The same number can tell different stories
Imagine that a type of event is followed by a rise 54% of the time. If comparable events usually rise 50% of the time, the observed difference is four percentage points. If the relevant market rises 60% of the time during the same conditions, the result is actually below that baseline.
The headline rate has not changed. Its meaning has. That is why an isolated win rate is rarely enough to judge a claim.
Choose the comparison to match the question
Comparing with fifty-fifty asks whether two directions appear balanced under a simple model. Comparing with the market-wide rate asks whether the feature differs from the market environment. A matched group can control for time, sector or other characteristics. These are different questions, not interchangeable technical choices.
A fair baseline should use a compatible universe, period and outcome definition. Comparing a one-day technology-stock result with a long-term all-market average would create an impressive-looking gap that answers very little.
Why Tradour keeps the comparator visible
Tradour treats source labels such as Rise and Drop as directional features, not instructions. Their historical rate gains meaning only against a declared comparator and transformation. The sample, dates and outcome definition belong alongside it.
A comparator can also be imperfect. A simple 50% marker is easy to understand, but markets do not always rise and fall equally over every period. When a richer baseline becomes available, it should be shown without silently rewriting the older analysis.