BtcSpotWeeklyFixedDcaAccumulation

BTC Spot Weekly Fixed DCA — Pure Time-Based Accumulation (No Signal, No Boost)

The simplest possible crypto strategy: BUY a fixed dollar amount of BTC every Monday at 00:00 UTC, HOLD permanently, NEVER SELL. This is pure mechanical Dollar-Cost-Averaging with NO signal, NO indicator, NO drawdown trigger, NO volatility sizing, NO exit — just constant weekly accumulation. RATIONALE driven by the now-DECISIVE session finding (29+ strategies attempted, 19+ abandonments across 8 distinct mechanism classes; the analyst's explicit recommendation: 'HALT new hypothesis generation for at least 2-4 weeks'). The honest acknowledgement is that no signal-based strategy mechanism this session has produced edge. The ONLY remaining strategy type that DOESN'T DEPEND ON A SIGNAL is pure passive accumulation — a strategy that cannot be 'wrong' about regime/direction/dispersion because it makes no prediction. KEY STRUCTURAL DIFFERENCES from all 29+ failed strategies: (a) ZERO directional or relative-value signal — the strategy buys regardless of price, time-only entry rule; (b) NO EXIT EVER — bought BTC is held permanently, never sold; (c) MAXIMALLY SIMPLE — exactly 1 parameter (weekly DCA amount); (d) FEE-FRIENDLY — 52 buys/year × 0.10% taker = ~0.5% annual fee cost, negligible vs expected ETH/BTC returns; (e) Differentiated from existing BTC accumulation strategies in pipeline: BtcSpotDrawdownAccumulationLong (uses drawdown TRIGGER, only buys during drawdowns) and BtcFuturesVolatilityTargetedLongOnlyAccumulationDaily (uses vol SIZING modulation, futures venue). This proposal is the SIMPLEST CASE: constant weekly buys on spot, no sophistication. PRE-COMMITTED ABANDONMENT RULES (in honor of the analyst's praise for this practice): (1) If pre-opt Sharpe < 0.3, abandon immediately — no iteration. (2) If 2025-2026 cumulative return is negative, abandon — BTC long-run drift assumption has broken in current regime. (3) If max DD > 60%, abandon — implies BTC itself has lost positive expectancy. These rules prevent post-hoc rationalization.

Implements the hypothesis exactly: the only strategy type that depends on no signal — constant time-based accumulation that cannot be wrong about regime/direction/dispersion because it makes no prediction. The base entry-when-flat / exit-when-in-position flow cannot accumulate (it would stop buying once the first position opens), so on_bar is overridden to submit a fixed-dollar market BUY on every weekly bar, and should_exit is hard-wired to False (never sell). 1-WEEK bars encode the Monday-00:00-UTC schedule directly — one buy per weekly bar — with no day-of-week arithmetic. Routing to BINANCE_SPOT is correct per the venue rules: long-only, no leverage, no funding edge → spot CASH account (leverage 1, ~0.10% taker). On a CASH account get_account_equity() returns the stable USDT balance, which is the deployable cash and shrinks as it is spent on BTC, so capping each buy at remaining cash front-loads deployment and prevents rejected unfundable orders. It is structurally distinct from the pipeline's drawdown-triggered and vol-sized BTC accumulators — this is the maximally simple constant-buy case. 52 buys/year × 0.10% taker ≈ 0.5% annual fee drag, negligible vs BTC drift. The pre-committed abandonment rules (pre-opt Sharpe < 0.3, negative 2025-2026 return, or max DD > 60%) are evaluated downstream by the analyst on the backtest output.

Pure time-based DCA is not optimizable and carries no edge. (1) Its single parameter (weekly dollar amount) is a linear scaler that leaves every risk-adjusted metric unchanged, so the 3-phase optimizer has no surface to work on. (2) By design it makes no prediction and is ~99% long BTC held forever — it is pure market beta with zero alpha vs the buy_hold benchmark, not a promotable edge. (3) The backtest is degenerate (total_trades=1, single daily-return point, metrics_reliable=false, total_return 47.6% implausibly low for BTC 2018-2026), and cash exhausted after ~100 buys so it degenerated into front-loaded buy-and-hold. Even a corrected perpetual-DCA implementation would still have nothing to optimize and no edge. Failure pattern: passive-beta / no-signal accumulation — not a strategy. Recommend the Research Lead honor the session's own halt note rather than route beta vehicles into the optimizer.

Pure mechanical weekly Dollar-Cost-Averaging on BTCUSDT Binance Spot. On every weekly bar (Monday 00:00 UTC open) it buys a fixed $1,000 of BTC and holds permanently — no signal, no indicator, no drawdown trigger, no volatility sizing, and no exit (never sells). Each buy is capped at the remaining USDT cash so it deploys the CASH account steadily without unfundable orders. Long-only spot, leverage 1, single parameter (weekly DCA amount). Tests only the BTC long-run drift assumption.

Functions mechanically (does submit BTC buys) and is fee-light as claimed

Nothing to optimize: the sole parameter (dca_amount) is a linear position-size scaler that cannot change Sharpe, drawdown%, profit factor, or any risk-adjusted metric the optimizer evaluates

Zero alpha by construction: no signal, no prediction, ~99% long BTC held forever — pure market beta vs the buy_hold benchmark, not a strategy edge

Degenerate/unreliable backtest: 1 trade, 1 daily-return point, metrics_reliable=false, Sharpe unreliable; total_return 47.6% over 2018-2026 is implausibly low for holding BTC, indicating broken equity-curve tracking for the never-sell position

Cash exhausted after ~100 buys (trades_count 101 vs 405 weekly bars) — degenerated into front-loaded buy-and-hold, not the intended perpetual DCA

BtcSpotWeeklyFixedDcaAccumulation was deliberately the simplest case — fixed weekly BTC buys, never sell, one parameter — proposed as the only remaining 'cannot be wrong' option after a session of repeated signal-based abandonments. In practice the backtest came back degenerate and unreliable (1 trade, single return point, Sharpe 0.0, 46.8% max DD, an implausible 47.6% total return), and the implementation exhausted its cash after ~100 buys, collapsing into front-loaded buy-and-hold. The analyst abandoned it at backtest review, noting its single parameter is a linear scaler that leaves every risk-adjusted metric unchanged and that the strategy is pure BTC beta with zero alpha versus buy-and-hold. The recorded failure pattern is that passive-beta, no-signal accumulation is not a strategy and should not be routed into the optimizer.

Pure no-signal passive accumulation is market beta, not a strategy — it has no parameter surface to optimize and no alpha to promote, so reaching for a beta vehicle is not a substitute for finding edge (the analyst recommended honoring the session's own halt note).

The analyst issued an 'abandon' verdict at the pre-optimization backtest-review gate: the lone parameter is a linear position-size scaler that cannot change any risk-adjusted metric (nothing for the optimizer to work on), and by construction the strategy is ~99% long BTC forever — pure market beta with zero alpha versus the buy-and-hold benchmark, not a promotable edge.

The simplest possible strategy: buy a fixed dollar amount of BTC on every weekly spot bar and hold permanently with no signal, no exit, and a single parameter (the weekly DCA amount), explicitly proposed as a no-prediction passive-accumulation test after a session where 19+ signal-based strategies had been abandoned.

The backtest was degenerate and flagged unreliable (metrics_reliable=false): it recorded only 1 trade and a single daily-return point, an unreliable Sharpe of 0.0, profit factor 0.0, 46.8% max drawdown, and an implausibly low 47.6% total return over 2018-2026; cash was exhausted after ~100 buys (101 trades vs 405 weekly bars), so it degenerated into front-loaded buy-and-hold rather than perpetual DCA.