Source: Qilehui.

Is it really an execution problem?

One day, you find that a company you are interested in has a new information regarding supply and demand changes in the industry. Based on your understanding, there will be a supply-demand reversal in the next month, leading to a significant increase in product prices.

However, you also feel that based on past experiences, changes in supply and demand do not necessarily lead to price increases, or may not increase so quickly. It may be better to wait and see if there are more signs to further confirm this judgment.

But the market is very efficient. You see it, others see it too, and they act first. The market trend comes very quickly, and you see that the stock price has skyrocketed. If you hesitate again, you will miss the opportunity to enter.

So you regretfully think that you have identified this investment opportunity correctly, but your execution ability is too weak.

So you come to a conclusion: the next time you encounter this situation, no matter what happens, get on board first and then think about it.

Is it really because of your own weak execution ability?

In this article "Bayesianism: Three Methods for Investment Masters" on "Bayesian Probability", I analyzed the probabilistic methods of information processing in investment research. If you only qualitatively analyze the bullish and bearish information of the information, you will find a contradiction. All stocks are so complicated that they have a lot of bullish and bearish factors, while investments are so simple that they are just buying and selling.

How to use complex analysis to guide simple buying and selling operations? The end goal of investment analysis is probability.

However, many investors are psychologically resistant to the method of "probability analysis" because the probability in actual investment is often estimated by unreliable methods. This kind of probability analysis is very different from the statistical and probability calculation processes taught in school and is very suspicious.

This is the psychological barrier we need to overcome when using Bayesian methods to distinguish classical probability from Bayesian probability.

Second, classical probability and Bayesian probability

There is a saying in A shares that "in the fifth year, it is poor; in the sixth year, it is desperate; and in the seventh year, it reverses". This can also be verified by statistical data. In the past 20 years, the Shanghai Composite Index and the Shenzhen Component Index have risen in July 11 times, accounting for 55%.

The "probability of rising in July in the past 20 years" here is a classical probability, also known as the Neyman-Pearson statistical probability, which is a statistical result of data for a repeatable event and is also taught in middle school mathematics textbooks.

But what investors really want to know is the "probability of the rise and fall of July this year", which requires Bayesian probability. It differs from classical probability in three ways:

Firstly, "July" is a repeatable event, but "July this year" is a non-repeatable event, so the probability of rising and falling in July this year cannot be measured by classical probability;

Secondly, when July this year ends, it will either rise or fall. It cannot have a 55% probability of rising;

Thirdly, when July this year ends, its rise and fall will make the "probability of rising in July in the past 20 years" become 52% or 57%. This is the relationship between two probabilities.

In summary, the "probability of rising in July in the past 20 years" is an objectively calculated result, while the "probability of the rise and fall of July this year" is a subjective belief, directly related to your judgment ability.

Of course, the "probability of rising in July in the past 20 years" still has some guiding significance for the "probability of the rise and fall of July this year". Therefore, in Bayesian algorithm, taking the "probability of rising in July in the past 20 years" as a conditional probability can turn your prior probability of judging "the rise and fall of July this year" into a posterior probability, which is the core content of that public account article.

There are numerous probabilities similar to "the probability of rising in July for the past 20 years," such as macro analysis results, graphical analysis, and so on. Each one can be regarded as a conditional probability, which can be substituted into Bayesian calculation to obtain new posterior probabilities.

No matter how much information you find, the probability of "whether the market will rise or fall in July this year" cannot be a certain number. The Bayesian process is similar to detective reasoning rather than scientific calculation, and may not necessarily give a 100% correct conclusion. This is like a judge cannot satisfy both the defendant and plaintiff, so many people cannot accept this probability.

The invention date of Bayesian probability was in 1748, much earlier than the classic probability of Neyman-Pearson statistics born in the late 19th century. However, because it was too counterintuitive, it was regarded as "unscientific" and excluded from the scientific community for a long time until the rise of computer science in the mid-20th century.

Bayesian probability is accepted not because it is "correct," but because it is "useful." The three-door problem listed in the previous section is not necessarily correct when switching the door, but it is "useful." Someone simulated the results with a computer to prove that the number of times someone gets a car when switching the door is twice as many as not, and someone designed a gambling game based on the three-door problem and won a lot of money from people who did not accept Bayesian probability.

The calculation of Bayesian probability may not be accurate, but it is definitely useful. The uncertain "probability of whether the market will rise or fall in July this year" is far more important than the certain "probability of rising in July for the past 20 years."

For example, we can use Bayesian methods to analyze the case at the beginning of the article.

III. Investment does not require executive power.

The morning weather report said that the probability of precipitation today is 30%--how was this 30% calculated?

Since today is not repeatable, this probability is not a classical probability and is not statistically determined. Specifically, it is Bayesian probability that has been calculated by "counting after statistics."

The meteorologist first counts all the weather conditions in history that have the same meteorological conditions as today (temperature, pressure, humidity, etc.), and the proportion of rainy weather is 35%, which is the prior probability.

Then the meteorologist adjusts these data based on experience, and the experience is a "conditional probability." One is to consider more meteorological conditions that affect precipitation, and the other is to consider the great changes in climate in recent years. Whether the weather without precipitation has similar climate conditions as today.

The final probability of precipitation today is 30%. It represents the confidence of an experienced meteorologist announcing whether there will be precipitation today based on historical data and subjective conjecture.

You can even think of the Meteorological Bureau as a casino and upgrade this probability to odds. Essentially, the probability of precipitation is the same as the odds of a certain team.

The calculation method of probability in investment is similar to that of precipitation probability:

1. Add XXXX to the watchlist, representing that the prior probability of the company has reached the attention level (such as 60%);

2. By adding the company to the watchlist, you can pay attention to the company's new information more quickly. Every valuable new information represents a new conditional probability, thereby allowing the company to produce a new posterior probability.

3. Once the distinguishability of new information reaches a certain level, when the posterior probability exceeds a certain operating value (such as 70%), you can consider buying. The posterior probability can be divided into several levels in actual combat, which is directly related to "buy, add, reduce, and clear" actions. In this way, you can turn all the information and beliefs you master into probability events that guide operations.

Returning to the example at the beginning of the article, from these three steps, the supply and demand changes of that company have already produced new bullish information, indicating that the posterior probability has increased. At this time, if you do not take action, there are only two possibilities:

Possibility one: The posterior probability has not reached your operating line;

Possibility two: The posterior probability has actually reached the point where you take action, but you have not mastered this method.

It is obvious that a sharp rise in stock prices indicates that many people have determined that the posterior probability has reached the operating line. However, the problem is that you do not have an effective method for dealing with the relationship between new information and old news, and can only use the method of "delayed decision-making" to avoid this problem.

The reason you summarized as "weak execution" may cause you to step into the same trap again. When you encounter new information again, the posterior probability may not actually reach the operating line, but you think of the experience of missing opportunities due to weak execution last time and make the opposite mistake.

Investment is just buying and selling, only need judgement, no need for execution, unable to change your beliefs and operations with your new ideas.

Of course, accurate probability calculation is not necessary. As can be seen from the above example, the Bayesian algorithm is first and foremost a way of thinking.

Analyzing the past makes you confident, but predicting the future can be confusing.

In macro analysis, we often encounter situations where we analyze a certain phenomenon in the past market and obtain a clear rule, but when we use this rule to analyze the current market, either it is not very effective or we don't know how to use it.

So many people refer to it as a "rearview mirror" and think that this type of article is just attracting traffic.

However, historical rules do not give you direct conclusions and guide your operations, but they form a conditional probability - commonly known as "investment experience". If you have not mastered the correct method of probability judgement, it certainly cannot help you make decisions.

Feeling the past thread clearly is because when we observe past events, we only extract the data that has an impact and form a causal relationship in our brain.

And causality is actually a philosophical concept. In real life, what we can use is actually correlation. For example, whether smoking and lung cancer form a causal relationship is a philosophical judgement. Science can only tell you that they have a high correlation.

There are a large number of conditional probabilities in real investment, which affect each other, and it is necessary to find the factors with greater influence weight.

For example, the relationship between exchange rates and the stock market. Many people analyze based on recent intuition that when the RMB exchange rate falls, the stock market will also fall. However, there is no causal relationship between these two factors. The correlation between them is very complex, and it requires three steps:

Step 1: Determine the most important influencing factor of the current exchange rate on the stock market.

[Possibility 1] Positively correlated with the same factor: both are affected by economic recession.

[Possibility 2] Positive and negative correlation with the same factor: proactive devaluation policy may cause exchange rate to fall and the stock market to rise.

[Possibility 3] Produce positive correlation through funds: exchange rate falls, foreign capital outflow, stock market falls.

[Possibility 4] Produce negative correlation relationship through policy: new policies to stop the yuan devaluation have a negative impact on the stock market.

[Possibility 5] Not correlated: investors and policies are indifferent to exchange rate changes, and exchange rates and stock markets are insensitive.

In macro analysis, the first thing to do is to determine which of the above factors is the current explicit factor.

Step 2: Analyze the continuity of the impact conditions.

More importantly, macro analysis should not stop at explanation, but focus on prediction, which determines whether the follow-up of correlation is continuation, cessation or reversal.

The analysis is as follows:

In the above possibilities 1 and 2, exchange rates are only outcomes, and the actual judgment is about economic trends and exchange rate policies, without generating conditional probability.

In possibilities 3 and 4, exchange rates are a positive feedback factor, with continuity.

In possibility 5, exchange rates also do not generate conditional probability that changes the stock market.

Thirdly, predicting the impact of exchange rates on the market.

The use of macro factors to analyze the index requires acknowledging the rationality of the current trend, believing that the current trend has reflected the impact of all known information, and forming the 'prior probability' of the future rise and fall after forecasting.

Then, taking the exchange rate change as a new influencing factor, using conditional probability, we generate the 'posterior probability' of our predicted future rise and fall.

In this way, even if the actual trend is opposite to your prediction, you will know where you went wrong.

I believe many investors still have doubts when they see this: Can the stock market be predicted in this way?

That depends on how you define the concept of 'prediction'. Bayesian statistics is actually a 'thought' that attempts to combine past objective statistical results with the relevant beliefs in your mind to approach the truth that you have in your heart step by step.

V. Worldview in the eyes of Bayesianists.

The 'Monty Hall Problem' in the previous article is a derivative of Bayesian probability. I have written it many times, and every time there are a bunch of people questioning the result. I once thought it was an 'IQ test', but in fact, many highly intelligent people can understand the explanation but cannot fully accept it from the heart.

I later realized that it is more like a 'thinking style tester'. The more rigorous and rational your thinking is, the more difficult it is to accept.

When you finally evolve into a Bayesianist, you will find a completely different worldview unfolding in front of you:

1. The world is subjective, usefulness is more important than correctness, and belief is more important than understanding;

2. Even theories in textbooks are to be falsifiable, and before being falsified, we assume that they are correct;

3. People who hold different opinions from you are just subjectively different, and it is not necessary to try to convince them. It is more important to analyze the conditional probabilities that make up their beliefs and see if there is any information that you don't know;

4. When encountering an issue that has never been encountered before, boldly assume a prior probability, make a small bet, and slowly adjust through some information and feedback. When you have confidence in this probability, then formally bet on it;

5. Future events are a subjective probability that is constantly moving towards 0% or 100%, but can never reach 0% or 100%, and the more information disclosed, the more stable the probability becomes.

6. The change in probability is the distinctive conditional probability brought by new information, and the higher the distinction, the more dramatic the change in stock price. 7. New information itself cannot promote stock price, the real driving force is people's subjective perception of information.

Despite analyzing so much, I think many people still feel uneasy and are unwilling to estimate a data casually. Actually, in practical investment, truly effective information doesn't need specific calculation.

If you are using a Futubull account, please log in to access this feature.

In investment, real opportunities lie in conditional probabilities with strong distinction in the information you have found. Like the new information of customers 'talking about price' mentioned in the previous article, which can dramatically increase the probability of closing the deal from 23% to 70%. Experienced salespeople don't need to calculate but can realize the great increase in probability, thus using the ultimate weapon of offering discounts and win the customer.

If you are interested in this topic, I will launch the third article of this series: how to find information with distinction.

Editor / jayden