KILLED · curve-fit
Hypothesis
The Ichimoku Kinko Hyo system (cloud, conversion/base lines, lagging span) generates profitable trend signals.
Math
$$ \text{Tenkan}=\tfrac{\max_9 H+\min_9 L}{2},\ \ \text{Kijun}=\tfrac{\max_{26}H+\min_{26}L}{2} $$
Method
Full Ichimoku ruleset (cloud breakouts + TK cross + Chikou confirm) backtested across 40 symbols and parameter variants.
Results
| Default (9,26,52) Sharpe | 0.2 |
| Parameter sensitivity | extreme |
A bundle of moving-average crossovers with five tunable lookbacks — once you test the parameter grid honestly, performance is indistinguishable from a generic MA system and highly fit-dependent. No special edge. Killed.
Exotic-sounding indicator systems usually decompose into moving averages with more knobs. More knobs means more overfitting, not more edge.