Panic Journal (3) – Why Expert Virologists seem to be really bad (or careless) with statistics
Disclaimer: I am an anonymous investment blogger and not a Virologist, so please be aware of this. I also think that Covid-19 is a huge threat and that everyone should contribute to public safety by staying at home and avoiding contact for some time.
The Problem
As mentioned above, I am a “stock guy”, however mostly a fundamental/Behavioural type of investor. If I see a number, I want to know where it comes from and how it is derived.
The current discussion on Covid-19 is, at least in Germany, mostly about numbers. Two numbers stand out and get reported all the time and everywhere:
- Number of Covid-19 Infected persons (per day and daily change)
- Number of persons that died due to Covdid-19 (again per days and daily change)
The official source for this number is the Robert-Koch Institute (RKI) named after the German scientist who discovered the tubercolosis bacterium more than 100 years ago.
RKI publishes daily the number of infected persons that have been reported to them during the day. Most people look both, at the absolut change and the relative change from one day to the other and on every TV discussions, these numbers are the basis for whatever is discussed.
The numbers so far look dramatic, implying a huge increase in infections especially since last week.
For 18.03. at the time of writing, RKI reports 8.198 cases, a plus of 1.042 cases compared to the day before.
Eight days earlier, on 10.03. total cases were at 1.296, an increase of 157 against the 1.139 on 09.03. A significant part of the press and also politicians interprets this as proof for an explosion of infections in Germany.
The problem is that these number say actually very little. In order to be counted, a person needs to be tested first, and then test positive for the virus. There are stories everywhere that Germany doesn’t have enough test capacities. People want to get tested but get no response or are sent home. This again is seen as a further proof of a massive spread of the virus.
One thing is very clear: the number of actually infected persons is much much higher than the 8.198 reported ones, but no one knows how many.
The BIG issue: This is useless data
However the big issue in my opinion is that the RKI doesn’t know how many tests are actually performed each day. If you search for the number of test performed, there are no hard numbers available.
What is clear is that the capacity and actual test activity has increased a lot over the last few days.
One source, a public-scientific TV show published some estimates.
According to them, up until “Mid march”, Germany had a capacity to perform 12k of tests a week or ~1.7k per day. Now, the capacity seems to have increased by a factor of 12,7 to around 160K a week or 22.9k per day.
No comes something obvious but that I haven’t heard anywhere yet: The more you test, the more positive results you get, even if the infection rate stays constant or even goes down.
As there is no data on testing, everything is speculation. But if you just assume that the infection rates was constant, a 12,7x increase in tests should result in 2.124 confirmed cases, actually more than have been reported.
Conclusion:
My conclusion is simple: The current numbers for infected persons, especially the daily changes are simply useless in order to determine if and how the infection is spreading. Nevertheless everyone is focusing on them and using them in a very wrong way.
The big scandal in my opinion is that the “expert virologists” have not thought about this at all. The RKI now tries to get this numbers, but in the mean time, Germany is flying blind on the actual spread of the Virus.
However please be careful: I am not saying that the virus is NOT spreading like crazy in Germany and that the infection rate is not increasing. I simply don’t know. But neither the expert virologists nor our Government does. We are all flying blind at the moment. And don’t belief anyone who wants to tell you how things will develop over the next week, months or years with regard to the virus. No one knows, it is all speculation. The same goes for capital markets.
value and opportunity 
Italian official numbers include hospitalised and intensive care data, which is useful. Can’t find equivalent on RKI site. Is that available somewhere?
Thank you for your update. Here in belgium we are at about 2000 positives on 16000 tests. But the number of tests a day isnt released either.
In fact 50% of the people tested positive were asymptomatic (in the Princess ferry). So any figure should be multiplied by 2 (asymptomatic usually don’t test).
The only method is the corean: once someone is tested positive, test his contacts.
In Corea, a lot of 20-30 are tested positive afterwards but are asymptomatic.
Dr Raoult from Marseille in his last youtube video explains very well what should be the method (in english).
Spread the word, you are organized enough in Germany to copy Corea. Otherwise, you will have to make a confinment of the country like France.
My conclusion early on following this on the ground in Asia was that the fear factor is the largest part in this.Not the actual danger of the virus for the general population. Even so, stay safe mmi! We are witnessing a double black swan event with Corona virus and a oil price war at the same time, those kind of events doesn’t come around very often!
Not everything is super accurate what I wrote over a month ago, for example fortunately the actual mortality rate seems to be lower than my estimated span of 4-10%, but the rest of my thoughts still ring true I think: http://www.globalstockpicking.com/2020/02/09/the-situation-in-china-hong-kong/
I agree that the Robert Koch-Institute (RKI) should have published how many tests were performed. Since the RKI does not do all the testing themselves, they should have collected these data from all the laboratories. This is being done in countries like the U.S. or South Korea.
Furthermore, I believe that the initial testing capacity reported by quarks.de is too low. Why? Because the KBV itself reported that between 09.03. and 15.03. 100.000 tests were performed for walk-in patients alone (i.e. without the tests performed in hospitals). I believe that quarks.de confused 12.000 tests/day (this is a number I read elsewhere) and 12.000 tests/week. However, it is pretty safe to assume that in February, there were way fewer than 12.000 tests/day or even 12.000 tests/week performed (even though the capacity was there) because of the testing guidelines of the RKI. I believe that this was a huge mistake which allowed the virus to spread undetected within the population.
At the moment, people are only tested IF they show symptoms AND have been either in contact with a Covid19-positive person or stayed in a risk area (these are, as of now, Italy, Iran, the province Hubei in China, the province Gyeongsangbuk-do in South Korea, the region Grand-Est in France, Tyrol in Austria, the Spanish capital Madrid, the states California, Washington and New York in the U.S. and the area of Heinsberg in Germany) or show signs of viral pneumonia that cannot be explained otherwise. This means that people who got infected by another Covid19-positive person e.g. within Germany or elsewhere will not be tested and, if they do not show any symptoms or only very mild symptoms, will distribute the virus to others. It is therefore of utmost importance that a certain percentage of tests is performed on people who show symptoms but do not fulfil the rest of the criteria to get an estimate of how many infected persons have so far escaped detection. Most of these tests will (hopefully) be negative (and this is why even some experts think that this type of tests is unnecessary), but the few positive cases will allow novel contact tracing and therefore identification of novel infection routes.
According to reports in the press two days ago, the testing capacity in Germany is currently at its limits with around 20.000-30.000 tests/day. This means that further exponential growth in positive cases (which we will see for sure at least until the beginning of next week) cannot be attributed anymore to an increase in the number of tests performed. Only during next week we will see whether the measures taken by the federal and local governments will have some effects on reducing the spreading of the infection.
Thanks for the qualified comment. Next week will indeed be interesting If tests are actually counted we will be able to draw better conclusions.
Maybe something interesting to do – if you have some time before going to bed tonight:
1. Watch this video: https://www.youtube.com/watch?v=Kas0tIxDvrg&feature=youtu.be&fbclid=IwAR1S9tq_7JDcJcgSG7p5LkzzcirmivhAM4TCMssyiCgCYiH2-PN3y4PGD3o
2. Then ask yourself:
a. What is your estimate of the % of people who will get affected in your community/country/city?
b. What is the % of people that got infected in Wuhan and/or on the Diamond Princess Cruise ship (best conditions for the virus to spread).
3. Then read this: https://www.calcalistech.com/ctech/articles/0,7340,L-3800632,00.html
I hope you and your family and everybody else is well – and I think its important to keep some good social distance short term!
Good stuff. Thanks.
You and I must be getting very different news feeds….. Increasing numbers being a function of increased testing is all the messaging I’m seeing
As I have written in the post, I am not talking about my personal news feed, but the public discussion on TV especially in Germany. And the inability of the German institutes to count tests.
Great post, totally agree.
I think it’s typical of human beings acting in uncertainty. This peculiar species tends to grab hold anything that can serve as an anchor and will draw all kinds of conclusions that are many stretches too far.
Just like the daily story telling by the financial media on the stock market!
Stay safe all.
Virologists ahve thought about that and use critical cases numbers and number of deaths as a proxy for their best guess. In the case of covid-19, especially the critical cases, I think.
Well, in other countries like the US they actually track the tests. Estimating from Critical cases or death is useless for a new, inknown desease. The only help woudl be random testing. I reiterate, Virologists seem to be really bad/careless in statistics.
Good post and I couldn’t agree more on the lack of data and reliable information. I’d only question the final conclusion that the declining markets are also primarily the result of speculation.
There is no doubt, that the current (over-)reaction by international governments shutting down major cities, countries and economies will severely reduce economic output and cash flow for many industries in the near-term. This will lead to bankruptcies, increase in unemployment, decrease in consumer spending (other than toilet paper of course…) and potentially a significant recession. No one knows how long this might last, but the biggest capital market risk is that we had a bull market for more than a decade fueled by cheap money & leverage and ever increasing multiples which will fully deflate without earnings. Even if COVID-19 gets under control in a month or two (very optimisticly), it might still be too late for many companies and small businesses and there will be ripple effects throughout the global economy.
Thanks for this post. The hysteria is overwhelming currently. And that’s ok. The point being that modern academia treats statistics just as a side show so “experts” in this field (or any field for that matter) don’t know how to treat the numbers. The only ones in total disbelief are statisticians and (self-educated) number crunchers.
No, don’t take it easy. It most likely is serious. But keep a cool head. And most importantly:
Try to stay safe.
The economic consequences will be devastating. But, given enough time, the world will go on. This could result in a once-in-a-lifetime opportunity for buying real assets, including common stock of good companies. Just don’t underestimate how low we could go…
Best,
Tom