Your models must be Split, merge & Delist Aware. That is the only way you can solve this problem.
When a company announces a merger/delisting date, it's often after seeking Shareholders approval, which often will be approved by the nation/authority. If not, the company will be criticized for misleading shareholders. So the news often is reliable and doesn't change.
If a company, whose shares are trading at say 100 USD announces a delisting date at a price of say 150 USD. All the market makers will be onto it as the stake is going to provide 50$ (Guaranteed/Risk-Free Return) on the delisting date, from their purchase date.
If you look at the above company data through your models, all models just consider it as an optimistic situation as the prices are rising consistently and they do the buying. Sometimes, the price would even exceed the buyback price, which is an opportunity to sell. But the models wouldn't be aware of such information.
The only way of doing it is to make the models aware of the splits, mergers and delistings. This information is public and could be processed. It is expensive, complex but this is the only way to go forward.
The information is provided by the respective indices.
For NasDaq Its here, NDAQ Press Releases by Company
For LSE, LSE Press Releases
For DAX, DAX Press Releases
For ASX, ASX Press Releases
For NSE, NSE Press releases
You should be able to find this information from the respective indices, if the index isn't mentioned above.
The data could be very specific [depending on the index]. Some indices just upload the pdf or file received from the companies. In such a case, you might have to manually process the information. Often, this wouldn't be the case, if you are validating top companies of an index. For micro/mid/small cap, this could be a problem.
The Index maintains the information of those companies, which are delisted too. However it might be provided in a different sub section. you might make a note of this.