I usually say "I only look at on-chain data," especially focusing on mempool and slippage. Whenever I see a bunch of transactions with the same route and timestamp, I suspect being sandwiched... But recently, I got proven wrong: address profiling and tag clustering can be referenced, but don't treat them as IDs. A "smart money" tag might mean the team is moving funds around via multi-signature, or it could be stolen funds being washed; the same flow of funds may look like one line, but actually, it's separated by N intermediaries and intentionally fragmented.

The fee rates have been extreme these days, with the group arguing whether to reverse or keep pumping the bubble. At first, I thought: just compare on-chain net inflow and outflow, and you'll know. But I found everyone has become savvy—using CEX, cross-chain bridges, aggregators... what you see on-chain is just a shadow. Now I treat on-chain data as a radar, emotions as noise filtering. If I really want to act, I turn on slippage, private transactions, and MEV protection first. Don’t wait until you’re sandwiched and then review—it’s annoying.