Most people picture a market top through a handful of much-reproduced images: crowded trading floors, magazine covers, taxi drivers offering stock tips. Those pictures create an expectation that tops arrive with visible mania, and that mania is something an attentive person could notice in time. Real tops seldom look like that at all. Through the highest stretches of a market, excitement is largely absent. What fills the space instead is a weary sort of conviction. Matters were always going to turn out this way, the reasoning runs, and the doubters have been wrong for far too long to deserve another hearing.
This conviction is hard to identify as optimism because it rests on evidence that is entirely genuine. Earnings really are growing, the structural story really does hold together, and the bearish arguments really have been dismantled one after another. Investors holding this view are not ignoring the facts, and their command of the facts has rarely been better. Nothing in those facts is wrong. The trouble is that by this point the good news has already finished happening. The stock of evidence capable of supporting prices has reached its maximum, and from there the only available direction is downward.
A second difficulty belongs to the structure of the thing itself. A top is a concept that exists only in retrospect, because it requires the subsequent decline in order to complete its own definition. In the moment, one new high is indistinguishable from any other new high. Both set records, both attract the same explanations, and both are experienced by participants as confirmation. Any method claiming to identify a top in real time is therefore passing judgement on a dataset that has not yet finished arriving.
Part of the illusion of obviousness comes from the way such periods are described afterwards. Once a decline has occurred, the account written later retains only those details that pointed towards it. The far larger body of evidence pointing elsewhere is quietly discarded. The resulting narrative is coherent, and its coherence has been manufactured entirely by omission. Any reader will reasonably conclude that the warnings were plain to see at the time. They were plain only to someone who already knew the ending, which is precisely the reader the account was written for.
If the mood itself cannot be measured, what remains observable is a change in the surrounding behaviour. Three such changes recur with some regularity across historical episodes. Discussion of risk disappears. Nobody refutes it, and it simply comes to seem a poor use of anyone's time. Time horizons shorten, as the same participants begin assessing the same assets over progressively shorter intervals. Patience with doubt declines, and those raising contrary arguments stop being treated as participants in a conversation and start being treated as people requiring correction.
What these changes share is that all of them occur in language, where they are easiest to dismiss. They forecast nothing at all, and they specify no timing whatever. What they describe is a condition, namely that the ability of participants to process contrary information is steadily deteriorating. Such a condition can persist for a very long time, and it may resolve into nothing. Its value has nothing to do with timing. It reminds an observer what sort of environment their own judgement is currently being formed in.
For a long-term investor the practical point is not learning to identify tops in advance. It is accepting that one is standing inside the same environment as everybody else. When somebody notices that contrary arguments have begun to feel unworthy of a reply, the observation says little about the market. It says a great deal about the state of their own thinking. That is one of the few things observable without any data at all. It is also, in hindsight, among the things people most often admit they noticed at the time and chose not to act upon.