How we score NGX signals differently from US stocks
Dangote Cement and Apple are both equities. Both trade on an exchange, both have volume, both have price history. But they do not trade the same way, and a model that treats them identically will fail on one of them — usually the African one.
When we built Pulse's signal engine, we made a deliberate decision: NGX equities get their own feature set. Here's what that means in practice.
The structural differences
US Equities (e.g. Apple)
- Deep liquidity, tight spreads
- Continuous price discovery
- Earnings every quarter, widely covered
- Moves with S&P500 beta
- USD denominated, FX-neutral for US investors
- Millions of retail and institutional participants
NGX Equities (e.g. Dangote Cement)
- Thinner liquidity, wider spreads
- Price moves in NGN — FX exposure is real
- Earnings irregular, less analyst coverage
- Moves with NGN stability and oil revenue
- Corporate action sensitivity (rights issues, dividends)
- Retail-dominated order flow on many names
What we changed in the model
For NGX names, Pulse adds three layers that don't exist for US stocks:
NGN FX sensitivity. A company like Seplat Energy earns USD but reports in NGN. When USDNGN moves, Seplat's reported naira earnings change even if the underlying business is flat. We model this relationship explicitly so the signal reflects the real driver.
Sector-specific liquidity gates. Some NGX names trade fewer than 500,000 shares on a typical day. A signal on a name that thin isn't useful unless we know you can actually get in and out. Our gate system checks volume conditions before any NGX signal passes.
Macro regime detection. Nigerian equities are unusually sensitive to the CBN policy cycle — rate decisions, FX window changes, and fuel subsidy adjustments all move the broad market. We track these as a regime signal that adjusts confidence thresholds on NGX signals during high-uncertainty windows.
"The signal in NGX equities is real. It just requires knowing what you're actually measuring."
What this looks like on a signal card
When Pulse fires a signal on DANGCEM.LG, the SHAP breakdown will show features like volume_ratio (how today's volume compares to the 20-day average), ngn_fx_momentum (recent USDNGN direction), and sector_flow (whether money is rotating into or out of Nigerian industrials). That's a different explanation than you'd see for an Apple signal, where earnings momentum and S&P sector rotation dominate.
That's the point. The explanation matches the instrument. If it didn't, the model wouldn't be telling you anything useful.