Session goal: Execute the planned dividend reinvestment study (DRIP vs BH) — answering V2 question “Evaluate dividend reinvestment impact on total return” from trade-pattern-v1.md. The script test_dividend_reinvestment.py was written but had bugs; needed to be fixed and run.
Context:
- Phase 1 complete: Buy-and-hold + quarterly rebalancing is optimal (arch-phase-1-conclusions.md)
- trade-pattern-v1.md lists “Evaluate dividend reinvestment impact on total return” as next V2 step
- Previous session 2026-05-22 planned this exact research but didn’t execute it
Progress log:
- 00:30 — Read project index, arch-phase-1-conclusions.md, trade-pattern-v1.md. Confirmed Phase 1 is complete and DRIP study is the pending V2 research item.
- 00:45 — Reviewed existing
test_dividend_reinvestment.pyscript. Found two bugs: (a) yfinance Dividends column wasn’t being extracted from history, (b) pandas str accessor bug on float column. Also noted no results file existed at expected path. - 01:15 — Rewrote and ran corrected DRIP vs BH simulation on 13 ASX large-caps (BHP, RIO, FMG, WOW, CBA, ANZ, NAB, WBC, TLS, MQG, QBE, CSL, REA). Period: 2015-2026. $2K initial capital.
- 01:30 — Analyzed results and wrote findings.
Results summary:
| Metric | Value |
|---|---|
| Universe | 13 ASX large-caps |
| Period | 2015-2026 (11.4 years) |
| Avg BH price return | +369.3% |
| Avg DRIP total return | +854.8% |
| DRIP edge over BH price-only | +485pp |
| Avg dividends reinvested per ticker | AUD $4,350 |
| Avg extra shares via compounding | +69.3% |
| Avg fees paid (23 trades × $9.90) | AUD $227.70 |
Key findings:
-
DRIP provides substantial compounding benefit on ASX compounders. The average DRIP strategy outperforms pure price BH by ~485pp over 11.4 years. This is larger than expected given that ASX compounders aren’t typically known for high yields.
-
Mining stocks dominate the compounding effect. FMG.AX (+2014% BH, +174% extra shares), BHP.AX (+413%, +103% extra), and RIO.AX (+557%, +102% extra) drive most of the DRIP benefit. High dividend payments from commodity supercycles create massive compounding opportunities.
-
Fees are trivial relative to benefit. Average fees per ticker: AUD $227 over 11 years — less than 1% of total return. The transaction cost drag is negligible compared to the compounding benefit.
-
Dividends do NOT explain IDEA-002’s +253pp dividend-screen alpha. The high-dividend screen worked because it captured miners at peak commodity profitability (yield rotation), not because reinvestment alone generates massive returns. Without the yield-selection effect, DRIP adds ~486pp over BH — significant but not enough to replicate IDEA-002’s edge on its own.
-
Low-yield compounders benefit less. CSL.AX (+34% BH, only +13% extra shares) and REA.AX (+286% BH, only +10% extra shares) show that low-dividend stocks get minimal compounding benefit from reinvestment. The DRIP effect scales with absolute dividend payments.
What this means for Phase 2:
- DRIP should be implemented — the ~486pp edge is substantial and comes “free” (minimal fees). Any Phase 2 execution layer should default to DRIP-style reinvestment on ex-dates.
- The V1 strategy needs refinement. The current “quarterly rebalance + BH” doesn’t account for dividend reinvestment mechanics. Adding DRIP would boost expected returns significantly.
- Fee structure matters less than we thought. At 227/yr) are <5% of dividend income for high-yield miners.
Outputs:
/opt/data/wiki/projects/asx-trading/results/dividend_reinvestment_study.json— full results data- Session log: this file (2026-05-23.md)
Next V2 action: Update trade-pattern-v1.md to incorporate DRIP mechanics. The “Evaluate dividend reinvestment impact on total return” question is now answered with evidence.
Status: done