Teaching You to Trade Stocks 35: A Remedial Lesson for Students with Weak Foundations
2007/3/9 11:51:34
This afternoon, the media organization is holding a formation planning meeting that this ID must attend, so I'm posting the lesson early. The market trend was already indicated yesterday: "Currently there still exists the possibility of falling back into this 5-minute hub, thereby expanding into a new 30-minute hub, so 2,915 cannot be effectively broken — otherwise a new 30-minute hub will expand." This morning, a classic 5-minute top divergence made this outcome inevitable. Yesterday, after the third-type buy point, one of the two theoretical possibilities was evolution into a larger-level hub — today is a classic demonstration. The hub starts from around 2,911 at 13:00 on March 7th. After formation, the operation is the same as the 30-minute hub from the past few days — the market gives another opportunity for the same type of operation. Enough said — very busy lately, sorry for not being more attentive. Apologies.
The gap in individual comprehension ability is too large, so naturally some learn faster and some slower. Therefore, dedicating a lesson to remedial work for students with weak foundations is appropriate. And even those who consider themselves well-grounded or already understand should benefit from reading — some subtleties may not be fully grasped. The previous lessons covered two foundational aspects: first, hubs; second, trend types and their connections. These two aspects depend on each other — without trend types, hubs cannot be defined; without hubs, trends cannot be classified into types. If the theory stopped here, a circular definition would be unavoidable. To resolve this circularity, the concept of levels is indispensable. With levels, a rigorous recursive definition can be developed.
The so-called lowest level is like the quantum concept in quantum mechanics — the physical world is not naively infinitely continuous, and neither is market trading. In the strictest definition, each individual transaction is the lowest level, and three consecutive transactions at the same price constitute a lowest-level hub. A movement containing one lowest-level hub is the lowest-level consolidation trend type; a movement containing two lowest-level hubs is the lowest-level trend type — if the second hub is higher than the first, it's a rising trend type; otherwise it's a falling trend type. Generally, assuming there exist N (N>2) hubs in sequence, as long as successively the Nth hub remains higher than the (N-1)th, it's a continuation of the rising trend type; as long as successively the Nth hub remains lower than the (N-1)th, it's a continuation of the falling trend type. Obviously, based on the above definition, within the lowest-level rise, only when (and if and only if) the Nth hub is no longer higher than — i.e., equal to or lower than — the (N-1)th, can we say the lowest-level rise has ended. The lowest-level decline is the reverse of this.
The above uses the lowest-level hub to completely classify trends at the lowest level. When three consecutive lowest-level trend types overlap — meaning the price ranges they pass through have a common intersection — they form a hub of the next higher level. With this hub definition, following the same classification method used at the lowest level, trends can be completely classified at the higher level too. This process can be pushed level by level upward, allowing rigorous definition of hubs and trend types at all levels without any circular definitions. But if one were to operate by strict definitions, one would have to start from the lowest level and progressively confirm each level — too tedious and not very meaningful. That's why we have the subsequent classification of 1-minute, 5-minute, 15-minute, 30-minute, 60-minute, daily, weekly, monthly, quarterly, and yearly levels. Under this framework, one can somewhat imprecisely say that three consecutive 1-minute trend types overlapping form a 5-minute hub, three consecutive 5-minute trend types overlapping form a 15- or 30-minute hub, and so forth. In actual operations, this imprecise formulation causes no principled issues and is very convenient, so it's used — this must be made clear once again.
All of the above has been repeatedly mentioned in previous lessons, but many people still seem confused, so let me say it one final time. Obviously, standing within any fixed level, trend types can be rigorously distinguished. For example, consider a 5-minute trend type — it obviously cannot contain a 30-minute hub, because by definition, no matter how a single 5-minute trend type extends, a 30-minute hub can never appear. To form a 30-minute hub, it obviously requires the connection of 3 or more 5-minute trend types. Trend types and the connections of trend types — these two concepts could not possibly be ambiguous. A 5-minute trend type must contain and can contain at most 5-minute level hubs. Whether it's 1 hub or 5 doesn't affect its being a 5-minute trend type — it merely determines whether it's classified as 5-minute consolidation or trend type.
Obviously, a higher-level trend type is necessarily composed of several lower-level trend types connected together, but not necessarily all of the immediate sub-level. For example, in a+B+b, where B is a 30-minute hub composed of 3 five-minute trend types, and a, b are 1-minute trend types, then this a+B+b 30-minute trend type can be decomposed into the connection of 2 one-minute trend types and 3 five-minute trend types. But we can also rearrange by splitting and regrouping, so that any higher-level trend type is necessarily composed of several immediate sub-level trend types connected together. Since a hub contains at least three sub-level trend types, we get the "Chan Theory Trend Decomposition Theorem Two": Any trend type of any level consists of at least three or more sub-level trend types.
For example, using the same a+B+b example, many people probably can't see how this decomposition theorem must hold. Actually, assume B contains three 5-minute trend types, denoted B1, B2, B3. Then a+B+b = a+B1+B2+B3+b = (a+B1)+B2+(B3+b). Obviously (a+B1), B2, and (B3+b) are all 5-minute trend types — this is exactly what the decomposition theorem states. Anyone who has studied a bit of abstract algebra will easily understand the above. In abstract terms: the connection operation of trend types is associative. But the connection operation of trend types is not commutative — that's what's special about this operation. As long as you understand the associativity of trend type connections, you'll simultaneously understand the "Chan Theory Trend Decomposition Theorem One": Any movement of any level can be decomposed into the connection of three movement types of the same level: "consolidation," "decline," and "rise."
Actually, just as quantum mechanics has multiple mathematical formalisms, this ID's theory can also be treated using abstract algebraic methods — but that would make it even less accessible. And the abstract method is not only more concise but better reveals the essence. We'll discuss this later. For now, let's understand using the simpler, geometry-like approach.
Note: trends are objective, but what level you use to analyze them is subjective. According to the "Chan Theory Trend Decomposition Theorem One," any movement of any level can be decomposed into the connection of three same-level movement types — "consolidation," "decline," and "rise." This means that operating at a certain level is equivalent to forever dealing with only three same-level movement types and their connections. Using the a+B+b example again: from the 5-minute perspective, there are three connected movement types; from the 30-minute perspective, there's only one movement type. So what's been repeatedly discussed — determining your operating level — means determining at what level you analyze and operate. For instance, three 5-minute-level segments of up-down-up means at the 5-minute level there are 2 bottom divergences and 2 top divergences. Following the principle of buy at buy points and sell at sell points, that's 2 complete round-trip operations. But from the 30-minute perspective, there are no buy or sell points here, so no operations are needed.
From a purely theoretical standpoint, the lower the operating level, the higher the efficiency. But practically, the operating level can't be arbitrarily low, and what level to analyze and operate at relates to your capital and other specific conditions. For example, under T+1 rules, operating below the 1-minute level faces the risk of not being able to settle smoothly. And systematic operations must account for all possible situations, so operating entirely below the 1-minute level is impossible — unless it's T+0. Additionally, the smaller the level, the smaller the average price swing between buy and sell points. For those very small levels, the swings are insufficient to make transaction costs, execution errors, etc. small relative to the buy-sell point range — from a long-term perspective, such operations are meaningless. The so-called execution errors can include many things — for example, from the moment you see a buy point to when you actually complete the transaction, there's inevitably a time lag, and therefore a price difference. For large levels this is insignificant, but for very small levels, extreme precision is required — and this is impossible to maintain long-term.
Therefore, based on various conditions, you can accordingly determine your operating level, then analyze and operate at that level. That is, once a buy or sell point appears at your level, you must enter or exit. This means at your operating level, you do not participate in any consolidation or decline trend types. Someone once asked this ID why, for four years after 2001, I didn't look at stocks at all. Simple — because a sell point appeared at this ID's operating level, so everything was exited. Wait for the corresponding buy point to reappear. Having set the level, whether to conduct partial operations at sub-levels and below is a matter of operating style. In practice, such operations should be arranged, especially when entering a sub-level consolidation or decline of your operating level — the largest-level non-rising movement you can tolerate. Of course you should operate to lower your cost. If your operating level is very large, the sub-level's sub-level can also be used for partial operations. This gives the overall operation a certain dimensionality, further reducing risk — which is the only risk-reducing activity: lowering costs. Only when cost is zero is one truly free of risk.
Based on the "Chan Theory Trend Decomposition Theorem," it's easy to prove the "Chan Theory Buy/Sell Point Level Theorem": A large-level buy/sell point must necessarily be a buy/sell point at some level at or below the sub-level.
The proof is simple — I won't write out the details. Using the same example for illustration: a+B+b = a+B1+B2+B3+b = (a+B1)+B2+(B3+b). The final (B3+b) forms a 30-minute buy/sell point, so naturally its extreme point lies in b. Decompose b — if that extreme point isn't a buy/sell point of b, then b hasn't completed yet and must continue extending, which means this extreme point naturally can't be the extreme point of (B3+b) — a contradiction. But note: a large-level buy/sell point isn't necessarily a sub-level buy/sell point. In this example, b could be at the 1-minute level, which isn't the sub-level of the 30-minute level. So we can only say it's a buy/sell point at "some level at or below the sub-level." This is also why sometimes a 1-minute divergence can trigger a large-level decline. In the most standard movement, a large-level buy/sell point happens to be a buy/sell point at every level below it. Of course, this is still a rather rough theorem — future lessons will offer more refined ones, but that's for later.
缠中说禅 2007/3/9 11:54:01
Please leave your questions. If I'm back early tonight I'll answer them. But study the lesson thoroughly before asking — that's more efficient.
Logging off. Bye.