A scanner that shows thousands of price gaps has not handed you thousands of good bets. Advanced +EV filters are about deciding which candidates deserve action. This lesson lays out a five-stage pipeline: coverage, normalization, fair pricing, friction and risk review. Each stage knocks out a different kind of false signal, such as a mismapped market, a stale quote or an illiquid book. A worked example takes a +4 percent gross edge, subtracts execution cost, then shows how a three-point probability error can erase it. The top raw number is often the least trustworthy one. You leave with a way to keep your own filters visible, separate costs from model doubt, and check quote age before you click.
Learning objectives
- Explain this technique and its assumptions.
- Reproduce the worked numerical example and interpret its limitations.
- Apply a documented research workflow and decide when evidence is insufficient.
