Skip to content

A machine learning project/exercise. Financial data was collected, cleaned, treated and used in 2 different ML models, in an attempt to predict a specific candlestick pattern used in currency trading.

Notifications You must be signed in to change notification settings

bmserafim/doji-analysis

Repository files navigation

doji-analysis

The candlestick chart is an important technical analysis tool, used by traders of different financial products. Among a variety of techniques, the investigation of specific candle formats is widely used trying to evaluate the market bias, therefore predicting future price movements.

One of these candle formats is known as 'doji', and it consists in fairly symmetrical wicks and a small body size.

In this project two different machine learning models were used to predict if the candle following a doji (under specific conditions) would close after a determined target (take profit), and if that would happen without hitting a determined stop loss.

To do that, in the first case (Doji_ML.py) data from 200.000 candles from 2020 was used, both for training and testing (splitted data). Secondly (Doji_ML_v1.py), the previous data was used only to train the models, and a separate source (GBPUSD_M15_10k_2021.csv) used to perform the testings.

All results are shown within the code.

About

A machine learning project/exercise. Financial data was collected, cleaned, treated and used in 2 different ML models, in an attempt to predict a specific candlestick pattern used in currency trading.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages