Set3 PPT

Detecting and Analysing
Candlestick Pattern in Stock

Market Using Deep Learning
KARTHICK RAJA V
22MDT1083
M.Sc. Data Science

Division of Mathematics
School of Advanced Sciences
Guide: Dr. Saroj Kumar Dash
Table of contents
✔Abstract
✔Aim and Objective
✔Literature review
✔Methodology
✔Reference
2
Abstract
• This project explores how integrating technology, data science, and deep learning can
enhance efficiency in the stock market. Focus on automating the identification of
patterns in candlestick charts through the application of object detection techniques.
Candlestick charts are widely employed in financial market analysis, and recognizing
patterns within them is crucial for informed decision-making in trading. The research
explores the integration of object detection models, specifically addressing how these
models can enhance the efficiency of identifying and classifying various candlestick
patterns.
3
Aim and objectives
• The project’s objective is to employ a variety of object detection models on candlestick

charts for discerning price action patterns within the financial market.
• The primary focus is on evaluating the efficacy of each model and determining their
respective performance levels.
• Additionally, a sophisticated dashboard utilizing PowerBI will be crafted to empower
traders and investors, providing them with actionable insights derived from the
detected patterns.
• This aims to enhance the profitability of trading setups through the integration of
advanced technological analyses.
4
Review of Literature
YEAR ARTICLE NAME AUTHOR METHOD USED TOOLS USED
This paper explores the

potential of applying deep
Is Deep Learning for Image learning methods like image
HyunSikSim ,HaeInKim
2019 Recognition Applicable to recognition, specifically CNN
,andJaeJoonAhn
Stock Market Prediction? Convolutional Neural Networks
(CNNs), to the challenging
task of stock market prediction.
Author proposes a new method

Encoding candlesticks as (GAF-CNN) to automatically
images for pattern Jun-Hao Chen and Yun- identify these patterns.
2020 CNN, GAF
classification using Cheng Tsai GAF-CNN first encodes price
convolutional neural networks data as images, then uses a CNN
to classify them.
5
YEAR ARTICLE NAME AUTHOR METHOD USED TOOLS USED
Xin Huang, Xinxin Wang1, Wenyu Lv,

A Practical Object Xiaying Bai1, Xiang Long1 Kaipeng PP-YOLOv2 is a new object detector that
2021 Detector Deng1, Qingqing Dang1, Shumin Han1, improves upon the previous PP-YOLO PP Yolo v2
Qiwen Liu1, Xiaoguang Hu1 Dianhai model by adding several refinements.
Yu1, Yanjun Ma1, Osamu Yoshie
This paper has two maincontributions. The

first is that the paper proposes a GASF-
CNN model to construct an innovation field
Explainable Deep of financial vision research for candlestick
Jun-Hao Chen, Samuel Yen-Chi Chen,
2020 Convolutional recognition. The second is that the paper GASF-CNN
Yun-Cheng Tsai, and Chih-Shiang Shur
Candlestick Learner proposes an approach based on modified
local search adversarial attack to explain the
reason for the GASF-CNN model on how to
determine the different candlestick patterns.
Leveraging the YOLO (You Only Look

YOLO object
Once) object detection system, the
recognition algorithm
Serdar Birogul, Günay Temür, Utku model recognizes and classifies 'Buy-
2020 and “buy-sell decision” YOLO
Kose Sell' objects on visual candlestick
model over 2D
charts, achieving a remarkable 85%
candlestick charts
success rate in decision-making. 6
WORKFLOW
Data Model Dashboard

Data Collection Model Fitting
Preprocessing Evaluation Creation
7
Data Collection
8
9
Annotation of images:
1. Install LabelImg: LabelImg is an open-source graphical image annotation tool. You can install
it by following the instructions on its GitHub repository: tzutalin/labelImg.
2. Annotate Images: Open LabelImg and load the images you want to annotate. Then, draw
bounding boxes around the objects you want to annotate and assign appropriate labels to them.
3. Export Annotations: After annotating the images, export the annotations in the desired format.
Roboflow supports various annotation formats such as COCO JSON, Pascal VOC XML,
YOLO TXT, etc. Choose the format that suits your needs.
4. PATTERNS:['Head and shoulders bottom', 'Head and shoulders top', 'M_Head', 'StockLine',
'Triangle', 'W_Bottom']
10
11
12
Loading the dataset
Choose Download Format: Select the download format that suits your needs. Roboflow
supports formats such as TFRecord, COCO JSON, Pascal VOC XML, YOLO TXT, etc.
Choose the format that is compatible with your machine learning framework or tool.
Total: 6636
Train:5318
Test:449
Val:869
13
14
YOLO
YOLO (You Only Look Once) is an object detection algorithm that operates on the
principle of dividing the input image into a grid and predicting bounding boxes and
class probabilities for objects directly from this grid. YOLO works by processing the
entire image in a single forward pass through a convolutional neural network (CNN)
and simultaneously predicting bounding boxes and their corresponding class
probabilities.
15
16
17
18
References
1. Sim, HyunSik, HaeIn Kim, and JaeJoon Ahn. “Is Deep Learning for Image Recognition Applicable to Stock Market
Prediction?” Complexity, vol. 2019, 2019, pp. 1-10., doi:10.1155/2019/4324878.
2. Chen, JH., Tsai, YC. Encoding candlesticks as images for pattern classification using convolutional neural networks.
Financ Innov 6, 26 (2020).
3. PP-YOLOv2: A practical object detector. X Huang, X Wang, W Lv, X Bai, X Long, K Deng, Q Dang, S Han, Q Liu, ...
arXiv preprint arXiv:2104.10419, 2021.
4. J. –H. Chen, C. –H. Wu, Y. –C. Tsai and S. Yen-Chi Chen, “Explainable Digital Currency Candlestick Pattern AI
Learner,” 2022 14th International Conference on Knowledge and Smart Technology (KST), Chon buri, Thailand, 2022,
pp. 91-96, doi: 10.1109/KST53302.2022.9727231.
5. S. Birogul, G. Temür and U. Kose, “YOLO Object Recognition Algorithm and “Buy-Sell Decision” Model Over 2D
Candlestick Charts,” in IEEE Access, vol. 8, pp. 91894-91915, 2020, doi: 10.1109/ACCESS.2020.2994282.
19
THANK YOU...!!
20

Set3 PPT

Uploaded by

Copyright:

Available Formats

Set3 PPT

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Set3 PPT

Uploaded by

Copyright:

Available Formats

Detecting and Analysing

Candlestick Pattern in Stock

M.Sc. Data Science

• The project’s objective is to employ a variety of object detection models on candlestick

This paper explores the

Author proposes a new method

Xin Huang, Xinxin Wang1, Wenyu Lv,

This paper has two maincontributions. The

Leveraging the YOLO (You Only Look

Data Model Dashboard

You might also like