The idea that smart criminals don’t get caught is born of a false dichotomy. Dumb criminals, like those who make the news and video clip shows, self-select for getting caught. But the contrary does not typically apply. Whether you get caught rarely has much to do with intelligence and more to do with dumb luck.
The reason for this is that no matter how meticulous the planning, no-one can forsee every detail that occurs in the course of a robbery or other elaborately planned crime except in fiction, where one authorial mind is omniscient. Indeed planning and surveillance may themselves draw attention to the culprit in real life. The more detail they know about a target, the more they are at risk of having exposed themselves by the very fact of trying to find those details out.
The problem for an offender is that they have to make a million unrehearsed, real time decisions to deal with the contingencies that arise as the offence unfolds, in circumstances where they do not have perfect information about what else is happening. Did that unexpected bystander see me or not? If I make the wrong decision and overreact or under react, I risk disaster. And as observed by others above, it is frequently necessary to balance risk against reward, such as in the case of bringing on board an accomplice who has necessary skills but who might rat you out. Sometimes, the risk crystallises into reality in a way that intelligence applied in advance can only ever have foreseen in terms of percentages, not absolutes. The more people you bring on board, the greater the chance that the inevitable real life game of Prisoner’s Dilemma that arises comes to favour someone ratting you out.
The police, on the other hand, are not constrained by real time. They have endless time to go back and reconstruct what happened, the timelines of various actors, a whole-of-event picture of which you the offender necessarily had only an incomplete view, and unravel the decisions you had to make on the run which no amount of intelligence can have guaranteed you got right. This asymmetry between the positions of crim and police is what essentially makes it a matter of luck rather than intelligence whether a crime succeeds. It is only with hindsight bias that we are tempted to call the decisions of the uncaught criminal “intelligent”; in reality things could have panned out a thousand different ways beyond the criminal’s control and resulted in his getting caught.
None of this means that I am implying that all criminals get caught. of course there are some who got away with it. Given the amount of crime, there must statistically be some small number who do it repeatedly and are never caught. But that is a function primarily of luck, of the bets going your way, rather than “intelligence”.
It is also the case that just because some crime (say, the burglary of your house) has gone unsolved, it does not follow that the culprit is free. Frequent offenders typically might get away with a fraction of their offences, but they go to prison for the others. The fact that some famous crime has gone unsolved does not mean the perpetrator is living it up on the French Riviera. He might just as well be in prison for some other heist that didn’t go so well for him. The more often you roll the dice, the greater the risk of crapping out.
A caveat. I am talking here about the sorts of crimes that the OP seemed to be interested in - property crimes like robbery and burglary, or contract murder even. Internet crimes that exploit transnational boundaries are in a different category. Many of those crimes are effectively presently unenforceable in the traditional sense, and law enforcement uses alternative methods to suppress such offences.
Similarly, there are cases like Joe Kennedy, who it is said made his stake from bootlegging which he then converted into a legitimate empire. That too seems slightly different from what the OP is talking about, which assumes persistent criminality, without the subsequent legitimisation that occurs in cases like Kennedy’s.