Trading Ideas

Anything that improves trading. Start from post #20.

What do you think about AUD/USD? Signals are generated at a standard deviation of 2.
Looks promising!!!

EURUSD
2024-07 shows trend continuation
But if we look above AUDUSD it gave total reversal. Something to think about.

Also one simple idea is to hedge AUDUSD and EURUSD; but needs to be tested live to understand the behaviour.
But it's for later

Has anyone successfully predicted FX price using machine learning?
Which one is better? I have tried XGboost, LSTM and Prophet
___________________________________________________
Source: (ChatGPT)
There are several ML techniques used for stock price forecasting, depending on the type of prediction (regression or classification) and the complexity of the data:

1. Linear Regression: A basic method to model the relationship between independent variables (e.g., technical indicators) and the dependent variable (e.g., stock price). It’s a good starting point for simple problems.
2. Decision Trees and Random Forests: These algorithms can capture more complex, non-linear relationships between input features and stock prices. Random forests, in particular, work well for handling large datasets and avoiding overfitting.
3. Support Vector Machines (SVM): SVM can be used for both regression (SVR) and classification (SVC). It’s useful for forecasting when you need to classify whether the price will go up or down.
4. Neural Networks (Deep Learning): Deep learning models, particularly LSTM (Long Short-Term Memory) and GRU (Gated Recurrent Units) networks, are designed for time series forecasting and excel at capturing sequential dependencies in the data. These models are very effective for stock price prediction because stock prices are influenced by time-dependent patterns.
   1. LSTM models can "remember" past information (e.g., previous days' prices) and are particularly suited for forecasting stock prices based on historical trends.

5. XGBoost: This gradient-boosted decision tree model often performs well in structured/tabular data and can handle non-linear relationships in time series data, making it a popular choice for stock prediction.
6. Reinforcement Learning: Reinforcement learning (RL) is used when you want the model to learn trading strategies based on rewards and penalties (profit or loss). It's a more advanced technique that allows the model to simulate buying and selling decisions.

This prediction is from XGboost; it shows up momentum since 2024-11-22
Overall direction is good; but predicting fluctuations isn't easy with XGboost

Parameters changed

LSTM
Looks like overall direction is up taking into consideration XGboost

Test data shows good results; but you know it's always difficult to implement in real time.
As they say:
"It's easier to say than to do."

A litte lower TF
Analyze as above

i use NU to hedge my AU positions. Really bullish AU from a Macro perspective. #1 Iron Ore producer and #2 Gold producer, seems like from here to .6000 buy and hold for me. I have limit buys @ .6270 and .6170 respectively but bullish long term no matter the pullback.

BTC live trade

Longer term looks nice

it follows that the euro will go up

ok. dears
i m leaving this thread
you may follow me on FF on " GOLD ANALYSIS " thread. bye.like and subscribe .thanks

Trading idea with chart will help

As you can see above, I tested a few things. Now experimenting with Random Forests:
Results:
Cross-validation scores: [0.73138298 0.82446809 0.75797872 0.88031915 0.83244681 0.84574468 0.84840426 0.84042553 0.79521277 0.848 ]
Mean cross-validation score: 0.8204382978723406
Accuracy: 0.8457446808510638
Confusion Matrix:
[[302 86]
[ 30 334]]



Last Three Trading Signals:
EURUSD Signal
Date
2024-12-17 1.051735 Buy
2024-12-18 1.049505 Sell
2024-12-19 1.041558 Sell