Positive-month share as a consistency score input

否决评分研究来源: 混合 2026年8月19日

系统: 平台／方法论研究——不绑定单一系统。
Git 提交: c101f55

假设

Share of profitable months is a usable consistency measure for these systems as-is.

受控更改

None shipped in this form — the raw share was evaluated and rejected as a standalone input.

数据集与条件

Extracted monthly series, all 14 systems.

更改前 / 更改后

更改前

Systems posting 90%+ positive months: 9 of 14

更改后

Among them, systems with a 40%+ drawdown: 4

验证

样本外、步进、蒙特卡洛与前向结果仅在数据存在时显示。本条目不存在这些数据。

决定

否决

Grid- and martingale-style exits win almost every month by construction and give it back in one. Rewarding the raw share ranks the riskiest builds highest — the opposite of what a consistency score is for. Replaced by share × worst-month factor.

AI 参与

模型: Claude (Anthropic)
AI 做了什么: Proposed the raw share as a consistency input and then produced the counter-analysis that rejected it; the rejection was a joint decision.
人工审核: 已经人工审核并批准

AI 辅助的条目仅在人工审核后发布；未经审核的 AI 生成条目会被构建流程扣留。AI 不预测市场，也没有任何条目作此宣称。

备注

The replacement is EXP-000004.

回测的改进不代表未来的改进。实验越多，碰巧成功的概率越高——本日志的存在部分正是为了让多重检验风险保持可见。