There is about 7X surge (from about 2000 to 14840) in daily new cases of Coronavirus reported today in Hubei of China? No, this is statistical illusion due to different methods.
Old method requires full diagnosis (longer time, more steps) for confirmation. So, it is possible to have many undiagnosed cases which were not reported as confirmed # infected.
New method could confirm # infected after CT scan (faster), therefore number # infected “increased” suddenly overnight.
So, for 14840 “new” cases in Hubei for 12 Feb 2020, we need to divided into 2 methods for consistent comparison:
Old method = 1508
New method = 14840
Difference = 14840 – 1508 = 13332
So, for consistency in reporting, past data should have 2 system, either all old method or all new method (quite unlikely for past data with new method unless they have collected these data but not reported before).
So, it means for future data analysis, we need to add “adjustment” of difference in 2 methods for consistent trend analysis. Eg, 13332 has to be deducted from final 60327 world # infected for 12 Feb 2020, therefore only 46995 case if follow the consistent old method for all data collected so far, therefore only 1825 cases for 12 Feb 2020 (down from with 2071 cases on 11 Feb 2020).
It is important for authority to provide 2 different sets of number or do a one-time adjustment for all data. If not, readers have to analyze data using 2 different timeframes:
Old Method = 22 Jan 2020 to 11 Feb 2020
New Method = 12 Feb 2020 onward
====================
Inconsistent policy or reporting criteria is nightmare for analyst, both for Coronavirus and Financial Market (eg. Stock). In fact, both absolute number (eg. # infected on certain day, earning of company for certain quarter) and relative number (eg. weekly difference of # infected, quarterly or Y-O-Y earning difference of company) are important as they serve different purpose.
Absolute number is to benchmark against certain fixed criteria, eg max # infected for SARS vs Coronavirus, Price to Book ratio of an undervalue company (PB<1) .
Relative number is to show the trend, eg % share prices difference over last 1 week or 1 month during Coronavirus.
Adjustment is everywhere, not a nightmare if we are prepared. Eg, when company has changes in number of shares (stock split or consolidation), the price per share or earning per share can be adjusted accordingly.
The impact of new method in Coronavirus is probably similar to IFRS16 adopted in year 2019 by most countries (except for US, Europe and some countries), resulting in confusion as past financial data (before 2018) could not be adjusted easily, requiring reading financial reports to understand true impact of change.
Similarly, for the sudden surge (1 time) in Coronavirus case today should be viewed as policy change, analyst needs to adjust manually or best with help by authority to provide consistent data (more important than true data). For example, Singapore provide consistent data in # cases for Coronavirus and try its best to cover most cases (may not be all as some have no symptom, infected but not known), therefore reporting higher # infected than other regional countries outside China.
However, current global Coronavirus data of each country can still be trusted because the relative trend is consistent, despite each country may have somewhat different criteria and also effort in diagnosis.
====================
In fact, I have pointed out in earlier post that fatality rate of Hubei is 16 times higher than outside Hubei which is not reasonable. So, this confirms that there a significant reporting difference, resulting in # infected cases reported are less, therefore fatality in Hubei is much higher.
https://www.ein55.com/2020/02/hidden-statistical-analysis-of-coronavirus-with-stock-investment/
Bad News = Ignorant people becomes fearful as “actual” # infected is 7X higher (although could be 14X higher if we use same fatality rate to find hidden data)
Good News = higher # infected confirms the actual fatality rate of Coronavirus should be much lower, the “average” of 2% was based on old method, so if we use simple 2% / 7, actual average fatality could be around 0.3% which is reasonable.
Learn the right way of Technical Analysis in stock investment: www.ein55.com
