The Millennium Criteria Assessment: What It Is and How It Is Scored

If you have been invited to Millennium's Criteria assessment, here is the short answer: it is the middle step of Millennium's three-stage early-career process, it is delivered through a third-party vendor called Criteria Corp, and the module candidates most often describe is 50 multiple-choice questions in 15 minutes. The most useful thing to understand before you sit it is that "Criteria" is the name of the platform, not the name of one test — which is why candidate accounts of it contradict each other so often.

The Criteria assessment at a glance

Everything in the right-hand column above is either from Millennium's own careers page or from Criteria Corp's own documentation. Where this guide relies on candidate reports instead, it says so explicitly.

What "Criteria" actually means, and why accounts disagree

The single most common mistake in guides to this stage is treating "the Millennium Criteria Assessment" as one fixed exam with one fixed structure. It is not.

Criteria Corp is an assessment vendor. Its catalogue includes cognitive aptitude testing, work-preference and behavioural questionnaires, and game-based modules. An employer assembles an event from that catalogue, and different roles at the same employer can be assigned different combinations.

That is exactly what candidates describe. For Millennium's recent cycles, reports vary by role: some candidates received the Criteria event after a technical assessment, some received it first, and some report receiving only Criteria with no separate technical stage at all.

This also explains why the assessment feels unpredictable in a way that Millennium's interview process does not. The interview stage is shaped by a specific portfolio manager with a specific seat to fill. The assessment stage is shaped by which role family you applied into.

Where the assessment sits in Millennium's process

Millennium's own students page sets out three steps: Screening, where candidates are reviewed for eligibility; Assessments, described as completing evaluations to match you with a hiring team; and Interviews, covering technical and behavioural rounds.

The phrase worth pausing on is "to match you with a hiring team." At most firms an aptitude test is a filter and nothing more. At Millennium, whose structure runs to 330+ investment teams across the platform, the matching function is real: the firm is not only deciding whether to proceed with you, it is deciding which pod you might fit. That is consistent with everything else about how the firm hires — you do not interview with one central pipeline, you interview with a team.

The official page also notes two constraints worth knowing before you apply at all: Millennium lists four tracks (Quantitative; Engineering & AI; Markets & Products; Business Services), states that it aims to hire its analyst class from its internship programme, and caps applications at two. That cap is the reason to apply deliberately to the track that actually fits rather than spraying applications across the firm.

The cognitive module: what 50 questions in 15 minutes is like

The module most candidates describe matches Criteria's published CCAT format, so the vendor's own documentation is the best available description of it.

The test is 50 items in 15 minutes — an average of about 18 seconds per question. Items mix verbal reasoning, numerical reasoning and logical or spatial problems, rotating between types rather than grouping them. Criteria states plainly that very few test-takers complete all 50 items within the limit.

That last fact is the one that changes how you should sit it. If you go in expecting to finish, you will pace yourself as though completion is the goal and lose time on items that were never worth solving. Not finishing is the normal outcome. The test is calibrated so that the time limit itself is part of the measurement.

How the scoring actually works

Criteria's scoring model is straightforward once you separate the two numbers.

Your raw score is simply the number of items you answered correctly out of 50, with nothing deducted for incorrect answers. That raw score is then converted into a percentile, which places you relative to other test-takers — an 80th percentile result means you scored higher than 80% of them.

Criteria publishes two useful anchors. The average score is 24 out of 50. And general guidance places a score of roughly 31 at around the top 20% of candidates.

The critical point about the threshold: the passing score is set by the employer, per position. There is no universal CCAT pass mark, and Millennium has not published one. Criteria's own guidance notes that more analytical and senior roles typically carry higher thresholds than entry-level ones — which, for a firm hiring into quantitative and investment seats, suggests the bar sits toward the upper end rather than at the average.

The item types, and how to budget time across them

Criteria describes the cognitive test as covering verbal, numerical (maths and logic), and spatial reasoning, rotating between those types rather than grouping them into blocks. That rotation is deliberate, and it has a practical consequence: you cannot settle into a rhythm on one kind of problem, and every few items you are switching cognitive mode.

Broadly, the item families behave differently under time pressure:

• Verbal items — analogies, antonyms, sentence completion and short reasoning statements. These are usually the fastest to resolve or to recognise as unresolvable, which makes them the cheapest items on the test. Answer them briskly and bank the seconds.
• Numerical items — arithmetic, ratios, percentages, number series and short word problems. These are where candidates lose the most time, because a problem that looks tractable often is not. Note that calculator permissions vary by event, and your invitation page is what tells you.
• Logical and spatial items — pattern completion, sequences and shape rotation. These are the most bimodal: you either see the pattern within a few seconds or you are unlikely to see it at all. They are the best candidates for immediate abandonment.

The time budget that follows is simple. At roughly 18 seconds an item you have no slack, so spend your fast items fast rather than carefully, and treat any item that has consumed more than about twenty seconds without resolving as a guess-and-move. Accuracy on the items you do answer matters far less than the number of items you reach, because there is no penalty for being wrong.

How this compares with other pod shops' screens

It is worth knowing that an aptitude screen of this kind is not the norm across the large multi-manager platforms, which is part of why the Millennium stage catches candidates off guard.

Point72's Academy runs a structured, cohort-based process built around a résumé screen and a superday, with the investment case study as the centre of gravity — see the Point72 Academy guide for the stages. Citadel's process is known for its technical depth and case work rather than a timed cognitive filter, covered in the Citadel interview guide. Balyasny's Catalyst route is built around a discretionary, hedged equity pitch, as set out in the Balyasny interview guide.

Millennium's use of a vendor assessment as an explicit matching step is therefore genuinely distinctive among its peers, and it is consistent with the firm's scale. With 330+ investment teams, a standardised early signal is one of the few ways to route candidates toward the pods that might actually want them. For the wider comparison of how these platforms differ, the major pod shops guide sets them side by side.

Five mistakes that cost candidates this stage

Most failures here are procedural rather than intellectual, which is the good news — they are all avoidable.

1. Pacing for completion. Candidates who aim to finish all 50 items rush the easy ones and make careless errors on questions they would otherwise get right. Criteria states outright that very few people finish. Optimise for items reached, not for the finish line.
2. Leaving blanks. This is the single most expensive habit, and it is entirely mechanical. With no penalty for wrong answers, every unanswered item is a discarded expected score. Plan the final sweep before you start.
3. Sitting it on the wrong device or a weak connection. A fifteen-minute timed test gives you no room to recover from a technical failure. Check the device and browser requirements on your invitation page the day before, not five minutes before.
4. Engineering the work-preference answers. These modules probe the same traits repeatedly through differently-worded items. A constructed profile produces contradictions the instrument is built to detect, and inconsistency reads worse than an ordinary honest profile.
5. Treating the assessment as the hard part. It is a gate, not the decision. The stage that determines whether you get a seat is the PM conversation, and candidates who spend a week on aptitude drills and no time on their pitch have optimised the wrong thing entirely.

Work-preference and game-based modules

Beyond cognitive aptitude, Criteria's catalogue includes work-preference and behavioural questionnaires and game-based assessments. Candidate reports indicate some Millennium events include these; you should not assume you will see all of them, and you should not assume you will see none.

These modules are scored differently and require a different approach. They generally test for consistency — the same underlying trait is probed several times through differently-worded items, and a profile assembled to look ideal tends to produce contradictions that the instrument is specifically designed to detect.

The practical advice is unglamorous: answer them honestly and quickly. Overthinking a work-preference item usually makes it worse, because deliberation is what produces the inconsistency. If you find yourself constructing the answer a hedge fund "wants", you are already doing the thing the module catches.

Engineering and AI candidates take a different route

If you applied into Millennium's Engineering & AI track, your experience is likely to differ substantially. Candidate reports for that track describe a HackerRank-based online assessment covering Python and data-science material, which is a different instrument on a different platform from the Criteria event.

Some candidates report receiving both, in either order. This is the clearest illustration of why the "one fixed test" framing fails — the assessment stage is configured per track, and the four tracks are genuinely different hiring pipelines that happen to share a company name.

How to prepare in the week before

The cognitive module rewards preparation, but not the kind most candidates default to. You are not learning material; you are rehearsing decision speed.

• Take timed practice tests on the official format. The pace is the whole difficulty. Practising untimed builds the wrong instinct entirely, and practising on a differently-formatted test builds familiarity you cannot use.
• Drill the abandon reflex. Set a hard personal limit — twenty seconds, say — and physically move on when it passes. Most of the score improvement available to a prepared candidate comes from not sinking time into two or three unsolvable items.
• Rehearse the final sweep. Know how you will fill every remaining answer in the last stretch. This is free score, and candidates who have not planned for it routinely leave items blank.
• Check your setup the day before. Device requirements, browser, calculator permissions and a stable connection. A technical problem inside a 15-minute test is close to unrecoverable.
• Do not neglect the interview. Passing the assessment gets you to the stage that actually decides the outcome. The Millennium interview guide covers what a portfolio manager tests in your pitch, and the risk-limit mechanics explain the culture behind those questions.

What happens after you submit

Criteria's candidate FAQ is explicit on this, and it answers the question most candidates ask next. Your results are emailed directly to the employer that administered the assessment. They are treated as part of your confidential candidate record, they are not shared with other companies, and Criteria's own support team cannot access them — which means contacting the vendor for your score or your status will not work.

All follow-up therefore runs through Millennium's recruiting process, not through Criteria. And because the stated purpose of the stage is matching you to a hiring team, the next step is not necessarily a uniform one: candidates proceed into conversations with specific teams, at a pace set by which pods are actually hiring.

The bottom line

The Millennium Criteria assessment is best understood as a vendor-delivered evaluation stage whose exact shape depends on the track you applied into, most commonly including a 50-question, 15-minute cognitive test that almost nobody finishes.

Prepare for the clock rather than the content. Answer every item, because blanks cost you outright. Answer the work-preference sections honestly, because consistency is what they measure. And treat the whole stage for what Millennium says it is — a matching step on the way to the interviews that actually decide whether you get a seat.

For the stage that follows, and what a portfolio manager is really testing when you pitch, read the full Millennium interview guide. For how the platform itself works — 330+ pods, hard drawdown limits and pass-through fees — start with how pod shops work.