Trading Ideas

My question is: Do you believe the results are reliable?

Today

Random Forests according to Volatility
0 - down
1 - up
Cross-validation scores: [0.73138298 0.82978723 0.78191489 0.85638298 0.80319149 0.81914894 0.79521277 0.84308511 0.82446809 0.80533333]
Mean cross-validation accuracy: 0.8090
Standard deviation of cross-validation accuracy: 0.0334
Accuracy on the test set: 0.8231

Predicted EUR/USD movement for next day: Up

Accuracy on the test set: 0.8231

1. What it means: This number tells you that your model correctly predicted the direction (up or down) of the EUR/USD movement on the test dataset 82.31% of the time.
2. Test Dataset: This is a portion of your data that was held out during the training process. It's used to evaluate how well your model generalizes to unseen data, simulating real-world conditions.
3. Interpretation: An accuracy of 82.31% is generally considered good, especially in the context of financial markets, which are known for their inherent noise and unpredictability. It suggests that your model has learned some meaningful patterns from the data and can make reasonably accurate predictions.

Important Considerations

1. Baseline Accuracy: It's helpful to compare your model's accuracy to a simple baseline. For example, if the EUR/USD went up more often than down in your dataset, a naive model that always predicts "up" might achieve a certain level of accuracy. Your model should significantly outperform such a baseline to be considered effective.
2. Confusion Matrix: While accuracy is a useful overall metric, it's also important to look at the confusion matrix to get a more detailed understanding of the types of errors your model is making (e.g., false positives, false negatives).
3. Real-World Performance: Remember that test set accuracy is an estimate of how well your model might perform in the real world. However, actual performance can vary due to market conditions, unforeseen events, and other factors.

Any scalar multiple of something random is still random. In my opinion, the markets might seem random, but they definitely aren’t. Big money moves the markets, and the key is to align yourself with those big players. When you do that, things tend to work out because you’re just following the big players, and their actions aren’t random.

We don’t know about their actions.

Oh, we can figure it out. Just keep learning, and you’ll see for yourself. I’ve got plenty of examples to show we can figure out what the big players are doing. Here’s one example I shared earlier.

Testing something else now

I will try to combine Random Forests with this signals

Random Forests:

Mean cross-validation score: 0.8231923427011036
Accuracy: 0.8388814913448736
Confusion Matrix:
[[291 95]
[ 26 339]]

If your trades are truly random, you’d expect to be right about 50% of the time and wrong 50% of the time, assuming no fees and no slippage. But in reality, there are fees and slippage, and other issues, so this doesn’t hold up.

Markets aren’t random and can be predictable. But let's say they markets are completely random. In that case, you might be able to find inefficiencies to take advantage of, but doing that would need a clear and systematic strategy, not random trading.

Basically, trading randomly won’t make money over time, even if the trades are truly random and free of software bias. By the way, true randomness is also hard to achieve with software.

Overall the week was profitable
Tested 5-7 weeks; they gave the same results.
Also the prediction caught the waterfall

If you ran the simulation again, it would probably give you a different combination. There are 2 to the 5, or 32 possible combinations. Instead of just betting on one combination, I would run 32 separate demo accounts to cover all the possible combinations. That way, you’re guaranteed to include the "correct" sequence that matches the market movement over the 5 days.

Here are the 32 combinations:

B B B B B
B B B B S
B B B S B
B B B S S
B B S B B
B B S B S
B B S S B
B B S S S
B S B B B
B S B B S
B S B S B
B S B S S
B S S B B
B S S B S
B S S S B
B S S S S
S B B B B
S B B B S
S B B S B
S B B S S
S B S B B
S B S B S
S B S S B
S B S S S
S S B B B
S S B B S
S S B S B
S S B S S
S S S B B
S S S B S
S S S S B
S S S S S

Friday was also UP
In profit

Next week we have holidays;
So we might have a range for the next week

Random dates
At least we have breakeven

Another random week


It exactly caught the rocket