David Leitch has quietly become one of the most reliable stylists in American action filmmaking, a former stuntman who understands, better than almost anyone working at his budget level, that a sequence can be as thrilling in the cutting room as in the choreography. His films live or die on their craft rather than their scripts, and that trade has mostly worked out, Bullet Train and The Fall Guy both ran on sheer kinetic charm. How to Rob a Bank is the next entry in that mode, a social-media-age heist movie that is, visually, some of the most exciting work he has done. It is also a film whose script keeps tripping the style it is built to showcase, and the gap between those two things is exactly why it lands where it does.
How to Rob a Bank follows Ryan, the charismatic leader of a crew of bank robbers who broadcast their heists on social media, turning their crimes into viral content and themselves into folk heroes. Their growing fame draws the attention of a veteran FBI agent, who recruits a hacker living under house arrest to help track them through the digital trail they leave behind, setting up a cat-and-mouse that builds toward the crew’s most personal and ambitious job. Nicholas Hoult plays Ryan, and Zoë Kravitz, Anna Sawai, John C. Reilly, Christian Slater, and Pete Davidson also star. David Leitch directs from a screenplay by Mark Bianculli.
The high point of the film is the visual style, which is dazzling. The action is beautifully shot and cut in the way you now expect from Leitch at his best, and here he pushes it further into a hyper-stylized register, constructing the action sequences as though they are a piece of viral video, all kinetic energy and self-aware flash. It is a smart formal match for a story about criminals who package their crimes as content, and it is executed with real confidence. On the strength of its filmmaking alone, this would be a five; the craft is that assured.
What pulls it down are two fixable problems. The first is that too much of the movie, and especially its third act, runs on plot convenience. The reveal of the crew’s final target, the bank connected to Christian Slater’s character, is obvious long before the film treats it as a twist, because the moment Slater is introduced and the film lingers on that particular bank, any attentive viewer has already solved it. The frustration is watching the characters, and the police, take far longer to reach a conclusion the audience arrived at an hour earlier. The problem compounds across the back half, with crucial information surfacing at the exact moment the plot requires it and not a beat sooner. The FBI strike team does not work out what was actually being transported in an armored truck, an encryption key central to the climax, until essentially the minute before they need it, and the timing exists purely because the structure demands it rather than because anything in the story earns it. A heist movie lives on the audience believing the game of cat and mouse between the police and the team stealing whatever is being stolen being a believable battle of wits; this one keeps letting you get ahead of its characters, which drains the tension the style works so hard to generate.
The second problem is Zoë Kravitz, or more precisely the character she has been handed, because the performance itself is good. Her hacker is written inconsistently, lurching between positions as the plot requires rather than following any coherent internal logic, and the film never spends enough time developing her to make the lurches feel like a person rather than a function. There was a sharper version available: a character vocally opposed to the institution she has been coerced into helping, working against her own principles only because it is the price of her freedom, or one with a prior loyalty that forces a turn. Any of those would have given her an arc. Instead she mostly exists to deliver exposition, mostly when paired with John C. Reilly’s agent. Their dynamic too often amounts to one of them asking leading questions so the other can explain the plot, and their first real scene together, in her apartment, is essentially a stretch of her prompting him to narrate where the movie currently stands. Two talented actors are reduced to a mechanism for conveying information, which is a waste of both.
It is worth noting the film’s point of view, because it is more pointed than the genre usually bothers with. This is a heist movie with real anti-institutional teeth, a story in which the robbers are explicitly positioned against a banking system the film regards as the actual villain, their most personal target chosen because of a specific grievance against it. The messaging is a little muddied in places (someone as anti-capitalist as this would not be defensive of police as "just doing their job"), trying to be a crowd-pleasing action-comedy and a piece of economic anger at the same time, but it works well enough, and it gives the movie a spine that a pure heist romp would lack. It is also not hard to see why a film with this particular target got quietly dropped into the dead zone of mid-November, tucked between larger releases rather than given a real platform; a studio-backed movie that spends its runtime arguing banks are the bad guys is not always one its distributor wants front and center.
How to Rob a Bank is a very good time undercut by the sense that it was one script pass away from being a great one. The style is terrific, Hoult is a charismatic anchor, and the central idea, heists as viral content in an age of manufactured folk heroes, is a smart hook. That said, a heist film cannot keep letting its audience outpace its characters, and it cannot park two actors this good in purely functional roles, and those two failings are what separate the four it is from the five it could have been. It is well worth seeing in a theater for the filmmaking alone, and it may even sneak onto a year-end list, but it is the rare Leitch film where you walk out wishing the screenplay had been as sharp as the camera. See it for the style, and forgive the seams.
