Before you go, check out these stories!

Hackernoon logoUnderstanding Signal Versus Noise in Tech Investments and How to Leverage it [An Analysis] by@charliecliu

Understanding Signal Versus Noise in Tech Investments and How to Leverage it [An Analysis]

Author profile picture

@charliecliuCharlie Liu

Global Macro | Fintech | Sommelier

I have previously written about the investment philosophy of my friend Jinjian Zhang of Trustbridge Partners. A few months ago, Jinjian has founded his own venture capital fund, Vitalbridge Partners based in Shanghai, China. In a recent talk, Jinjian dived deeper into his “Noise vs Signal” investment philosophy. 

 — -

本文根据 42章经 线下活动「决策中的信噪比」整理而来。

This article is collated from the MyFortyTwo offline activity of “Signal-to-Noise Ratio in Decision Making”.

2019–2020 年市场很冷,有人选择离开、有人选择蛰伏、有人选择变化。从《投资中的信号与噪声》时的挚信资本,成长为文中的绿洲资本,并不是个容易的决策,张津剑在这篇文章中分享了他对创投市场核心决策的方法论。

The market is desolate from 2019 to 2020. Some people chose to leave, some were dormant, some chose to change. It is no easy to shift from Trustbridge in Signal and Noise in Investment to Vitalbridge in this article. Jinjian Zhang shares his methodology of the core decisions in the venture capital market in this article.


— — 量化经济学家 Fischer Black 
 Noise makes financial markets possible, but also makes them imperfect. By Econometrist Fischer Black.


The market is filled with signal and noise.

Each of us knows to pursue signals and abandon noise. However, without noise, we do not have transactions. Imagine that if everyone had the same signal, you would have no one sell what you want to buy, and the transaction would not be concluded.


Therefore, noise in fact drives the ecology of our entire venture capital industry. However, last year noise and noise traders slumped, thus weakening the liquidity of the market. In such a case, everyone was overwhelmed with anxiety and confusion.


When the overall market liquidity is reduced, “do something” always brings us a sense of security. As an inevitable result, noise traders will emerge because people struggle to find exclusive information and carry out transactions.


In fact, we must return to the essence and identify the signal when noise decreases.


Therefore, the best reaction at present is to stay peaceful. In a peaceful status, you improve the signal-to-noise ratio in your decisions.


What is signal-to-noise ratio? Let’s see the picture above.

图片左侧是我们理想中希望接受的信息,右侧则是实际接收的信息。左侧灰色面积 A 是理想消耗的功率,右侧是实际的。可以看到,噪声带来的“冗余”信息不单影响质量,而且消耗能量。

The picture on the left is the information we hope to receive, and the one on the right is what we actually receive. The gray area A on the left is the ideal power consumption and the right is the actual consumption. We can see that “excessive” information not only affects quality, but consumes energy.


Unfortunately, people’s way of thinking is closer to the analog circuit on the right rather than the digital circuit, thus inclining the expressions towards “redundant” instead of “simplified”.


What’s more, since it is an analog circuit, limitations lie in different circuit designs and component options. For us, the limitations are from our personal background, experiences and personality traits, which will increase the system noise in our communication.

就像上次文章《投资中的信号与噪声》中提到过的大学室友,他鼓起勇气给他暗恋的女孩(女孩也暗恋他)写了一封情书,洋洋洒洒 6 页纸,讲述了自己的成长经历以及不可放弃的报国志向,最后加了一句:“你也很重要”。寄出去,室友兴奋了一夜,女孩却以为自己被委婉拒绝而哭了一宿。

Like the college roommate I mentioned in the previous article Signal and Noise in Investment, he summoned great courage to write a love letter to the girl he loved secretly (and the girl also loved him secretly). In the six-page letter, he shared his life experiences and irresistible aspiration to serve the country. The last sentence he added was “you are also important”. After sending it out, the roommate was excited all night, but the girl mistook the letter as a euphemistic rejection and cried overnight.

再比如,当我们沟通一个项目的时候,你如果问他投不投?其实你需要的答案就是 0 或 1,但大多时候这个人会告诉你说“这里有很大的机会,分别是 1、2、3,但是这里同时又有风险,分别是 1、2、3、4、5。”

For another example, when we talk about a project, if you ask him to invest or not, you need the answer of either 0 or 1. However, mostly, the person will respond “I can see enormous opportunities here, which are 1, 2, 3, but risks as well, which are 1, 2, 3, 4, 5.”


You ask again “invest or not?” In the end, his response can be “shall we bargain over the price?”


In this example, a lot of information is in fact noise, while the signal is uncovered. Therefore, the final conclusion is ambiguous.


Hence the signal-to-noise ratio is the ratio of signal to noise in the information. Mathematically, it is the signal power divided by the noise power. The larger the number, the higher the signal proportion, and the better the information quality.


However, all the components cannot avoid producing more or less noise as long as a signal is sent out. In investment, we make a lot of decisions every day. How can we effectively screen out signals and improve the signal-to-noise ratio?


In fact, our brains encounter similar problems every day. Limited by the processing capacity of the human brain, we cannot process all the information collected by the retina. Humans solve this problem with two methods: (1) ignoring the information at the edge of the retina (2) “focusing” on processing the information in a specific area.


Therefore, even if we can see an entire book, we can only “read” a line of text. The latter is the “attention” mechanism that evolved in humans. The mechanism has greatly improved the “signal-to-noise ratio” in visual information processing.


For example, in the picture above, the driver will immediately “focus” on the red signal of “STOP” and ignore other information. In contrast, computers, with unlimited computing power, will perform all-inclusive analysis instead of instant focus.


To be brief, the attention mechanism is a “top-down”-driving-“bottom-up” strategy.


In investment, “top-down” is induction: finding the internal connection and shared characteristics between matters, and generalizing the common features. For example, what are the shared features of excellent entrepreneurs? Or, what business models do high-quality projects commonly comply with?


“Bottom-up” is deduction: finding the external variations and different characteristics between matters, and demonstrating individual peculiarities. For example, what resources will the entrepreneur search for in this environment? What transformation will be made?

前者需要大样本量的抽象总结能力;后者需要具体事件中的洞察想象能力。各有利弊, 我们展开来说。

The former requires the capability of abstracting and summarizing large samples; the latter requires the capability of penetrating and imagining in specific events. Both have pros and cons. Let’s discuss.


Facing small samples in life, “bottom-up” is more flexible, so the deduction is the main strategy in our daily life:


For example, in ordinary life, you now need to choose one out of the three furnishing suppliers. What is the best strategy? Meet all the three suppliers and you will draw the conclusion. The core of the strategy is to compare, that is to identify the “optimal” solution by comparing all options.


Surely some businesses take advantage of this. For example, The Economist offers three classic subscription options:

1、电子版,59 美金 / 年

2、纸质版,125 美金 / 年

3、电子版+纸质版,125 美金 / 年

1. Digital, US$59 per year

2. Print, US$125 per year

3. Digital + Print, US$125 per year

你会选择谁?如果没有选项 2,你会选择谁?选项 2 的出现,让 3 成为了全局最优选择。

Which one will you choose? If without Option 2, which one will you choose? The existence of Option 2 makes Option 3 the optimal choice.


However, everyone must know that in fact, the comparison itself is exactly the source of most noise in life and decisions, because, in the process, we believe our choice to be the best one by constantly comparing. With the increasing options under comparison and the aggravating complexity of the system, we drive ourselves into an overloaded system.

就像人和人之间一样,毕业之前可能大家只是比比成绩,但毕业之后发现成绩已经不能够支撑我们之间的比较了,就变成了薪酬待遇、社会地位、名望、你的妻子、儿子,从一个点变成一百个点,这也就是日益剧增的 Peer Pressure 的来源。

Similarly, people have the issue in the everyday relationship. We compare scores at school. However, after graduation, you find that scores can no longer establish a comparison. It shifts to salary, social status, reputation, your wife, son… The comparison extends from one aspect to a hundred, which becomes the source of increasing peer pressure.


Therefore, the bottom-up comparison method fails when people face an environment of large samples.


For example, today I tell you that you only have one single opportunity to choose from one hundred top doctors in China. The choice may determine your life span. How will you choose? Will you spend two years meeting all of the one hundred doctors? Even you will, you could hardly make a decision after the meetings.

2012 年的创投市场,正是一个小样本量的环境。简化来说,就像上面这幅图,中国 2012 年整个市场可能就是这五家公司,竖着的表示有价值的公司,横条就是没有价值的。

The venture capital market in 2012 was exactly an environment of small samples. To be brief, it was like the picture above: China’s entire market in 2012 was composed of the five companies. The vertical bar represents the valuable company and the horizontal bars represent the worthless ones.

那个时候生态也不成熟,比如 FA 机构,整个中国就那么几家,你只要认识一家,就能知道市场上发生的所有变化,每年年末的时候所有投资人坐在一起,大家都知道说今年哪些项目融到钱、这个项目怎么样, 所有人都见过所有的项目。

The ecology at that time was not mature as well. For example, China had only a few Finance Advisor institutions. As long as you knew one, you knew all the changes in the market. At the end of each year, all investors gathered to share the financing and development of specific projects. Everyone viewed every project.


At that time, the “bottom-up” comparison strategy worked best. The volume of the entire market was small. You could view all the projects in a short time. As long as you view all five projects, you would be capable to identify the best one. You could make decisions by comparing. Therefore, “sourcing” was the core of the stage.

但今天市场变了,今天市场变成上图这样, 变成了大样本量的环境。

However, today the market has changed. Today’s market is like the picture above, which is an environment of large samples.


Some investors happen to meet a column of companies, which are all bad targets and say negatively that the market no longer has innovation because no company is worth investment. On the contrary, if some investors coincidentally viewed three good target companies, they will say that the market is as blooming as spring.


If we still want to find the two most valuable companies in the market, what can we do with our past bottom-up comparison strategy?

由于策略是比较,所以不能“遗漏”,进而投资机构开始围绕“Soucing”来扩大,有的组织以一横为单位分赛道,有的组织以一纵为单位分产业。试图通过分组的方式,把一副图形分成一、二、…七、八组,然后每一组有一个 leader,leader 之间再相互讨论,得出今年项目要投什么的结论。

Because the strategy is to compare, nothing can be “missed”. Further, investment institutions began to expand around “sourcing”. Some organizations horizontally divide the whole into tracks and others vertically divide the whole into industries. By dividing and grouping, institutions try to divide the graphics into one, two, …, seven, eight groups. Each group will have a leader, and the leaders discuss with each other to draw a conclusion of what to invest this year.


But what’s wrong at the moment?


First, no one can see the entire picture of information. That is to say, no one can make a comparison based on complete information. If what you compare is different from mine, how do we conduct cross-validation?


Secondly, a single comparison method fails when the information dimension multiples. We must continue to add “comparison points”. With more “comparison points”, noise increases and the market begins to diverge.

这里还有两点。 第一点是之前提到的单位元器件的结构性差异。 就是当一个人已经没有办法去覆盖整个市场的时候,他只能通过雇佣更多的人去做这件事。所以你会发现国内 VC 这个行业的层级变的越来越多,不像美国的 Benchmark 到现在只有合伙人和分析师。 但很多人忘记了一个东西,就是任何一个原器件在发出信号的时候一定是有噪声的。

There are two more points. The first is about the structural difference of the unit components mentioned earlier. When one is not able to cover the entire market, only by hiring more people can he accomplish the research. Therefore, you will find the domestic VC industry has increasing layers of hierarchy. Unlike Benchmark in the United States, the company only has partners and analysts. Many people forget a critical point that any component makes noise when it sends out a signal.

第二点是比较对象过多会造成整个系统的过载。 就像去年一样,突然有一天,你会觉得这个市场怎么这么冷,你会发现周围突然很多人都疲惫、迷失了,是因为市场在高速流动性的刺激下,这个系统已经没有办法支撑这么庞大信息的搜集了。

Second, excessive comparison objects will overload the entire system. Just like last year, one day you unexpectedly encountered the desolateness of the market and found people around exhausted and lost. It was because the market system, under the stimulation of high liquidity, failed to support the huge collection of information.


We have mentioned earlier that noises show differences but signals demonstrate similarities. When the sample becomes larger and comparison cannot help us choose, we need to identify the similarities to find the signal.

而“自上而下”的归纳法正是探索大样本事物之间的相同点,而非不同点。 我们将这种相同点叫做特征,而这些特征构成了事物的本质。我们经常提到复盘,复盘就是通过归纳法不断总结成败特征,进而了解本质的过程。

The “top-down” induction method is exactly exploring the similarities, instead of differences, among a large sample of things. We call the similarities peculiarities. We often refer to reviewing, which is exactly a process of understanding the essence by summarizing the peculiarities of success and failure through induction.


For individuals, it means thinking upon the life plan and the defining of certain peculiarities, which in return pilot the decision making. For the investment in organizations, it means the type of entrepreneurs and companies we look for, fast-growing or high-quality? The essential peculiarities finally become the consistent code of conduct for individuals and organizations, which is what we often describe as “vision, mission, value”.


When we understand and strengthen the essence, we will make a decision around the essence. When you decide to invest in a high-quality company while your opponent chooses to invest in a high-growth company, the organization will know that his financing news is full of noise. When you choose a simple and joyful life while the era chooses the hustling and highlighting, you will understand that his anxiety has nothing to do with you.

所以借助这样一个冬天和长假,我们真正应该关注的是作为个体和组织,我们的立身之本是什么,我们的特征是什么,我们的本质又是什么。 这种方法可以过滤掉大量噪声,但是不足之处在于这种一般性会缺乏在具体事件中的应用,因而变得空洞且无法落地。

Therefore, after such winter and long holiday, what really deserves the attention of an individual or organization is what our core of standing is, what our peculiarities are, and what our essence is. The method can filter out a lot of noise, but the disadvantage is that the generalization lacks the application in specific events and becomes hollow and impracticable.


The attention mechanism is the combination of these two strategies — optimizing the comparing process of specific cases by inducing the characteristics of things in advance. Induction drives deduction; essence drives comparison. Finally, the signal-to-noise ratio in the information will be improved.

例如开头驾驶的例子。驾校学习让我们了解到一些危险标示的特征,然后在实际场景中,当这些特征出现,大脑会优先处理这部分信息,忽略其他信息。 我们经常讲“不忘初心”,这份初心,就是个人生活一以贯之的特征,这个特征就会反向指导日常的具体决策,减少比较带来的噪声。

Take the example of driving at the beginning. The learning at driving schools allows us to understand some features of danger warning signs. In real-world scenes, when the features appear, our brains will prioritize this part of the information and ignore others. We often talk about “remaining true to our original aspiration”, and the original aspiration is one’s consistent features in life, which pilot the specific decisions in life and reduce the noise caused by comparison.

现在市场很冷,很多人觉得没有机会了,这是基于“比较”得出来的结论。然而我们回归到“本质”,科技驱动的中国消费市场,真的没有机会了吗? 这里面,我们回到两个特征,科技特征和人口特征,这两个特征没有了吗?

The market now is desolate and people mostly have a pessimistic attitude towards opportunities. It is the result of “comparison”. However, if we return to the “essence”, in the technology-driven Chinese consumer market, are we really short of opportunities? If we return to the two characteristics — technology and demography, are they gone?

我们先看科技。 这张图是过去 100 年里主要科技产品在美国家庭的渗透率。 “电力”大概用了 40 年完成普及,“手机”大概用了 20 年,“平板”还没有完成,目前来看会更短。我们在这幅图中可以看到,科技的渗透在变的越来越密,也越来越陡峭。

First, let’s take a look at technology. This chart shows the penetration rate of major technology products in American households in the past 100 years. It took about 40 years to complete the popularization of “electric power” and about 20 years for “smartphones”. “Tablets” have not finished the popularization yet, but the time is supposed to be shorter. In this chart, we can see that the penetration of technology is getting denser and steeper.

我们在科技上的探索和创新,只会更加频繁和快速,从 0 到 90% 只会更快,更让人意想不到。特别是接下来十年随着 5G、AI、区块链等很多科技爆发,我们可能会迎来完全崭新的纪元。 所以对于科技这个特征而言,一点没有减少。

Our exploration and innovation in science and technology will be nothing but more frequent and faster. It will be faster and unexpected to grow from 0 to 90%. Especially in the next decade, with the outbreak of 5G, AI, and blockchain, we may usher a brand-new era. Therefore, regarding the field of technology, the opportunities are not diminishing.

我们再看人口。 中国 1980 年到 1995 年这 15 年出生人口数量,基本等于美国的人口总数,而且这代特殊的独生子女还因为特殊的成长背景和社会经历正在迸发出独特的消费需求。

Second, let’s talk about demography. The number of Chinese people born during the 15 years from 1980 to 1995 is basically equal to the entire population of the United States. This unique generation under the only-child policy has evolved unique consumption needs because of the special upbringing background and social experience.

另一方面,中国的人口产业结构也在发生迁徙。 比如下面这张图,美国作为农业大国,第一产业的人口大约 3.2%。1970 年之后,第三产业人口大部分来自第二产业,大部分是从过去的工厂转型过去,所以今天的美国服务业以中高端服务业为主。

On the other hand, China’s population size in industrial structure is also migrating. For example, the following picture shows that the United States, as a major agricultural country, had about 3.2% of its population in the primary industry. After 1970, the most population in the tertiary industry were from the secondary industry and most of them were transformed from the factories in the past. Therefore, the US service industry today is dominated by high-end service industries.

但中国第三产业人口目前主要来自第一产业,大部分都是中低端服务业。中国农林畜牧还有多少人口呢?33.7%,而美国是 3.2%,日本则是 3.7%。 这里接近 10 倍的结构性差异将在中国未来 20 年里发生迁徙,进而形成巨大的变革。

However, China’s tertiary industry population currently mainly forms the primary industry and most of them are in the low-end service industry. How many people are there in agriculture, forestry and animal husbandry in China? It’s 33.7%. The percentage is 3.2% in the United States and 3.7% in Japan. The structural difference of nearly 10 times here will cause migration in China in the next 20 years, which will evolve into a drastic transformation.

在这个巨大的变革中,这些人口去哪里,他们怎么稳定适应新的产业,如何帮助他们,什么样的科技公司会受益?这里面有很多可以思考的问题。 所以无论是科技还是人口的特征,噪音是在变多,但信号并没有减少,希望我们提到的注意力机制,可以帮助大家不断提高投资和决策中的信噪比。

In this drastic transformation, where do the people go? How do they adapt to new industries? How can we help them? What kind of technology companies will benefit? Many issues deserve our thinking. Therefore, in the characteristics of both technology and demography, noise is increasing but the signal has not decreased. I hope that the attention mechanism can help everyone continuously improve the signal-to-noise ratio in investment and decision-making.

那么在日常的生活和投资决策中,我们如何不断提升注意力机制的应用呢?基于前面所述,有两点在这里总结: 第一、了解自己。 了解自己的成长背景,生活经历和家庭关系,更好的还原自己作为的原始参数和性格特点。这些特点决定了我们在乎什么,我们喜欢比较什么。这是自己的功课,也是组织领导者对于团队和组织建设发展的功课。 只有这样,我们才能更好的利用自己的特点,去提升决策质量。

How can we continuously improve the usage of attention mechanisms in daily life and investment? Based on the content above, I would like to conclude with two points. First, know yourself. Know your background, life experience, family relationship, and restore your original parameters and personality traits. The traits determine what we care about and what we like to compare. It is your own assignment, as well as the organization leaders’ assignment for team and organization development. Only in this way can we make better use of our own peculiarities to improve the quality of decisions.

第二、回归本质。 老子讲“为学日益,为道日损”,不断去除噪声,回归到事物最基础的特征上,就是本质,就是道。商业有商业的道,企业家有企业家的道,这个道有一个不断去繁就简和相信的过程。

Second, return to the essence. Lao Tzu said, “devotion to learning increases one’s knowledge; devotion to Tao diminishes one’s doings”. That diminishing noise and returning to the basic characteristics of matters is the essence, the Tao. Tao exists in businesses and entrepreneurs, witnessing the process of constantly simplifying matters and building faith.


For example, we emphasize the “power of life” inside entrepreneurs. Experiences can be accumulated and skills can be purchased. Only the power of life is inherent, distinguishing the living from the nonliving. However, power is not preserved by many people but consumed by the noise gradually.

英文里面有个单词,“Vitality”,说的就是这种“生命力”,韦伯词典进一步解释,说这是生命自我忍耐和进化的能力。说的真好,这也是我们认为企业家最本质的能力。一个企业家能走多远,一定不取决于他的背景、经验和资源,而是他是否不断自我认知、自我忍耐和进化。我们希望能发现这些有生命力的企业家,架起生命力之桥,这就是我们基金名字 Vitalbridge(绿洲资本)的来源。

The English word — vitality — exactly describes the power of life. In Webster’s Dictionary, the word means the power of enduring and the capacity to live and develop. The definition is marvelous. It is what we think of as the most essential ability of entrepreneurs. How far an entrepreneur can go must not depend on his background, experience, and resources, but depending on whether he has constant self-awareness, perseverance, and self-evolvement. We hope to find these vital entrepreneurs to build a bridge of vitality. This is the origin of our fund name — Vitalbridge.

愿每一个生命,能保护好这种勃勃生机的生命力,胜过于保护一切。 很高兴和大家分享过去一年的感受,也希望在这个噪声纷扰的时代里面,我们大家能够了解自己,回归本质,提高决策中的信噪比。 最后,就像42章经讲的,回到朴实的生活里,思考事物本质,谢谢大家!

May every life be better saving the vitality than saving anything else. I am very happy to share with you my reflections on the past year. I hope that in this age of noise and disturbance, we can get to know ourselves, return to the essence, and improve the signal-to-noise ratio in decision-making. Just like what MyFortyTwo asserts, “go back to the simple life and think about the essence of matters”. Thank you.


The Noonification banner

Subscribe to get your daily round-up of top tech stories!