Analysis Of My Lichess Bullet Games

2023-10-16 639 words 3 minutes

Hello Guys,

In this post we will analyze some of my bullet games on Lichess. I love playing chess, and I try to play bullet games almost every day, at least 3-4 games. I thought, it is time to analyze some of my games with Python. To be able to do it, we need to use the Lichess API to export my games. There is already a client package implemented in Python3 called Berserk, so I will be using it.

We will use pandas to manipulate the data and matplotlib to plot some charts (hopefully we will get some meaning based on them).

I have more than 5000 games as of the data 16th Oct 2023. We will be analyzing last 1000 games.

Let’s begin.

I got help from ChatGPT while creating this notebook.

Fetch the Games

1
2
3
4
import berserk
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt 

1
2
3
4
with open('./lichess.token') as f:
    token = f.read()
session = berserk.TokenSession(token)
client = berserk.Client(session)

1
client.games.export_by_player('IsaacNewton29', opening=True, perf_type="bullet", max=1000)

<generator object Games.export_by_player at 0x1455dbcf0>

1
games = list(_)

1
2
3
df = pd.DataFrame(games)
df = df.drop(columns=['id', 'rated', 'variant', 'speed', 'perf', 'clock', 'lastMoveAt'])
df.head(2)

	createdAt	status	players	winner	opening	moves
0	2023-10-16 17:34:35.325000+00:00	mate	{'white': {'user': {'name': 'Chessington008', ...	white	{'eco': 'B01', 'name': 'Scandinavian Defense: ...	e4 d5 exd5 Qxd5 Nc3 Qd8 Bc4 Nc6 Nf3 Nf6 d4 e6 ...
1	2023-10-16 14:06:40.830000+00:00	outoftime	{'white': {'user': {'name': 'IsaacNewton29', '...	black	{'eco': 'C00', 'name': 'French Defense', 'ply'...	e4 e6 Bc4 d5 exd5 exd5 Be2 c6 Nf3 Nf6 d4 Bd6 N...

1
df.createdAt.min().date(), df.createdAt.max().date()

(datetime.date(2022, 12, 15), datetime.date(2023, 10, 16))

It seems like I played the last 1000 games between the dates 12th December 2022 and 16th October 2023.

Let’s extract the white and black players for each game. It will be useful

1
2
df["white"] = df["players"].apply(lambda x: x["white"]["user"]["name"])
df["black"] = df["players"].apply(lambda x: x["black"]["user"]["name"])

1
df[df["white"] == "IsaacNewton29"].shape

(490, 8)

1
df[df["black"] == "IsaacNewton29"].shape

(510, 8)

1
df[((df["white"] == "IsaacNewton29") & (df["winner"] == "white")) | ((df["black"] == "IsaacNewton29") & (df["winner"] == "black"))].shape

(495, 8)

1
df[(df["black"] == "IsaacNewton29") & (df["winner"] == "black")].shape

(224, 8)

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
# Count the occurrences of each category
category_counts = df['winner'].value_counts()

# Plot a bar chart with counts displayed on top of each bar
plt.figure(figsize=(8, 6))
bars = plt.bar(category_counts.index, category_counts.values, color='skyblue')

# Add counts as labels on top of the bars
for bar in bars:
    yval = bar.get_height()
    plt.text(bar.get_x() + bar.get_width()/2, yval, int(yval), ha='center', va='bottom')

plt.xlabel('Player')
plt.ylabel('Count')
plt.title('Winner Counts Based on Color')
plt.xticks(rotation=0)  # Rotate x-axis labels if necessary
plt.show()

Let’s create a new column to check if I won that game. To get the value

If the winner is white check if I play white
If the winner is black check if I play black

1
2
df["did_i_win"] = ((df["white"] == "IsaacNewton29") & (df["winner"] == "white")) | ((df["black"] == "IsaacNewton29") & (df["winner"] == "black"))
df["did_i_win"].value_counts()

did_i_win
False    505
True     495
Name: count, dtype: int64

Let’s see on the chart to understand it better.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
# Count the occurrences of each category
category_counts = df['did_i_win'].value_counts()

# Plot a bar chart with counts displayed on top of each bar
plt.figure(figsize=(8, 6))
bars = plt.bar(category_counts.index.values, category_counts.values, color='skyblue')

# Add counts as labels on top of the bars
for bar in bars:
    yval = bar.get_height()
    plt.text(bar.get_x() + bar.get_width()/2, yval, int(yval), ha='center', va='bottom')

plt.xlabel('Is the Game a Win')
plt.ylabel('Count')
plt.title('Did I Win the Game')
plt.xticks(range(len(category_counts.index)), category_counts.index)  # Set x-tick labels
plt.show()

It seems like I lost 505 games whereas I won the 495 games, it is almost 50/50. No improvement at all 😄

It will be continued…

Analysis Of My Lichess Bullet Games

Lichess Server Crash Image

Fetch the Games