Because the stock exchange does not provide investors' account information to the public, the chip distribution in all kinds of software is an approximate value calculated through historical transactions. Assuming that the throwing probability of chips is related to the floating profit and holding time, we can sample a certain number of investment groups to get the function of this throwing probability, and then according to this throwing probability, we can determine which old chips are cancelled in daily transactions and are now replaced by new chips. Let's put the question simply: according to the profit-taking habits of a considerable number of investors, especially retail investors, it is easiest to sell stocks between 10%-20%; For the main force, it is difficult to sell most of his positions when the profit is less than 30%. Then, the profit-making disk with a profit of 15% contributes more to the day's trading than the profit of 25%. This is a more accurate method to calculate the chip distribution. Sometimes, for the sake of calculation, equal throwing probability can be used instead of real throwing probability statistics, which will lead to certain errors, but this error is tolerable. Because in the actual investment analysis, the chips at a certain price will not affect the final conclusion more or less. After the introduction of CYQ of compass, almost all domestic software manufacturers imitated a chip distribution map. Some of them are well done, but some algorithms have too big errors. Suggest that these software vendors modify it. This inaccurate algorithm is to weight historical transactions by time, and the longer the time, the lower the chip distribution ratio. On the surface, it seems that you can get something very similar to the distribution of chips, but it is actually useless. Because the activity of each stock is very different, it is difficult to reflect this difference by artificially determining the volatilization speed of historical chips. Because it involves trade secrets, the algorithm of compass chip distribution cannot be made public for the time being. I can only say some principled things here. I hope readers will forgive me. Comments: Indeed, I have seen a lot of software, and its chip allocation is a simple average transfer. The calculation principle is roughly as follows: first, the sum of all chips should be 100% or circulation disk, and the chip distribution refers to how many chips are distributed at each price; Then, the chips of each price are calculated from a specific date, and they are added together to get the total number of chips of that price until the chips reach 100% or the circulation disk stops counting. The formula is roughly as follows: cost distribution = cost distribution of the previous day x( 1- turnover rate)+cost distribution of the current day. The biggest problem is that human factors and time factors are not considered. For the same trading record of 100 lot, the investors who make profits in the trading group are definitely more than those who make less profits, and the investors who hold shares for a long time are easier to throw out than those who hold shares for a short time, so the chip transfer is absolutely possible. Here, Professor Chen Hao brushed it off with a throwing probability, but did not elaborate on it, which may be regarded as involving trade secrets. I guess this throwing probability should exist. How to distinguish between banker chips and retail chips? In my opinion, there are two factors that must be considered, one is the increase, and the other is the turnover rate or volume. If it is a long position, the chips at the bottom are generally cancelled (this part is the dealer's chips). If the volume shrinks, the most traded chips should be retail chips near the price. If the trading volume is stagnant, it means that the institutional chips near the price are in opposition ... Generally speaking, the transfer of chips should be related to the increase or decrease and the trading volume. If you can sit in front of the mainframe of the exchange, make up a small program to convert the data, arrange the order and draw a picture, you can figure out the real chip distribution of a stock. But there is another problem, that is, there will be multiple accounts belonging to the same person, and if he uses his own multiple accounts to toss, it will also worry analysts. The above content is taken from the famous Arabian Ancient Literature "Arabian Nights". The real chip distribution should be like this (very simple, but it is estimated that not many people think about it): for a stock on a certain trading day, all the chips actually held in the account of that trading day are accumulated and drawn on a graph with the price as the vertical axis according to the original buying price. When we can't sit in front of the main computer of the exchange (don't laugh, someone can sit, as far as I know, several people have the right), we can only simulate (or guess) the chip distribution according to the open market data (that is, the daily opening, closing, high and low, trading volume and amount). The starting point of simulation, I think, will generally start with the perfect turnover cycle (vol, capital) (please note that this perfect term was put forward by me here today, and the quotation must be noted). There are several differences (also some problems to be solved) between the simple chip distribution in a perfect hand-changing cycle and the real chip distribution: 1. Some investors bought stocks and held them for several years, some made miniskirts, and even T+0 came out the day before yesterday. As long as the stock price rises, retail investors and institutions are selling. The key is that the ratio is different. If it is a heavy volume, it should generally be the fastest chip write-off at the bottom, because this part of the chip is the most profitable. Generally, people who buy at the bottom will choose to take profits when the volume is high. This is a prediction of trading mentality. We don't have to pay attention to whether it is an institutional bargaining chip. As long as we know, when the stock price rises, the chips at the bottom generally transfer the most, and the greater the volume, the faster the transfer. This is also the reason why the turnover will not rise. Whether we can study a price-price model. The current chip distribution can be said to be a price-time model, that is, the distribution of trading volume at different price points is calculated by the time factor of perfect turnover period. When I say price-price model, I mean to consider the distribution of trading volume from the historical trading price.