docs(spec): accumulation policy -- value averaging and band rebalancing vs static DCA (#831) - #834
Merged
Merged
Conversation
…ebalancing vs static DCA (#831) Draft for review; no code or run until approved. Declares the account model (equal $500 monthly deposits, idle cash earns nothing, allowance-aware fees: buys free to $500/month, sells always at taker), four arms with every parameter stated, the rails a live version would meet (sell_only_on_rule for trims), the KB section 51 prior against fixed-weight rebalancing, and six decisions (D1-D6) for the review. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
eaitbrahim
added
research
…strap block lengths (#831) Arm D trims only assets above their upper band, down to target, and redeploys the proceeds to underweights by shortfall; a lower-band breach alone triggers no trade. The band is max(0.15 w_i, 0.015), so the 6% sleeves trade on a 4.5-7.5% corridor, not on noise. The bootstrap reports 20- and 60-day blocks side by side, and the decision rule must hold under both, so the block length cannot be chosen after the fact. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
eaitbrahim
added a commit
that referenced
this pull request
Sep 27, 2026
…tic DCA (#831) Pre-registered (design #834; driver 4113ed0 pushed before the run, and unchanged since). On equal $500 monthly deposits with allowance-aware fees, over 5 years without PAXG and 1.4 years with it: no arm clears P(dSortino>0) >= 0.95 under both bootstrap block lengths. Band trimming leads the historical path by 13.5% but improves Sortino in only 36-43% of paths; value averaging is the most consistent at 58-61%. 16 ablation rows appended to the trials ledger. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
eaitbrahim
added a commit
that referenced
this pull request
Sep 27, 2026
…831) (#835) * feat(sim): accumulation-policy harness -- four allocators, allowance-aware account, joint stationary bootstrap, D6 rule (#831) Implements the approved design (docs/superpowers/specs/2026-09-27-accumulation-policy-design.md, revision 2) as a pure module, apart from portfolio_sim: static DCA, bounded buy-only value averaging, shortfall-steered DCA and DCA with selective upper-band trimming; an account loop with next-open fills, the monthly fee-free buy allowance (sells always taxed; a flat-taker mode), per-asset slippage and average-cost lots; a deposit-neutral TWR index; a stationary block bootstrap of joint daily rows; and the D6 verdict over both block lengths. Arm A reproduces benchmark.dca_into_allowlist to the cent under zero fees on gapless candles; the one divergence (next open vs same-day close) is pinned by its own test. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(experiments): pre-register the accumulation-policy experiment -- driver written, NOT yet run (#831) The driver's docstring is the pre-registration: spec §3-§8 frozen with D1-D6 settled, the expectation recorded, both runs (5-year PAXG-free primary to now_ts=1790455996; ~1.4-year with-PAXG secondary from PAXG's first cached daily bar), both fee modes, both bootstrap block lengths (20, 60; 2,000 paths; seed 831) and the D6 decision rule. No real candle has been run through it; it was smoke-tested only on synthetic candles in a throwaway database. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(experiments): accumulation policy after fees -- no arm beats static DCA (#831) Pre-registered (design #834; driver 4113ed0 pushed before the run, and unchanged since). On equal $500 monthly deposits with allowance-aware fees, over 5 years without PAXG and 1.4 years with it: no arm clears P(dSortino>0) >= 0.95 under both bootstrap block lengths. Band trimming leads the historical path by 13.5% but improves Sortino in only 36-43% of paths; value averaging is the most consistent at 58-61%. 16 ablation rows appended to the trials ledger. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Refs #831. A draft for review. No code and no simulation until it's approved. Merging this PR is the approval; comments or edits here are the review.
docs/superpowers/specs/2026-09-27-accumulation-policy-design.mdcovers:ΔC = clamp(w·D·t − H, C_min·w·D, C_max·w·D), with a zero-growth target path, buy-only, and C_max = 3.portfolio_sim.dca_into_allowlistexactly, as the oracle for the whole harness.sell_only_on_rulerequires arule_kindon every sale, so rebalancing trims need a policy decision. The two buy-only arms avoid that question entirely.Revision 2 (from review): selective trimming in arm D, a ±1.5-point band floor for the small sleeves, and 20- and 60-day bootstrap blocks reported side by side, with the decision rule required to hold under both.
Decisions for you (§9)
max(0.15·w_i, 0.015), monthly; selective: trim only upper-band breaches, proceeds to underweightsOnce D1–D6 are settled, §3–§8 become the driver's pre-registered docstring, committed before any run, exactly as for #830.
🤖 Generated with Claude Code