We can predict behavior before the address is reported somewhere. Our thesis is based on the analysis of an address's account statement and we have trained artificial intelligence to recognize the usage pattern of an address through various sources (elliptic.co, kaggle, bitcoinabuse). Essentially, we look at the transactions of an address and predict the likelihood of it being a scammer. If the address is new, we can't make any predictions and advise treating it with caution, because the general rule of payment systems applies, namely that the creditor side should not seek anonymity.If, on the other hand, the address has at least 3 transactions, we can generate a prediction.
That's pretty much like I can just google on specific address I want to search and if listed on bitcoinabuse or any site related about it, the address is belong to scammer. Not all report in bitcoinabuse is legit, someone can just blindly report it including you, me or anyone else.
I think you should add walletexplorer.com as one of your source, that's a good explorer because you can search the particular address and you will find the wallet, you can also track the other address associated with the wallet as long as the address have been used.