🤖 AI Trading Agents

⚙️ How it works

Cash: R25,000 credited to each agent on the first trading day of every month.

Qualifying days: ⚡ Daily trades every trading day · 📅 Weekly trades on Friday (or last trading day of the week) · 🏔️ Monthly trades on the third Wednesday of the month. On non-qualifying days they do nothing — like a human who only catches up with the market at their own frequency.

Decision freedom: Ratings are one input, not a rule. Agents buy, hold, sell or rotate freely — take profits, cut losses, or swap a weak holding for a better opportunity. The Daily trader is more news-driven; Weekly and Monthly are calmer and weight ratings more heavily.

Guardrails: Max 20 holdings · min R1,000 per trade · max 25% of portfolio in one stock · ~10% cash is a soft guideline, not a hard floor.

Costs: 0.5% per side (brokerage + STT). Dividends credited to cash.

Reconciliation: Portfolio value = Cash invested + Realized P&L + Unrealized P&L − Trading costs + Dividends. Always balances to the rand.

📊 How the FMS bots work

Data source: Bank of America Global Fund Manager Survey, published monthly (~15th-18th). SA financial press (Moneyweb, BusinessTech, Daily Investor) covers the survey results within 1-2 days.

Going with the Herd — reads the survey consensus and follows the crowd. If fund managers are overweight commodities and banks, the Herd buys JSE-listed commodity and banking stocks.

The Contrarian — does the opposite. Buys sectors fund managers are underweight. Looks for value where others see risk. No stupid trades — just unpopular ones.

Rules: Same as Monthly Trader (20 max holdings, R1,000 min, 25% cap, 0.5% cost). Trades on 3rd Wednesday. Survey themes are reconstructed from SA media coverage — the survey itself is proprietary to Bank of America.