Although both are data analysis, various types of data analysis have their own characteristics and flaws.
For example, in our financial forecast analysis, we will adopt water level management and control for the occurrence of risks. Different water levels correspond to different risks. In big data analysis, we will focus more on the linearity and interactivity of the data. The main purpose of the data on the chain is to infer various inferences. Due to price correlation, these data are divided into technical aspects of market transactions, news aspects of market value, etc. The meaning of the data composed of these aspects will naturally have corresponding flaws. We divide these shortcomings into the following points to explain: 1. Dependence on historical data: Price is a market phenomenon, so we will make predictions based on past market data. But the market is constantly changing, and past patterns may not accurately predict future conditions. Just like past weather conditions may not accurately predict tomorrow's weather. And history is a process of cumulative change. Just like the Bear Bull in 2009 is different from the Bear Bull now. We can say that some data models can still be used to speculate out of the Bear Valley, but they may not be applicable during the climbing power period. 2. Inability to consider all factors: Each model may only focus on a few specific indicators or data, while ignoring other factors that may affect the market. The reason is simple, because a model that takes everything into consideration is not a model, it is a model of God. (This is actually a programming joke, similar to a cold joke. If you don’t understand it, just skip it.) If you want the model to be established, you must delete the data that you think is disturbing. In addition to data interference, there is also news interference, including global economic conditions, political events, market sentiment, etc. Once the influence of the news occurs, hedging is a good choice, but you can also choose to go against the wind. After all, when blood flows into a river, opportunities emerge. This phenomenon is easy to understand. Just like from the data, it is clear that capital is about to withdraw, and suddenly a country says "adopt Bitcoin as a legal currency", which will trigger an emotional surge. The effect of this emotional surge will gradually decrease with the number of occurrences. Having said that, the more mature the market, the lower the frequency of this kind of phenomenon. 3. Data latency: Since the model uses past data to calculate indicators, there is a certain delay. In situations where markets are moving rapidly, instruments may not be able to respond to market changes in a timely manner. For example, suppose a tool calculates an indicator using data from the past 30 days, and the market experienced significant volatility during those 30 days. In this case, the instrument may not accurately reflect the latest market conditions, and investors may miss the best opportunities to buy or sell. But this is not bad. Usually, people will interact with four to six models for comparison. If the comparison is divided into four models: seven days, 30 days, 111 days, and 200 days, it will be less likely to fall into the blind spot of delay. 4. Risk management affects decision-making: The model does not provide "buy or sell" signals. Investment operations are actually a decision-making process for the use of funds and risks. What the model can do is provide reference, not decision-making. To put it simply, when you have 1,000 times the capital for leverage, you won’t care even if the sentiment of 10% surges. But if you only have one time the capital, even if the model shows a buy signal, you may not be able to withstand it once. accident. Many trading teachings will talk about buy and sell signals, but in the field of data analysis, decision-making and analysis are not the same thing. Decision-making includes betting on the possibility, that is to say, assuming that the signal does not appear, but the expectation value is high enough, or the resistance If the pressure ability is high, the operator may still make a buying or selling action.
No matter what kind of data analysis, the most important thing is hedging, not only the trading part needs hedging. For example, I am currently purchasing a batch of goods from abroad. The quantity of goods is within the safe range. I have a great chance that I can almost sell out this batch of goods within three months, and there will be no shortage of goods. But at the same time, I will consider what happens if there is any unsold situation, such as parallel imports entering the market, or the emergence of homogeneous competing products, so I will also formulate what data to observe. Once these additional reference data If it goes out of the estimated range, start the filing or even the second filing immediately. Regarding financial data analysis, I will write about it after I finish all the basic courses on on-chain data analysis. The fun of data analysis lies in quantifying reality and achieving verifiable recalculation and replicability through quantified deducibility. I will talk about this slowly when I have the opportunity.
That’s it for today.