EGX Scanner — Semsem

Tested — and the honest result is mixed. Read this before using it. Backtested across 282 EGX listings on weekly bars (10y+ of daily history, resampled; liquidity-filtered; entry at the next weekly open; net of an estimated 30bps round-trip cost). The rule returned about +20% per trade over a ~17-week median hold — but the market returned about as much over those same weeks. Measured against the proxy EGX30 index across each trade's own holding period, the excess is about −1%, and only ~27% of trades beat the market. In other words this no longer shows even a modest edge over simply holding the index at the moment of the cross — the positive average return rests entirely on a small number of large winners. Treat this as a way of participating in EGX uptrends, not as a stock-selection edge.

The "cross must be recent" rule still does not survive testing — more clearly now. Entering 3, 8 or even 12 weeks after the cross performed as well or better than entering immediately (every delayed cohort beat the at-cross excess; 12 weeks was the best of all). The age and run-up controls above are therefore offered as a way to narrow the list to something readable — not because the evidence says fresher crosses do better. One earlier finding did not hold up on re-test: crosses already up more than 75% used to look like the weakest cohort, but the latest run shows the opposite — that group is now the single best by rate of beating the market. That reversal isn't well understood either, and shouldn't be read as a new rule in its place.

This strategy defines no stop-loss. Its only exit is the opposite weekly cross, which can take many weeks to arrive, so drawdown between weekly closes is unbounded. The ATR column is a position-sizing reference, not a stop this strategy places. Rows describe current indicator state, never a buy or sell instruction. Regime splits use a proxy EGX30 (an equal-weighted basket composite), not the real index. Results are in-sample plus a single walk-forward split, on a present-day constituent list, so survivorship bias is not resolved — and it bites harder here than elsewhere, because a multi-week holding period is exactly where a delisted name's collapse would have landed. A • next to a price means the current week hasn't closed yet.