA.get(key, {}).get(str(game_id), None)
+    if cached is not None:
+        return cached
+    result = analyze_game(str(GAMES_PATH / str(game_id)) + ('-{}'.format(strategy) if strategy is not None else ''))
+    if key not in DATA.keys():
+        DATA[key] = {}
+    DATA[key][game_id] = result
+    save_cache()
+    return result
+
+
+def save_cache():
+    with open(DATA_FILE, 'w') as f:
+        f.writelines(json.dumps(DATA, indent=2))
diff --git a/games.py b/games.py
new file mode 100644
index 0000000..41a5264
--- /dev/null
+++ b/games.py
@@ -0,0 +1,27 @@
+import json
+
+from typing import Dict, Optional
+from pathlib import Path
+
+from hanabi.live.site_api import get
+
+GAMES_PATH = Path('games')
+
+if not GAMES_PATH.exists():
+    GAMES_PATH.mkdir(parents=True)
+
+
+def get_game_json(game_id: int, strategy: Optional[str] = None) -> Dict:
+    filename = GAMES_PATH / (str(game_id) + ('-{}'.format(strategy) if strategy is not None else ''))
+    if filename.exists():
+        with open(filename, 'r') as f:
+            return json.load(f)
+
+    if strategy is not None:
+        print('Failed to load replay of {} strategy version of game with id {}'.format(strategy, game_id))
+        return {}
+
+    game = get("export/" + str(game_id))
+    with open(filename, 'w') as f:
+        f.write(json.dumps(game, indent=2))
+    return game
diff --git a/get_sheet.py b/get_sheet.py
index 4674427..3e4a78e 100644
--- a/get_sheet.py
+++ b/get_sheet.py
@@ -3,24 +3,38 @@ import requests_cache
 import json
 import csv
 import pandas
-
-from bdr import describe_game
-from hanabi.live.site_api import get
+from typing import List, Optional, Dict
+from pathlib import Path
 
 from hanabi.database import global_db_connection_manager
+from hanabi.live.hanab_live import parse_json_game, HanabLiveGameState
+from hanabi.live.compress import link
+
+from bdr import describe_game
+from endgames import analyze_game_cached, full_analyze_game_cached
+from games import get_game_json
+
+# Init db connection
 global_db_connection_manager.read_config()
 global_db_connection_manager.connect()
 
-session = requests_cache.CachedSession('.hanab-live.cache',expire_after=300)
+session = requests_cache.CachedSession('.hanab-live.cache',expire_after=30000)
+
+OUT_PATH = Path('out')
+if not OUT_PATH.exists():
+    OUT_PATH.mkdir(parents=True)
 
 player_mapping = {
         'RamaNoVarjan': 'Ramanujan',
+        'RamaNoVarjan2': 'Ramanujan',
         'purplejoe2': 'PurpleJoe',
         'PurpleJoeVar': 'PurpleJoe',
+        'PurpleJoeVar2': 'PurpleJoe',
         'yagami_blank': 'Yagami',
         'yagami_black': 'Yagami',
         'MarkusKahlsen': 'Markus',
         'NoVarkusKahlsen': 'Markus',
+        'NoVarkusKarlsen': 'Markus',
         'spring': 'spring',
         'str8tsknacker': 'str8tsknacker',
         'novarknacker': 'str8tsknacker',
@@ -45,7 +59,8 @@ player_cols = set()
 for _, p in player_mapping.items():
     player_cols.add(p)
 
-class Entry():
+
+class Entry:
     def __init__(self, game_id, seed, score, players, bdr=0):
         self.game_id = game_id
         self.seed = seed
@@ -60,7 +75,8 @@ def get_player_games(player: str):
     if r.status_code == 200:
         return json.loads(r.text)
 
-def collect_player_games():
+
+def collect_player_games() -> Dict[int, Entry]:
     global_games = {}
     for player in player_mapping.keys():
         print('Parsing games for player {}'.format(player))
@@ -89,15 +105,33 @@ def collect_player_games():
 
     return global_games
 
+
 def analyze_games(games):
     retval = {}
     for game_id in games.keys():
-        game = get("export/" + str(game_id))
-        bdrs, termination = describe_game(game)
+        bdrs, termination = describe_game(get_game_json(game_id))
         retval[game_id] = bdrs, termination
     return retval
 
 
+def analyze_endgames(games, strategy: Optional[str] = None):
+    retval = {}
+    for game_id in games:
+        print('Analysing endgames {} of game {}'.format('with strategy {}'.format(strategy) if strategy is not None else '', game_id))
+        result = analyze_game_cached(game_id, strategy)
+        retval[game_id] = result
+    return retval
+
+
+def full_analyze_endgames(games, strategy: Optional[str] = None):
+    retval = {}
+    for game_id in games:
+        print('Analysing all endgames {} of game {}'.format('with strategy {}'.format(strategy) if strategy is not None else '', game_id))
+        result = full_analyze_game_cached(game_id, strategy)
+        retval[game_id] = result
+    return retval
+
+
 def sort_players_by_num_games(games_dict):
     nums = {}
     for _, entry in games_dict.items():
@@ -108,28 +142,88 @@ def sort_players_by_num_games(games_dict):
                 nums[col] = num + 1
     return sorted(player_cols, key = lambda col: -nums[col])
 
-if __name__ == "__main__":
-    games = collect_player_games()
-    analysis = analyze_games(games)
+
+def lookup_val(endgame_dict, clue_modifier) -> str:
+    if clue_modifier > 0:
+        for lookup in range(clue_modifier, 0, -1):
+            val = endgame_dict.get('+' + str(lookup), None)
+            if val is not None:
+                return val
+    if clue_modifier < 0:
+        for lookup in range(clue_modifier, 0):
+            val = endgame_dict.get(str(lookup))
+            if val is not None:
+                return val
+    retval = endgame_dict.get('+0', None)
+    return retval
+
+
+def make_endgame_tables(ids: List[int], strategy: Optional[str] = None):
+    endgames = analyze_endgames(ids, strategy)
+
+    postfix = '-{}'.format(strategy) if strategy is not None else ''
+    main_fname = OUT_PATH / ('endgames' + postfix + '.csv')
+
+    fieldnames = ['Game ID'] + [str(i) for i in range(1, 16)]
+    with open(main_fname, 'w', newline='') as f:
+        f.writelines([','.join(fieldnames), "\n"])
+
+    with open(main_fname, 'a', newline='') as f:
+        writer = csv.DictWriter(f, fieldnames=fieldnames)
+        for game_id, endgame in sorted(endgames.items()):
+            endgame['Game ID'] = "<a href='https://hanab.live/replay/{}'>{}</a>".format(game_id, game_id)
+            writer.writerow(endgame)
+
+    special_fname = str(OUT_PATH / ('endgames' + postfix + '_modifier_{}.csv'))
+
+    x = pandas.read_csv(main_fname)
+    x.to_html(main_fname.with_suffix('.html'), escape=False)
+
+    all_endgames = full_analyze_endgames(ids, strategy)
+    fieldnames = ['Game ID'] + [str(i) for i in range(1, 11)]
+    for clue_modifier in range(-2, 3):
+        filename = special_fname.format(clue_modifier)
+        with open(filename, 'w') as f:
+            f.writelines([','.join(fieldnames), "\n"])
+        with open(filename, 'a') as f:
+            writer = csv.DictWriter(f, fieldnames=fieldnames)
+            for game_id, endgame in sorted(all_endgames.items()):
+                # print(endgame)
+                row = {'Game ID': game_id}
+                for deck_size in range(1, 11):
+                    val = lookup_val(endgame.get(str(deck_size), {}), clue_modifier)
+                    if val is not None:
+                        row[str(deck_size)] = val
+                    else:
+                        print("WARN: No results found for game {} and deck size {}: {}".format(game_id, deck_size, endgame.get(str(deck_size))))
+                writer.writerow(row)
+
+        print('processed file {}'.format(filename))
+
+        x = pandas.read_csv(filename)
+        x.to_html(Path(filename).with_suffix('.html'), escape=False)
+
+
+def make_results_table(games, analysis):
     streaks = {}
     fieldnames = ['Game ID', 'Seed', 'Player #', 'Result', 'BDR']
     fieldnames += sort_players_by_num_games(games)
     fieldnames += ['Other']
 
-    with open('games.csv', 'w', newline='') as f:
+    with open('out/games.csv', 'w', newline='') as f:
         f.writelines([','.join(fieldnames), "\n"])
 
-    with open('games.csv', 'a', newline='') as f:
+    with open('out/games.csv', 'a', newline='') as f:
         writer = csv.DictWriter(f, fieldnames=fieldnames)
         for game_id, entry in sorted(games.items()):
             bdrs, termination = analysis[game_id]
             row = {
-                    'Game ID': "<a href='https://hanab.live/replay/{}'>{}</a>".format(entry.game_id, entry.game_id),
-                    'Seed': "<a href='https://hanab.live/seed/{}'>{}</a>".format(entry.seed, entry.seed),
-                    'Player #': entry.num_players,
-                    'Result': 'Win' if entry.won else termination,
-                    'BDR': len(bdrs),
-                    }
+                'Game ID': "<a href='https://hanab.live/replay/{}'>{}</a>".format(entry.game_id, entry.game_id),
+                'Seed': "<a href='https://hanab.live/seed/{}'>{}</a>".format(entry.seed, entry.seed),
+                'Player #': entry.num_players,
+                'Result': 'Win' if entry.won else termination,
+                'BDR': len(bdrs),
+            }
             for player in entry.players:
                 col = player_mapping.get(player, None)
                 if col is not None:
@@ -145,5 +239,99 @@ if __name__ == "__main__":
                     num_others = row.get('Other', 0)
                     row['Other'] = num_others + 1
             writer.writerow(row)
-    a = pandas.read_csv("games.csv")
-    a.to_html("games.html", escape=False)
+
+    a = pandas.read_csv("out/games.csv")
+    a.to_html("out/games.html", escape=False)
+
+
+def make_team_table(games, analysis):
+    stats = {3: {}, 4: {}, 5: {}}
+    for game_id, entry in games.items():
+        normalized_players = sorted(map(lambda player: player_mapping.get(player, 'Other'), entry.players))
+        if 'Other' in normalized_players:
+            continue
+
+        num_p = len(normalized_players)
+
+        key = ", ".join(normalized_players)
+
+        bdr, termination = analysis[game_id]
+
+        if key not in stats[num_p]:
+            stats[num_p][key] = {
+                'Games': 0,
+                'BDR': 0,
+                'Win': 0,
+                'Loss': 0,
+                'Discard crit': 0,
+                'Bomb crit': 0,
+                'Strikeout': 0,
+                'VTK': 0,
+                'Lost Endgame': 0
+            }
+
+        stats[num_p][key]['Games'] += 1
+        stats[num_p][key]['BDR'] += len(bdr)
+        if entry.won:
+            stats[num_p][key]['Win'] += 1
+        else:
+            stats[num_p][key]['Loss'] += 1
+            stats[num_p][key][termination] += 1
+
+    for num_p in range(3, 6):
+        filename = Path('out/teams_{}.csv'.format(num_p))
+        with open(filename, 'w') as f:
+            writer = csv.DictWriter(f, fieldnames=['Team', 'Games', 'Win', 'Loss', 'BDR', 'Discard crit', 'Bomb crit', 'Strikeout', 'VTK', 'Lost Endgame'])
+            writer.writeheader()
+            for team, row in sorted(stats[num_p].items(), key=lambda item: -item[1]['Games']):
+                row['Team'] = team
+                writer.writerow(row)
+
+        x = pandas.read_csv(filename)
+        x.to_html(filename.with_suffix('.html'))
+
+
+def create_replay_links(ids: List[int], strategy: str):
+    outfile = Path('out/{}_links.csv'.format(strategy))
+    with open(outfile, 'w') as f:
+        writer = csv.writer(f)
+        writer.writerow(["Game ID", "Result", "{} Replay Link".format(strategy)])
+        for game_id in ids:
+            replay = get_game_json(game_id, strategy)
+            instance, actions = parse_json_game(replay)
+            game = HanabLiveGameState(instance)
+            for action in actions:
+                game.make_action(action)
+            bdrs, termination = describe_game(replay)
+            writer.writerow([game_id, 'Win' if game.is_won() else termination, link(game)])
+
+    x = pandas.read_csv(outfile)
+    x.to_html(outfile.with_suffix('.html'), render_links=True)
+
+
+def main():
+    games = collect_player_games()
+    analysis = analyze_games(games)
+    game_ids = sorted(int(key) for key in games.keys())
+
+    # This is the main table, tracking streaks, loss reasons, BDRs
+    make_results_table(games, analysis)
+
+    # This tracks team statistics
+    make_team_table(games, analysis)
+
+    # Additional endgame analysis stats:
+    # For the real games
+    make_endgame_tables(game_ids, None)
+    # For the cheating strategy
+    make_endgame_tables(game_ids, 'cheat')
+    # For the information strategy
+    make_endgame_tables(game_ids, 'info')
+
+    # Create JSON replay links to the bot games to watch
+    create_replay_links(game_ids, 'cheat')
+    create_replay_links(game_ids, 'info')
+
+
+if __name__ == "__main__":
+    main()