该模型在大多数时间里准确预测了以太坊的走势

现在我们进入最后几个话题，是关于阅读的。

Now we are getting on to our last few bits, which is about reading.

你实际上是怎么解读网络需求的？

How do you actually read the demand for the network?

如果不是公司，那我们怎么给它估值？

If it's not the company, then how can we value it?

有两件事我们和很多研究机构和分析师做得不一样，我挺自豪的，就是这些。

And two things that we will do quite different to a lot of research firms and analysts, I'm quite proud of, is those things.

第一，我们不关心费用有多高，因为开发者们故意让以太坊变得更便宜，就像我们看到的那样。

One, we don't care how high the fees are because the developers are making Ethereum cheaper on purpose as we have seen.

我们关注的是变化的速度。

And what we read is the rate of change.

也就是说，费用或者我所说的费用流，相对于以太坊自身常态的涨跌速度。

So how fast the fees or fee flows, as I call it, rise and fall against Ethereum's own normal.

而且这种变化实际上是分阶段的。

And that moves actually in phases.

就像经济经历商业周期的扩张、放缓、收缩和复苏一样，几分钟后你们会再次看到这四个阶段。

So much like an economy moves through its business cycle with expansions, slowdowns, contractions and recoveries, you will see those four names again in a few minutes.

第二重要的是，我们不会把手续费换算成美元。

And the second most important thing is we don't turn fees into dollars.

通常的做法是拿手续费，按今天的价格换算，比如称之为现金流之类的，然后用它来给以太坊估值。

So the usual approach is take the fees, convert them at today's price, for example, call it cash flow or whatever, and use it to value Ethereum.

但问题是，美元手续费等于以太坊手续费乘以以太坊价格。

But the problem is dollar fees are ETH fees times the ETH price.

所以价格进去，价格又出来。

So the price goes in and the price comes out.

这样你就把法币计价的循环引入了你的模型，这是我们不想要的。

What you have is you introduce fiat denominated circularity into your model, which we don't want.

我们想把手续费保留在以以太坊、以太币计价。

We want to keep the fees in Ethereum, in Ether.

也就是它们实际支付的单位。

The units they are actually paid in.

所以价格就不会渗入到任何预测里。

So the price can't leak back into any type of projections.

所以我们做的其实很简单。

So what we do is actually quite simple.

我们把人们支付的所有费用都拿过来。

We take all the fees people pay.

然后除以所有存在的代币数量。

We divide it by all the theorem that exists.

这样我们就得到一个以每枚代币费用为单位的指标。

So we get a metric that is denominated in fees per coin.

你很快就会明白为什么。

You will soon find out why.

然后我们找出正常水平，比如过去九十天的中间值，以太坊自身的正常活动水平，注意是活动，不是任何人的目标。

And then we find the normal, for example, the middle day of the last ninety days, Ethereum's own normal activity, to say, activity, not anyone's target.

然后我们衡量两者之间的差距。

And then we measure the gap between it.

比如说，第一步，回想一下我们刚才聊的所有内容。

So for example, step one, think back to everything we've talked just about.

美元在流动，稳定币、代币化基金、DeFi，等等。

Dollars are moving and stablecoins, tokenized funds, DeFi, whatever.

它们都支付了应得的费用，每一笔费用我们都放进了一个区块里。

They all pay their fair share of fees, and every one of those fees we put into a block.

当然，所有费用都是用ETH支付的，而不是用美元。

And obviously, all fees are paid in ETH and not in West dollar.

所以所有这些活动都汇入我们的池子，我们称之为费用，然后我们用这个池子除以所有以太坊的数量，因为这是以太坊的整个货币基础。

So all that activity collapses into our pot, which we call fees, and then we divide that pot by all the Ethereum there is because this is Ethereum's whole monetary base.

得出的结果是一个全新的指标，我在我的研究论文中首创了这个概念，这篇论文也是开源的，每个人都可以查看，我们称之为“流量强度”，也就是每枚币的费用。

And what comes out of is an entirely new metric that I've coined in my research paper, which also is open source and everyone can check on, which we've named flow intensity, which are fees per coin.

这里用实际数字举个例子。

So here's it with real numbers, for example.

我们以9月26日，也就是上周五为例，过去三十天里，平均每天的费用是158.7 ETH，也就是每天6.68 ETH。

We take the September 26, last Friday, and over the last thirty days, the average fees per day have been 158.7 ETH or 6.68 ETH per day.

当时大约有1.22亿。

And there were about 122,000,000.

所以，我们取这些数字，具体计算其实只是分类一下。

So we take the obviously the exact calculations is just a bit of a classification.

这两个数字都是公开的，而且是以ETH为单位，不是美元。

So both numbers are public and then ETH, not in dollars.

现在我们来做除法。

So we divide it now.

比如，我们现在得到的这个流量强度数值，我为什么称它为区块链的人均GDP呢？

And, for example, what we actually have now, the flow intensity number, why do I call it GDP per capita for a blockchain?

你想，一个国家的GDP在增长，但你可以除以人口，看看每个人大概生产了多少。

If you think about it, a country's GDP grows, for example, but you can also divide it by population in order to get how much each person is producing more or less.

这样你就可以随时间比较，不仅仅是通过通货膨胀。

So that's what you can compare over time, not only via inflation.

而且，以太坊的供应量也会变化，费用也一样，因为新ETH在发行，有些被销毁，有时供应甚至会变成通胀性的。

And for example, Ethereum supply changes too as well as it does with its fees because new ETH are being issued, some are being burned, sometimes the supply even turns inflationary.

如果你把手续费除以所有的以太坊，你看到的就是每个币是否在做更多或更少的工作。

And if you divide the fees by all the Ethereum, what you see is whether each coin is doing more or less work.

这样你得到的就是一个不受时间影响的变量。

So what you get is a timeless variable.

那我们拿它跟什么比较呢？

Well, what do we compare it now against?

我们不拿它跟股票比。

So we don't compare it to a stock.

也不跟比特币或其他人的目标比。

We don't compare it to Bitcoin or to someone else's target.

我们只跟以太坊自身的基本面比。

We compare it solely to Ethereum's fundamentals itself.

这里你看到的是过去九十天的手续费情况。

Here you see the last ninety days as they happened in terms of fees.

你看它们波动很大，很剧烈。

You see they are very spiky and volatile.

现在我们可以把它们从最低的一天排到最高的一天。

And we can now line them up from the lowest day to the highest.

然后我们取中间的那个。

And now we take the one in the middle.

所以这是我们的中位数，不是平均值。

So this is our median, not the mean.

我们取中位数，是因为少数几天显然可以把平均值拉高一到两个百分点。

We take the median and we take it because a few days obviously can drag up the average up to one or two percentage points.

但中间值几乎不受影响，因为我们想要一个稳健的以太坊基础经济水平。

But the middle barely notices because we want to have a sound fundamental economic level for Ethereum.

然后我们取过去三十天手续费的平均值。

And then we take the average of the last thirty days in terms of fees.

现在你看到的是我们的常态值和近期值。

And what you see now here is our normal and our recent.

另外，在幻灯片的右中侧，你可以看到过去三百六十五天的完整手续费图表。

And also you see on the middle right side of the slide the entire fee chart for the last three hundred sixty five days.

现在我一按按钮，你就会看到一个动画，我们会看到这个基础值呈灰绿色，而最近的值则显示为蓝色。

And when I now press my button, you will see an animation going where we see this fundamental base in gray green, so to say, and we see the most recent values in blue.

这样做的效果就是。

So this does one thing.

你现在可以看到蓝线在橄榄线上方的情况。

You can now see where the blue line sits above the olive one.

差距是绿色的，说明以太坊的经济比正常情况更热。

The gap is green, so Ethereum's economy is running hotter than normal.

当它跌到下方时，差距变成红色，表示比正常情况更冷。

And where it drops below, the gap turns red and it's colder than normal.

所以我们不需要做任何假设。

So we don't have to assume anything.

我们只是测量以太坊区块空间上实际产生的基本需求。

We just measure the actual fundamental demand which is being created on Ethereum's block space.

说实话，你可以...

And you could honestly...

我们在研究产品里把它作为服务提供，但你完全可以用几个订阅自己在今晚搭建出来。

We we offer it as a service on our research products, but you could obviously entirely rebuild it with a few subscriptions yourself tonight.

这些都是公开数据。

It's entirely public data.

我们不需要做任何假设。

We don't have to assume anything.

而且……

And...

好，这是它一年的走势。

Well, this is what it looks like over a year.

你可以做一个基础分析，因为你不仅能看到相对于基准线的上下波动。

There's a fundamental exercise that you can do because you don't only have the over and under from the baseline.

你当然也能衡量它的动能，比如扩张、放缓、收缩和复苏，就像我们现实中的经济周期一样。

You can also obviously measure the momentum, so you get expansion, slowdown, contraction, recovery, just like a normal actual economy that we have.

我们还可以更进一步，因为我们构建了一个所谓的流量状态矩阵，用来衡量这种偏离流量强度的偏差，也就是流量偏离度。

And you can even take this one step further because we've built a flow regime matrix, so to say, which measures this deviation from the flow intensity, which is called flow deviation.

现在你可以看到过去一整年以太坊经济的表现如何。

And you see now the entire past year in terms of how Ethereum's economy has fared.

所以你会看到它从衰退走向扩张，然后又回到衰退，这真是一个很好的工具和图表，能告诉你以太坊经济最近的表现。

So you see it move out of recession into expansion and back into it again, which is quite a good tool and charts to tell how Ethereum's economy is performing most recently.

你甚至可以更进一步，把我们积累的这个流量偏差数字最终转换成美元价格，因为这是目前还缺少的一环。

And you can even take one step further and put this flow deviation number that we have now accrued and ultimately translated into a US dollar price because this is the one thing that's still lacking.

显然，一切都以以太坊计价。

Have obviously everything denominated in Ethereum.

一个简单的步骤就是乘以以太坊兑美元的价格，把它转换成美元，这看起来可能有点简单，但我们很快就会解释为什么这可能是一个不错的预测方法。

And one simple step is just translating it into US dollar with multiplying with the Ethereum US dollar price, which might seem a bit simplistic, but we will soon cover why it's probably a good projection method.

所以这基本上就是一个温度检查，对吧？

So it's basically just a temperature check, right?

这个模型只是根据网络相对于自身常态的运行热度来判断，它对于以太坊在绝对水平上应该值多少美元并没有真正的看法。

So the model reads one thing if the network is running hot or cold against its own normal, and it has no real opinion on what Ethereum should cost in dollars on an absolute level.

它只关心相对变化。

It's only about the relative change.

但有趣的是，在金融市场里，我发现关键往往在于变化的速度和相对的水平。

But funnily enough, in financial markets, I find that it's often about the rate of change and relative levels.

所以现在最重要的问题是，我们创建了一个全新的模型，用流量偏差来预测，但它真的管用吗？

So the most important question now, we've created an entirely new model with the flow deviation, but does it actually work?

这就是统计学发挥作用的地方。

This is where the statistics part comes in.

整篇论文当然也已经开源了。

And the entire paper has obviously been open sourced.

它可以在好几个研究网站上找到，当然也包括我们的网站，而且已经提交给《管理科学》期刊进行评审。

It's available on several research sites and obviously also our website, and it has been submitted to management science for a review.

嗯，它……

Well, it come...

结果出来了，它预测以太坊市场价格未来六周或四十五天的走势方向，准确率达到了76.4%。说实话，这已经相当不错了，因为如果你一直押注以太坊会跌或者会涨，效果还不如我们的模型。

It came back with the data, and it called the direction of Ethereum's market price over the next six weeks or forty five days, 76.4% of the time, which is quite good, to be honest, because if you would have bet it's the entire time on Ethereum falling or going up, it would still have not been as good as our model.

这里你可以看到整个时间段的情况。

And here you see the entire time frame.

嗯，论文自己的窗口看起来就是那样，一天一天地。

Well, the paper's own window looks like that day by day.

最上面一行，你每天都能看到读数偏热还是偏冷，资金流是高于还是低于正常水平，下面则是以太坊接下来四十五天的表现。

And on the top row you see every day if the reading was hot or cold and the fee flows above or below the normal, and below you see what Ethereum did over the next forty five days.

绿色表示走势符合读数的倾向，红色则表示不符合。

So green is where it went the way the reading leans and red where it isn't.

基本上十五个时间窗口里，它预测对了十二个。

And it called basically 12 out of 15 time windows.