Trading glossary
Confidence score
By the PatternX team · Published
Definition
A confidence score is a single number, often on a 0–100 scale, that summarizes how well a trading setup meets a system's criteria. It grades the setup against defined rules; it is not a measured probability that the trade will profit.
How it works
Most confidence scores are weighted sums. Each factor is scored on a common scale, such as 0–100, multiplied by its weight, and the results are added:
The weights express the designer's view of what matters. The total is then often mapped to categories so people can act on it quickly, and compared with a minimum below which setups are ignored. A trustworthy score has inspectable factors and weights, and is clear about which inputs are measured from the market and which are fixed assumptions.
Why it matters
A score turns many separate judgments into one decision: take it, watch it or skip it. It lets you hold every setup to the same bar instead of deciding case by case. The danger is false precision. A score of 82 versus 78 reflects the formula, not a measured four-point difference in the chance of winning.
Example
For example, imagine a simple three-factor score that weights pattern quality at 0.5, volume at 0.3 and trend at 0.2. A setup scoring 80, 60 and 40 on those factors totals (0.5 × 80) + (0.3 × 60) + (0.2 × 40) = 40 + 18 + 8 = 66. With a minimum of 65 it passes, just. That's a prompt to check why the trend factor was weak, not a reason to feel sure.
Common mistakes
- Reading it as a win probability. It's a grade against rules.
- Raising the minimum to feel safer. Fewer signals isn't the same as better results; test it with backtesting.
- Ignoring the breakdown. Two setups with the same score can be strong in completely different ways.
How PatternX uses it
In PatternX, the confidence score is the PatternX Score: a deterministic 0–100 total of 13 weighted factors. Pattern geometry and a fixed per-pattern reliability prior, which is not a measured statistic, make up 24% of the weight; the rest is market context such as breakout strength, volume, trend alignment, liquidity and spread. Categories are Elite ≥ 80, High ≥ 65, Medium ≥ 50 and Weak below 50. The engine floor is 50, and your minimum confidence defaults to 65, adjustable from 50 to 95. The score isn't used to sort or rank signals, and the "probability" shown with it is a heuristic derived from confidence, not a measured win probability. In PatternX AI mode, the displayed confidence is re-weighted by the AI context layer.