Skill-vs-Chance Labels Reorder 26% of Trainee Risk Rankings

Picture two cohorts of finance trainees at the same Mumbai institute, given an identical set of risky lending scenarios. One group is told the exercise measures analytical skill; the other is told it measures luck. Same numbers on the page, same payoff structure, same trainees on paper. Yet when the 26% reshuffle in risk rankings appears, it is not because anyone's judgment changed. It is because the label did. The question this raises is uncomfortable for anyone who designs training programmes in banking: how much of what we call "risk appetite" is actually a response to how we have framed the task?

The Label Is Part of the Instrument

In assessment design, we tend to treat framing as noise to be controlled for, a contaminant to be scrubbed out before the "real" signal emerges. The skill-versus-chance literature suggests the opposite: the label is a load-bearing part of the instrument. Ellen Langer's 1975 experiments on the illusion of control showed that people behave more confidently and take more aggressive positions when a task feels skill-relevant, even when outcomes are demonstrably random. The cue is not the odds; it is the perceived handle one has on the odds.

For training programmes in finance, this matters because almost every exercise we run carries an implicit label. A credit-risk simulation framed as "judgment under uncertainty" invites a different posture than the same simulation framed as "stress-testing your instincts." Trainees read the frame before they read the numbers. If a 26% reordering of risk rankings can be produced by relabelling alone, then our competency matrices are measuring something more entangled than we usually admit: a compound of decision quality and frame sensitivity.

Why 26% Is the Interesting Number

A quarter of the ranking shifting is neither trivial nor catastrophic. It is exactly the size of effect that gets dismissed in debriefs as "noise" and simultaneously large enough to move a trainee from the top quartile to the middle. In a hiring pipeline where the top decile gets the front-office rotation, a 26% reshuffle is the difference between two careers. The number is a warning that our rank orderings are less stable than the decimal places suggest.

What Behavioural Psychology Actually Predicts

The mechanism is not mysterious. Kahneman and Tversky's work on loss aversion established that the pain of a loss looms larger than the pleasure of an equivalent gain, and crucially, that this asymmetry is frame-dependent. When a scenario is presented as a skill test, losses feel like personal failure and invite defensive, conservative play; when the same scenario is presented as chance, losses feel like cost of doing business and invite bolder positions. The trainee's risk ranking moves not because their risk tolerance moved, but because the meaning of a loss moved.

There is a second mechanism worth naming: variable-ratio reinforcement. Behaviour that is rewarded unpredictably is the most persistent behaviour to extinguish, a finding with deep roots in operant conditioning. In a training context, intermittent feedback on risky decisions, some wins, some losses, no clear pattern, produces trainees who keep taking positions long after the evidence says stop. This is not a character flaw. It is a schedule effect. Programmes that give erratic, dramatic feedback on simulated trades are, perhaps unintentionally, building exactly the persistence profile that variable-ratio schedules are known to produce.

The Skill Illusion in Trading Floors

The classic demonstration is the "hot hand" and its cousin, the illusion of control in financial markets. Studies of retail traders have repeatedly found that people attribute skill to outcomes that are statistically indistinguishable from chance, and that this attribution increases with the frequency of feedback. More feedback, more illusion. A trainee who sees a real-time P&L ticker every few seconds gets a denser feedback loop than one who reviews decisions weekly, and the denser loop manufactures confidence faster than it manufactures competence.

Designing for Frame Awareness, Not Frame Removal

The practical implication is not to strip framing from exercises. You cannot; every exercise is framed by something. The implication is to make framing visible and to train against it.

Consider a concrete design: run the same risk scenario twice, weeks apart, once labelled "analytical judgment" and once labelled "scenario under uncertainty," and ask trainees to compare their own decisions across the two sittings. The learning is not in either score. It is in the gap. A trainee who discovers they took materially different positions under the two labels has learned something no lecture on cognitive bias can teach: that their risk appetite is partly a mood, partly a frame, and only partly a considered view.

Three Practical Moves for Programme Designers

Rotate the label deliberately. If your risk module always frames scenarios as skill tests, you are systematically sampling a defensive posture. Alternate frames across cohorts and record which frame each trainee saw. Over two or three cycles, you build a frame-adjusted score that is far more predictive than a single-sitting ranking.

Separate feedback density from feedback drama. Variable-ratio effects are strongest when feedback is both frequent and emotionally loud. Give trainees frequent feedback but make it boring: small, structured, comparable. Reserve the dramatic debrief for the end of the module, once, so it functions as a lesson rather than a schedule.

Score the reasoning, not the outcome. A trainee who takes a bold position for a well-articulated reason and loses is not the same as one who takes the same position on a hunch and wins. If your ranking rewards outcomes, you are ranking luck and calling it skill. Rubrics that weight the quality of the prior reasoning, blind to the result, are slower to administer and far more stable across frames.

The Forward View: Frame-Adjusted Competency

The next generation of finance training will not be able to afford the fiction that assessments are frame-neutral. As simulation tools become cheaper and more immersive, the temptation to add realism, tickers, live feeds, competitive leaderboards, will grow, and each addition is a frame with behavioural consequences. The programmes that get ahead of this will treat frame sensitivity as a competency to be measured rather than a nuisance to be suppressed. A trainee who knows their own decision-making shifts under skill labels, and can correct for it in the moment, is more valuable to a bank than one with a marginally better raw score.

The 26% figure is not a reason to distrust assessments. It is a reason to design them the way a good actuary designs a model: aware that the instrument changes the reading, and calibrated for it. Begin by auditing the labels on your existing exercises. You may find that a quarter of your rankings, and a quarter of your conclusions about who is "risk-averse" and who is "risk-hungry," were written by the frame rather than the trainee.