Reduce complexity
Editor’s note: This article is from the WeChat official account “Seal of Thought” (ID: sxgy9999), The author works hard together.
“Sao operation” and “Mediocre operation”
When stockholders meet, it is inevitable to use their own cases to “learn from each other”. There are no more than several cases:
Rapid rise in a short period of time;
Accurately escape the top and bottom at the “point” level;
Successful at the “minute” level before the crash;
Successful continuous share swap……
On the contrary, few people take “holding Moutai this month, up 5%” as a case, the latter’s difficulty coefficient seems to be too low.
If the stock trading is a gymnastics competition scoring system, these “sao operations” are naturally 10 points, all of them are stock gods, but the real stock gods always have a lot of “banal operations”, which makes people unable to complain.
Once, I discussed this phenomenon with a friend who is good at Go. A friend recommended me an article that analyzed the “Go Man-machine War” in 2016, which gave me a deeper insight into the “banal operation” in investment. significance.
The “vulgar hand” of “Alpha Dog”
Returning to the topic of “Alpha Dog” vs. Li Shishi, in the plate, the commentators repeatedly used contemptuous comments that “Alpha Dog” was another “vulgar hand”. Of course, as a result, we also know that Alpha Dog 4 Battle 3 wins, and small wins every time.
Popular hand refers to the uninspired game of Go. The vulgar hand is different from the “wrong move”. The latter refers to a calculation error that leads to significant losses. The “vulgar hand” is a dumb game often played by low-level chess players. The most common one is to move a game that can be contested. dead”.
In master-level contests, there are very few vulgar players, which is why the commentator’s ridicule is aroused, but why do programs often vulgar players?
As a non-emotional program, “Alpha Dog” predicts the probability of the opponent’s next position for each step, and every possible change in the overall probability of winning or losing. In some cases, there are many possibilities for the opponent who plays a certain method to move next. When the calculation is too large, the vulgar hand of “Alpha Dog” is part of the strategy. Under the premise of satisfying the final winning rate, Sacrifice the quality of a move to reduce the computational complexity of the chess pattern.
If we divide the computational complexity of the chess pattern into 10 levels, 1 is the easiest and 10 is the most difficult, assuming thatThe highest level of complexity that can be calculated by human super masters like Li Shishi is level 5, while the “Alpha Dog” can reach level 7 through extensive machine learning.
If the computational complexity of the current chess type is level 9, because it greatly exceeds the human understanding of complexity, the human understanding is still level 5, and I still want to rack my brains to complete this “impossible task” , Will continue to make mistakes in vague estimates, and the competition of human masters in complex chess games depends on who makes the mistakes that are not so fatal.
But the computer has its own method. If the complexity exceeds 7 levels, the system will warn you: the amount of calculation is too large, it will time out, and the complexity must be reduced. The method is to use “vulgar hands” to fix the chess pattern of a certain piece of land (human understanding is to sacrifice the quality of the move), and forcibly reduce the complexity of the chess pattern to level 7, which is still higher than human understanding, but the computer has been able to calmly response.
The reason is a bit similar to that of the “Zhen Long Game” in “Dragon Ba Bu”. After filling in, “The world is wide. You don’t have to worry about the life and death of this big white chess, let alone have your own white chess.” It’s always constrained, but it can move freely. It’s not as difficult as before.”
The second game of the Alpha Dog self-play, which is talked about by chess friends, is a typical game of this strategy.
This is the “Fast rollout”, one of the four major strategies of “Alpha Dog”. On the premise of satisfying the winning rate, sacrificing the quality of the move and increasing the decision-making speed by 1,000 times.
The result of playing chess is three, win, tie or lose. Alpha Dog is a small win in almost every game. At first, everyone felt that this was a performance of the same chess strength of both sides, but after a long time, everyone figured it out. This is the dimensionality reduction blow of the disparity in strength-as long as the winning rate is satisfied, what All chess can be played.
But mankind has too much understanding of victory and defeat: wild victory, big victory, victory, small victory, tie, regrettable defeat, defeat, big defeat, miserable defeat…, there is also the turning of defeat into victory and negligent loss of Jingzhou, Hat trick…
Humans have made victory and defeat so complicated, precisely because they know that their calculations are not so accurate, and try to accumulate more advantages to secure the victory, and the result is quite uncertain. Uncertainty makes the game full of suspense, and the audience is satisfied, but it is not a good thing for the players.
Human chess strategy has its last resort, but to a certain extent, humans overestimate their own judgment. Therefore, in stock investment that also has a certain game component, when faced with too many uncertain factors , To lower certainUncertainty, in order to save resources and time to better grasp the opportunity, this is also a Fast rollout strategy.
Difficult opportunities
Compare the following two opportunities, which one is more worth fighting for:
A: Three consecutive years of 10% income, 80% chance of probability
B: 20% one-year income, 80% probability
The former opportunity is calculated by compound interest and probability weighting, and the three-year total internal rate of return is 26.5%.
The latter is transformed into three years for analysis, because every year we have to look for opportunities with 80% probability. In fact, the total internal rate of return for three years is 37.2%.
It seems that the latter is more profitable. But the problem is not that simple. The opportunities in the market are uneven. The 80% probability is based on calculations under current conditions, including market position, theme style, etc. When you have achieved the first 20% and find the next 20% chance, the probability may not be 80%. It is calculated based on future conditions, and the current conditions evolve into future conditions. For a fairly complex forecast, the calculation difficulty will increase exponentially.
Moreover, the hypothetical research of the former has ended, while the latter has two more researches, which is obviously more expensive.
With the computing power of the human brain, it is actually difficult to judge which of these two opportunities is more worth fighting for. Maybe the future quantitative investment can be calculated based on a complex strategy based on some combination of factors, but it is obviously not possible now.
Therefore, these two types of opportunities can be a combination of positions, a certain proportion of “three consecutive years of 10%” and “one year 20%” combination.
The proportion of the portfolio. In theory, the larger your stock pool is, the more likely it is to find stocks with a 20% increase in probability of 80% in one year. Therefore, retail investor positions should be more biased towards the former and have better research capabilities. Strong institutions should prefer the latter.
But in fact, retail investors are usually attracted by numbers and are used to overestimating their research capabilities. They like to allocate more “20% a year” varieties; many funds are too large to find enough The “20% a year” variety was forced to hold an excess of “10% for three consecutive years.”
In investment, as long as there are weaknesses, there will be opponents who use your weaknesses to make money. And the truly terrible opponent based on the insurmountable weakness of human beings is the kind of “Alpha Dog”-quantitative investment.
Whose money is made by quantitative funds?
Starting in 2015, domestic quantitative funds have begun to develop, especially in the bear market in 2018, and their returns have exceeded most active funds and index funds.
We usually say that value investing makes money for business growth, and technical analysis makes money in the pockets of opponents. So, whose money is quantified?
When I first entered the stock market in 1996, using simple technical indicators such as moving averages, KDJ, and MACD, I could get good returns and there were many opportunities.
In the future, a combination of multiple technical indicators must be used, and the frequency of opportunities is greatly reduced. Only by expanding the stock pool can profits be realized, and a method fails after a period of time, and a new method must be found.
Technical analysis is a behavior-based game. It is a probability-based analysis and judgment of current data. The congenital defect is that the strategy will fail if more people are used. The complexity needs to be deepened. The brain cannot handle it, and it will inevitably be defeated by quantitative investment.
Technical analysis is naturally suitable for quantitative investment, so most of the current domestic quantitative strategies are based on technical analysis, so they earn money that is also “technical”, so in the past few years, do you feel There are fewer and fewer technical masters around?
Why are there not many quantitative strategies based on fundamentals in China? It is precisely because value analysis has some natural “naïve” tendencies that make its analysis focus on the analysis of “business models”, “moats” and “management capabilities” that cannot be quantified at present. This is the area of human advantage.
In terms of trading, value investing also emphasizes more watching and less action. The calculation of a few fundamental indicators is not large, which leads to the fact that quantitative strategies can make more money for value investors than for technology.
So value analysts should abide by the “Less Is More” principle and strictly control the complexity of analysis within their own control, instead of easily running to the opponent’s home court.
The most typical crossover is the “timing strategy” for judging the rise and fall of the index.
In all technical analysis, the judgment of the index is the most complicated, because the factors that affect the index are exponential growth compared to the factors that affect individual stocks. People who frequently use it will find that it often conflicts with the “stock selection strategy”, which greatly increases the complexity of the system.
This is equivalent to giving up weapons and fighting with beasts.
Two different strategies
Finally, let’s summarize:
The strategy of general investors is: set up the risks they can accept, and on this premise, pursue the maximization of the risk-return ratio.
This strategy looks perfect, but once we encounter an opportunity that is so complex that the human brain cannot calculate it, we are susceptible to the temptation of high returns and forced judgment.
The strategy of “Alpha Dog” is to set a small win strategy of “at least one eye”. Under the premise that the current board can maintain a high winning rate, judge the probability of the opponent’s next move. If calculated The resources are still insufficient, so use “vulgar hands” to reduce the complexity of calculation.
The conversion of this strategy into an investment strategy is: first preset an income target, such as 15%, and then search for various asset combinations that meet this condition, and select the execution transaction with a higher probability. When the price changes, through the comprehensive calculation of the odds, constantly adjust the target varieties and proportions, so that the one-year rolling return rate is constantly at the level of 15%.
The biggest difference between these two strategies is that human’s best thinking is to constantly pursue a higher rate of return within a certain risk-return ratio; the optimal state of the program (just based on Alpha Dog’s assumptions, not the current Quantitative strategy) is to continuously pursue higher certainty under a certain rate of return.