Is the edge a ridge, or one lucky cell?
Anyone can find a moving-average crossover that looks brilliant on one setting. Algorythm runs the whole grid — every fast-over-slow window pair, on two weeks of real candles — and paints the return surface. A genuine edge shows up as a broad bright region you could have landed on by accident; an overfit one is a single island surrounded by losses.
Return surface — SMA crossover
What a sweep tells you
One backtest is a single sample from a noisy process. Scanning every window pair at once turns the question from "does this setting work?" into "does the neighbourhood work?" — a far harder thing to fake.
Reading the heatmap
Each cell is one complete backtest, coloured by its result. Indigo is profit, rose is loss, depth is magnitude. The gold-starred cell is the grid's best. Look for smooth gradients, not lone bright squares next to dark ones.
Honest limits
Fourteen days of one token is thin, and the sweep ignores fees, slippage and the yield forgone sitting in cash. Signals use only prior closed bars — no lookahead — but this is a research surface, not a trade plan.