Understanding the Strategy
I’ve taken a deeper dive into my self-created strategy to better understand how it functions and where its weaknesses lie.
Fitting to Historical Data
My strategy uses RSI (Relative Strength Index), but instead of common periods like [14, 20, 80], I’ve opted for extremely short periods like [2, 4, 96]. In hindsight, it feels like I forced these numbers to fit historical data.
Some may argue that unconventional settings like these are not advisable. However, if the win rate for 10,000 trades with [2, 4, 96] is 51.60%, compared to 49.18% for 120 trades with [14, 20, 80], I don’t see a problem choosing the former.
Not a Fractal, but a Paper Analogy
My strategy is based on the idea of “catching rebounds,” yet its performance on higher timeframes is poor. This suggests that on longer timeframes, rebounds may not actually occur.
Perhaps rebounds weren’t a valid assumption in the first place. In the end, I concluded that I cannot identify the overarching trends or patterns of rebounds on higher timeframes.
Currently, I conceptualize charts not as “waves” but as “crumpled paper.”
- Someone crumples a piece of paper.
- They then flatten it out.
- This “crumple → flatten” process happens repeatedly.
- The paper is placed on a table.
- Imagine observing the paper’s surface from a side view.
Observing from a Moderate Distance
In this scenario, if you try to predict the movement from left to right from a distance, it’s impossible to tell whether it will go up or down with certainty.
Focusing on Small Bumps
While larger trends (higher timeframes) are difficult to interpret, small bumps and dips (short-term movements) often follow somewhat predictable patterns.
By leveraging RSI and considering the larger trend just slightly, I aim to capture over 50% of these smaller patterns.
Summary
- Larger trends are hard to predict.
- Small, short-term movements tend to show consistent patterns.*
- The strategy aims to capture over 50% of these smaller bumps and dips.