PHASE 1 / INTRODUCTION
Crypto attracts me because it is volatile, observable, and structurally rich. The same traits also make it dangerous for research: the market is noisy, reflexive, and very easy to overfit.
Crypto 吸引我，是因为它 高波动、可观察，市场结构也足够丰富。但这些特点也会让研究变得危险：噪音很多、反身性很强，也很容易过拟合。
I start with low frequency for a simple reason: the work should rely on judgments I can explain and audit, before it relies on infrastructure I do not have or plan to build.
我从低频开始，原因很直接：这一阶段的研究应该先依赖我能解释、能复盘的判断，而不是先去依赖一套我目前没有、也不准备搭建的交易基础设施。
In Crypto, the closer the game moves to order books and high-frequency execution, the more it favors teams with lower fees, faster routing, better connectivity, and stronger inventory systems. A personal machine or a rented server does not change that asymmetry.
Crypto 里机构参与度很高。越靠近高频和盘口，竞争就越依赖低费率、快速路由、稳定连接和库存管理能力。一台普通电脑，或者几台云服务器，并不会让我在这些地方突然拥有优势。
Short windows are tempting because they always contain moves that look meaningful. It is easy to turn those moves into a trading story, especially once a plausible logic can be attached. But many short-term price changes come from liquidity shocks, liquidation flows, spread distortion, or market-maker inventory adjustment. They may explain a moment in the market without becoming a repeatable signal.
短周期里总能看到很多像“规律”的波动。对个人投资者来说，这些波动很容易被理解成信号，尤其是当自己能讲出一套交易逻辑时。但很多短期价格变化，其实来自流动性冲击、强平流、价差扭曲，或做市商库存调整。它们可以解释某一段行情，却未必能成为可重复的研究对象。
That is why the first phase avoids competing in an institutional lane: order-book access, fee tier, routing speed, API limits, dedicated servers, fast quote-and-cancel loops, and inventory hedging. These are real advantages, just not the advantages I currently have.
所以第一阶段我不想把研究推进到机构更擅长的层级：订单簿接入、费率等级、路由速度、API 限额、专属服务器、快速报价撤单和库存对冲。这些都是真实优势，只是离我现在的研究条件太远。
Daily and weekly signals give me a more suitable starting point. The data is easier to check, the execution is slower, and the full test can still be reviewed after the fact.
日频和周频信号更适合作为起点。数据更容易检查，执行节奏也更慢，整套测试做完以后还能回过头来复盘。
This does not make Crypto easy. It only narrows the first question to something I can test seriously: can a slower signal remain useful after costs, execution delay, drawdown, and market-regime changes, without relying on speed, order-book position, or complex infrastructure?
这不会让 Crypto 变简单，只是让第一阶段的问题更适合被认真检验：一个更慢的信号，不依赖速度、盘口位置和复杂基础设施，经过成本、执行滞后、回撤和市场状态变化之后，是否仍然有效？