The Streak That Proves Nothing

Every business owner has a decision they believe they got right and a mechanism they believe explains it. This article is about how little of that belief the evidence usually supports, and about a study where the people holding it most strongly were paid the least.

The Verdict, Stated First

Five claims, in descending order of confidence.

One. This is one of the better-evidenced findings in the series. A meta-analytic effect of 0.62 with a confidence interval well clear of zero, plus field data on professional traders measured against two independent outcomes including actual pay.

Two. The bias was inversely related to what traders earned. Not to a laboratory proxy, not to self-reported confidence, but to total remuneration. That is the strongest link between a behavioural finding and money that this series has reported.

Three. Its own meta-analysts warn the studies may not be measuring one thing. They report large heterogeneity in dependent measures and say the underlying phenomena can be of a different nature, which is the pattern this series has now found in five separate literatures.

Four. A run of successes is far weaker evidence of skill than anyone treats it as. On our own arithmetic, from a base rate of 25 percent, twelve consecutive wins leaves you only 74.8 percent confident, and reaching 90 percent takes nineteen.

Five. In any population large enough, somebody has the streak regardless. Where nobody is skilled at all, roughly 2.7 percent of people will string together eight straight wins in twenty attempts, which in a market of fifty thousand is over thirteen hundred people with a track record and a theory.

Our Grades For These Claims

Applying the scheme from the first article in this series.

Grade A for the effect existing in laboratory settings, from a meta-analysis whose abstract we obtained verbatim with a confidence interval.

Grade A for the field association with trader performance, from a peer-reviewed abstract obtained verbatim from four independent sources including the authors' institutional repository.

Grade B for the four inducing factors, which are consistently reported across sources and which we did not obtain from the 1975 paper itself.

Grade C for the mechanism, because a study named in the meta-analysis's own reference list is titled to contradict it.

Grade A for our own arithmetic, which is Bayes' rule and reproducible.

Our position: the finding is real, the field evidence is unusually good, and the mechanism is less settled than the effect. That combination is rare in this series and it is the right one to have.

A Note On Method

Everything here is verified to August 2026.

We obtained the traders study's abstract verbatim from four independent sources, including the lead author's own institutional repository[3][4]. We did not obtain the paper, its correlations, its effect sizes or its computer task.

We obtained the meta-analysis's abstract verbatim including its confidence interval[2], plus passages of its introduction and its reference list[5]. We did not obtain its analyses or its list of moderators.

We did not obtain the 1975 paper, and every statement about it here comes from later sources quoting or describing it, which for a paper of this influence is a limitation we flag rather than minimise.

We did not obtain the earlier 1996 meta-analysis, nor any of the challenging studies, and report them from citation records and titles.

All arithmetic is ours, uses invented parameters, and demonstrates a mechanism rather than measuring anything.

This article discusses research on judgment. It is not investment, trading or risk management advice, and nothing here is a recommendation about any strategy or security.

The Definition

What is being claimed, in the words attributed to the founding paper.

An academic source quotes it as "an expectancy of a personal success probability inappropriately higher than the objective probability would warrant"[1]. Another renders it as "an individual's tendency to overestimate their ability to control the positive outcome of an event."[5]

The citation is Langer, E. J. (1975), The Illusion of Control, Journal of Personality and Social Psychology, 32(2), 311–328, confirmed across at least four independent reference lists[1][6].

Four observations, ours.

The word doing the work is "inappropriately." This is not a claim that people are confident; it is a claim that confidence exceeds what the objective probability supports, which requires knowing the objective probability.

That requirement is why the research is almost entirely about chance tasks. You can only demonstrate an inappropriately high expectancy where the true probability is known, and in business it almost never is.

A source describes the common structure of the studies: "even though the participants' behavior was not the true cause of the result, they still believed they were in control of the outcome."[5]

And note what the definition does not say. It says nothing about whether the belief is harmful, which is a separate question the traders study addresses directly.

Four Factors That Carry No Information

What produces the illusion, and this is the practically useful part of the finding.

A source describing the 1975 paper records: "Langer (1975), who named the illusion of control, reports that factors typical of skill situations, such as competition, choice, familiarity, and involvement, can cause individuals to feel inappropriately confident."[1]

The meta-analysis expands the list: "some features of the chance situations make them resemble skill situations. These elements include familiarity with the task, choice, involvement, competition, or foreknowledge."[7]

Four observations, ours.

Read that list against a business meeting. Competition, choice, familiarity and involvement describe essentially every commercial decision anyone makes. They are not exotic laboratory manipulations; they are the standard conditions of work.

None of the four carries information about whether you control the outcome. Choosing your own lottery ticket does not change its odds. Being familiar with a market does not make it responsive to you. Competing for something does not make winning it a matter of skill.

Which means the four are cues that correlate with control in most settings and not in the setting being tested. That is a sensible heuristic misapplied, not a failure of reasoning, which is the same structure the twenty-sixth article found in a different literature.

And it produces a testable business prediction. The more a decision feels like your decision, the more likely you are to overestimate your influence on its result, and involvement is precisely what senior people have most of.

Why Those Four

The mechanism, as the meta-analysis states it.

"People experience the illusion of control in chance situations because they interpret the presence of skill-related factors as cues that one is actually being in control."[7]

Three observations, ours.

This is a cue-substitution account, and it is the same shape as the framework underlying the fifty-seventh article's subject: a hard question, being how much do I control this, is answered by an easy one, being does this feel like a skill situation.

It also explains why the illusion is not stupidity. In most of life the four cues genuinely do track control, and a person applying them is using a rule that usually works.

And it predicts where the illusion should be strongest: domains that look like skill and behave like chance. Trading is the paradigm case, which is why the field study went there.

And The Order Of Your Wins

A separate inducing factor, from the companion paper, and it is the one with the best title in this series.

Langer, E. J., and Roth, J. (1975), Heads I Win, Tails It's Chance: The Illusion of Control as a Function of the Sequence of Outcomes in a Purely Chance Task, Journal of Personality and Social Psychology, 32(6), 951–955[1][5].

The meta-analysis describes the finding: "if a person is expecting a series of particular outcomes, a larger number of successes at the beginning of a task favor the illusion of control."[7]

We did not obtain the 1975 companion paper and report this characterisation.

Three observations, ours.

The claim is that early wins matter more than total wins, which is a sequence effect and cannot be justified by any information the sequence contains. A run of successes at the start of a random series carries no more information than the same successes spread throughout.

The commercial implication is direct and uncomfortable. A venture that succeeds early will generate more confidence in its method than one with the same record achieved in a different order.

And the title states the asymmetry more sharply than any summary could. Heads I win, tails it's chance is a complete description of how outcome attribution works when the theory is not tested.

Point Six Two

The quantitative synthesis.

Stefan, S., and David, D. (2013), Recent developments in the experimental investigation of the illusion of control. A meta-analytic review, Journal of Applied Social Psychology, 43(2), 377–386, DOI 10.1111/j.1559-1816.2013.01007.x[2].

Its abstract: "The aim of the study was to estimate the effect sizes associated with different factors thought to induce the illusion of control considering different conceptualizations of the phenomenon. The overall weighted mean effect size (D) was .62, with the 95% confidence interval about the mean .49 to .75. Generally, experiments tend to obtain moderate to large effect sizes, and the illusion of control is constantly present as an effect of several situational and psychological factors."[2]

Three observations, ours.

A confidence interval of .49 to .75 is reported, which is more than most abstracts in this series have offered and which lets a reader see the precision rather than guess at it.

The lower bound of .49 is itself a moderate effect. Even the pessimistic end of this interval describes something worth knowing about.

And the phrase "constantly present" is strong language for an abstract. The authors are claiming robustness across conditions, not merely a positive average.

Which Is Large For This Series

Putting that number in context, using this publication's own record. Ours.

The fiftieth article audited every effect size this series had reported across its first forty-nine articles and found a median of 0.41 among values that were genuine measured effects.

Three observations.

At 0.62, the illusion of control sits well above that median and near the upper end of what this series has covered.

Using the translation the fifty-ninth article introduced, an effect of 0.62 corresponds to a correlation of about 0.30, which would order two people correctly around 60 percent of the time. Our own conversion, standard, and the assumption of normality is ours.

So this is a real and substantial effect and it is not deterministic. Both halves of that sentence matter, and this series has criticised writing that keeps only one.

The Warning In The Same Abstract

The sentence that follows the headline, and it is the one we would put first.

"Nevertheless, the large heterogeneity of the studies in terms of dependent measures suggests that the underlying phenomena can be of a different nature."[2]

Four observations, ours.

The meta-analysts are saying, in their own abstract, that the studies they aggregated may not be measuring one thing. That is a serious qualification on a pooled estimate and it is stated voluntarily.

This is the fifth literature in which this series has found the same structure. The twenty-eighth found three effects under framing; the forty-second found three under overconfidence; the forty-fifth found a goal and a method under confirmation bias; the fifty-third found the same dispute in dual-process theory.

The practical consequence is specific. A pooled effect size across heterogeneous measures tells you the average of several things, and if you are trying to predict one of them, the average may be the wrong number.

And we would credit the authors for it. Reporting an impressive headline and then qualifying it in the next sentence is the behaviour a reader should want, and it is rarer than it should be.

One Hundred And Seven Traders

The field study, and it is the strongest evidence in this article.

Fenton-O'Creevy, M., Nicholson, N., Soane, E., and Willman, P. (2003), Trading on illusions: Unrealistic perceptions of control and trading performance, Journal of Occupational and Organizational Psychology, 76(1), 53–68, DOI 10.1348/096317903321208880[3].

Its argument: "The authors argue that the task and environment faced by traders are conducive to the development of illusions of control and that individual propensity to illusion of control will be (inversely) related to trader performance."[3]

Its method and result: "Using an innovative computer task, designed to assess illusion of control in the field, data from 107 traders in four organizations showed individual differences in this bias to have a significant, inverse, association with performance, as measured by managers' ratings of trader performance and by total remuneration."[3]

Four observations, ours.

107 traders across four organisations is a real professional sample doing real work, which almost nothing else in this series can claim.

The design reasoning is stated in advance and is correct. Trading looks like a skill domain and contains a great deal of chance, which is exactly the combination the mechanism predicts will produce the illusion.

The task was purpose-built to "assess illusion of control in the field", and we did not obtain it, so we cannot tell you what it measured or how well.

And the direction is worth stating plainly. More of the bias went with worse performance, which is the prediction, and predictions confirmed in the field are worth more than predictions confirmed in a laboratory.

Measured Two Ways

The feature that makes this study unusually persuasive. Ours.

Performance was measured by managers' ratings and by total remuneration[3].

Four observations.

Those two measures have different weaknesses. Manager ratings are subjective and subject to everything the forty-fourth article documented about outcome-contaminated evaluation. Remuneration is objective and reflects institutional decisions rather than a rater's judgment.

A finding that holds on both is much harder to explain away than one holding on either, because the explanations that dismiss one do not dismiss the other.

Remuneration is the more remarkable of the two. This is a behavioural measure predicting what a bank actually paid someone, which is about as far from a laboratory proxy as this literature gets.

And we obtained no effect sizes for either association, only that both were significant and inverse. We would very much like to report the magnitudes and cannot.

Maladaptive, In This Context

The authors' conclusion, and the hedge inside it is deliberate.

"The authors conclude that, at least in this context, illusion of control is maladaptive and that it is productive to take an individual difference approach to the study of such illusions. Implications for debates about the costs and benefits of positive illusions are discussed."[3]

Four observations, ours.

"At least in this context" is doing real work. The authors are not claiming the illusion is universally harmful, and the closing sentence names a live debate about whether positive illusions have benefits.

That debate is genuine and this article does not resolve it. We obtained no evidence on the benefit side and report only that the authors regarded it as unsettled enough to reference.

The individual difference framing is a substantive claim: that propensity to this bias varies stably between people and can be measured. If so, it is a trait rather than a state, and traits are selectable for.

And the caution we would attach is the one the study cannot answer. An association between a measured bias and pay does not establish that reducing the bias raises pay, and nothing in the abstract claims it does.

Which Way Does It Run

The strongest objection to the strongest evidence in this article, which we raise because nobody else in our sources does and because it is not answered by anything we obtained. Ours.

The traders study reports an association between a measured bias and performance. Associations have at least three explanations, and the abstract establishes which one only by argument.

Four possibilities.

The bias causes poor performance. A trader who overestimates their influence takes positions the evidence does not support, and loses. This is the reading the authors argue for and it is the one the article has assumed throughout.

Poor performance causes the bias. This runs the other way and is not implausible. A trader whose results are poor has more unexplained variance to account for, and the illusion of control literature says exactly that people supply causal accounts for outcomes they did not cause. Losing could generate the belief rather than the reverse.

A third factor causes both. General analytical capability, or the disposition the fifty-fourth article's instrument was built to measure, could plausibly produce both lower measured bias and higher performance without either causing the other.

And the measure could be contaminated by the outcome. If the computer task was administered to traders who already knew how they were performing, the fifty-first article's finding applies: measurement taken by or from someone who knows the expected answer is not independent of it.

Three observations.

The authors' own framing is careful about this. They call the illusion "maladaptive" only "at least in this context" and describe the finding as an individual difference[3], which is a claim about a stable trait and is more consistent with the first and third readings than the second.

The remuneration measure helps against the fourth possibility, because pay is set institutionally rather than by the trader or the researcher, though it does not rule out a trader knowing their own pay when tested.

And we cannot resolve any of it, because we obtained the abstract and not the paper. A longitudinal design, or a measure taken before performance was known, would settle it, and we do not know whether this study had either. We report the association at the strength an association supports, which is high confidence that the two travel together and considerably less that one produces the other.

What A Winning Streak Is Worth

The arithmetic that makes this actionable, because the illusion of control is at bottom a claim about reading outcomes as evidence of skill. Our own calculation, Bayes' rule, invented parameters throughout.

Suppose a genuinely skilled person wins 60 percent of the time and an unskilled one wins 50 percent. Someone produces a run of consecutive wins. How confident should you be that they are skilled?

It depends entirely on the base rate, so we ran three.

From a base rate of 25 percent skilled: after 1 win, 28.6 percent. After 3, 36.5. After 5, 45.3. After 8, 58.9. After 12, 74.8. After 20, 92.7.

From a base rate of 10 percent: after 8 wins, only 32.3 percent. After 12, 49.8, which is still short of even odds.

From a base rate of 50 percent: after 8 wins, 81.1 percent. After 12, 89.9.

Four observations.

Twelve consecutive wins from a 25 percent base rate leaves you at 74.8 percent. That is a streak which would settle any argument in a boardroom, and it leaves a one-in-four chance you are looking at luck.

The base rate dominates the evidence, which is the whole lesson. The same twelve-win streak reads as 49.8, 74.8 or 89.9 percent depending only on how common skill is in the population, and nobody in a boardroom knows that number.

The reason streaks are so weak is that the skill gap we assumed is small. A 60 percent winner beats a 50 percent winner by ten points, and ten points takes a very long series to detect.

And if the real gap is smaller than ten points, which in most professional settings it plainly is, every figure above is optimistic.

How Many Wins To Be Confident

Running it the other way, which is the more useful form. Ours.

How many consecutive wins are needed to reach 90 percent confidence that someone is skilled?

From a base rate of 50 percent: 13 consecutive wins.

From 25 percent: 19.

From 10 percent: 25.

Three observations.

Nineteen consecutive wins is more than most professionals get opportunities for in the relevant window. In quarterly terms it is nearly five years without a single miss.

Which means that in most business settings, the evidence required to establish skill from outcomes alone does not exist, and confidence about who is skilled is coming from somewhere other than the record.

And that is the illusion of control operating at the level of an organisation rather than an individual. The four cues, competition, choice, familiarity and involvement, are supplying the confidence the outcomes cannot.

And Somebody Has The Streak Anyway

The complementary calculation, which connects this to the previous article. Our own simulation, 200,000 runs, invented parameters.

Suppose nobody in a population is skilled and everyone wins exactly half the time. How many people will nonetheless produce a run of eight or more consecutive wins within twenty attempts?

About 2.7 percent of them, by pure chance.

In a department of 50: about 1 person. In a firm of 500: about 14. In an industry of 5,000: about 138. In a market of 50,000: about 1,360.

Four observations.

In a market of fifty thousand, over thirteen hundred people have an eight-win streak having contributed nothing but a coin flip, and every one of them has a theory about why.

This is the fifty-eighth article's finding arriving from the other direction. That article showed what studying only winners produces from noise; this one shows how many winners noise produces, and the two are the same phenomenon.

The department figure is the one to hold onto. In a team of fifty, roughly one person will look exceptional on an eight-win run for no reason at all, and that person will be promoted, imitated and asked to explain their method.

And note the interaction with the illusion. That person is not lying. They experienced the wins, they were involved, they made choices, and the four cues are all present. The illusion of control is what turns their genuine experience into a confident causal account.

A Direct Challenge To The Mechanism

The contrary finding, which we report because the meta-analysis's own reference list contains it.

Listed among the meta-analysis's included studies: Martinez, F., Bonnefon, J. F., and Hoskens, J. (2009), Active involvement, not illusory control, increases risk taking in a gambling game, The Quarterly Journal of Experimental Psychology, 62, 1063–1071[5].

We did not obtain the paper and report its title and citation only.

Three observations, ours.

The title makes a specific claim: the behavioural consequence comes from involvement, not from the illusion. If that holds, the illusion is a correlate of involvement rather than the operative variable.

That would not eliminate the finding but would relocate it, in the way the forty-eighth article reported for a different effect. The practical advice would change from correcting a belief to reducing involvement in the judgment.

And it is inside the meta-analysis's own reference list, which means the pooled estimate of 0.62 includes at least one study whose authors dispute the mechanism being pooled. That is consistent with the heterogeneity warning and is a reason to hold the mechanism loosely.

The Boundary That Matters Most

The qualification with the largest implications for applying any of this, from a title we obtained and a paper we did not.

Also in the meta-analysis's reference list: Matute, H. (1996), Illusion of control: Detecting response-outcome independence in analytic but not in naturalistic conditions, Psychological Science, 7, 289–293[8].

We did not obtain it and report the title, which states a boundary condition directly.

Four observations, ours.

The title claims people can detect that their responses do not affect outcomes, under analytic conditions, and cannot under naturalistic ones.

If that is right, it is good news and bad news in the same sentence. The good news is that the illusion is correctable by changing how the question is posed. The bad news is that ordinary working conditions are the naturalistic ones.

It also suggests why the traders study found what it did. Trading is naturalistic, continuous and immersive, which is the condition under which the title says detection fails.

And it points at the only remedy this article can offer with any confidence, which is to make the question analytic: stop, write down the base rate, and compute rather than perceive. That is what the arithmetic sections above are for.

Three Variants Of One Citation

The bibliographic entry, since this series keeps a running count and this topic produced an unusual cluster.

The traders study, published in 2003, appears in one reference list dated 2010, twice, while carrying the correct DOI[9].

The lead author's surname appears in one source as "Fento-O'Creevy", missing a letter[10].

And the earlier meta-analysis's second author appears as "Benassi, V. A." in one source and "Benassi, V. S." in another, both otherwise identical[1][5].

Three observations, ours.

None of the three changes anything. Volumes, pages and DOIs are correct throughout, so every paper remains findable.

But three variants on one topic is the densest cluster this series has recorded, and the year error is the most consequential because a seven-year displacement changes which literature a paper appears to be responding to.

That brings the running count of bibliographic variants across this series to twenty-one.

What Actually Survives

Our reading, stated directly.

Five statements.

The effect is real and moderate to large. A meta-analytic 0.62 with an interval of .49 to .75, which is above this series' own median.

It predicts real outcomes in a real profession. Inverse associations with both manager ratings and total remuneration across 107 traders in four organisations.

The mechanism is less settled than the effect. One included study is titled to say involvement rather than illusion drives the behaviour, and the meta-analysts warn their studies may measure different things.

The inducing cues describe ordinary work. Competition, choice, familiarity and involvement are not laboratory artifices.

And the arithmetic is independent of all of it. A twelve-win streak is weak evidence of skill whether or not any of this literature holds, because that is a fact about Bayes' rule.

Your Forecasts

The first application. Ours, untested, and not investment or risk management advice.

Four points.

A forecast you made and involved yourself in is subject to at least three of the four cues by construction: choice, familiarity and involvement. Competition often supplies the fourth.

The specific error to expect is not overconfidence in the forecast's accuracy, which the forty-second article covered, but overestimation of your influence on the outcome, which is a different thing and licenses different actions.

The distinction matters because the remedies diverge. Overprecision is corrected by widening an interval. Illusory control is corrected by asking what would have happened without you, which is a counterfactual nobody constructs.

And the analytic-versus-naturalistic boundary suggests the format of the fix. Write the counterfactual down before the outcome, because a title we could not obtain says detection works analytically and fails naturalistically.

Your Best Performer

The second application, and the more uncomfortable one. Ours.

Four points.

On our own arithmetic, in a team of fifty where nobody is better than anyone else, roughly one person will produce an eight-win run by chance alone.

That person will have an experience of control that is entirely genuine, having been present, involved and choosing throughout, and the illusion of control literature says they will read that experience as method.

The organisational consequence is what the fifty-eighth article described from the other side. You will ask them to explain their approach and they will explain it, and the explanation will be sincere and uninformative.

And the practical test is the one that article gave. Ask what the base rate is: how many people with the same approach did not get the run. If nobody has counted, the account explains nothing.

The Cheapest Test We Know

One diagnostic, derived from the definition rather than from any study, and it costs nothing. Ours.

The definition requires an expectancy higher than the objective probability would warrant, which means the illusion is only detectable where you have stated a probability in advance.

So the test is: write down the probability before you act, and keep the note.

Four observations.

Without a recorded prior estimate, the illusion is undetectable by construction, because after the fact you cannot recover what you expected. The forty-third article established that memory reconstructs toward outcomes, so asking yourself later does not work.

With one, the illusion becomes arithmetic rather than introspection, which is exactly the analytic-versus-naturalistic distinction the boundary condition above points at.

And the second half of the test is the part people skip. Record what you expected would happen without your involvement, not only what you expected to happen. The illusion is about the difference between those two numbers, and only recording one of them makes the difference uncomputable.

This is the same instrument the series has now recommended in four articles for four different failures, which we take as a sign it is the durable one. A contemporaneous written number defeats hindsight, defeats outcome bias, defeats overprecision and defeats this, and it requires no insight into anybody's psychology.

The Opposite Error

A caution we want to state, because this article could be read as licensing fatalism. Ours.

Three points.

The literature concerns chance tasks, chosen precisely because the objective probability is known. Most business decisions are not chance tasks, and control over them is real and substantial.

Underestimating control has costs too, and the traders study's own closing line references "debates about the costs and benefits of positive illusions"[3], which is an open question we obtained no evidence on.

And the honest formulation is narrower than either extreme. The claim is that four specific cues inflate perceived control without carrying information about it, not that perceived control is generally wrong.

What To Do

Learn the four cues and notice them. Competition, choice, familiarity and involvement raise felt control without carrying information about actual control, and they describe ordinary commercial work.

Ask for the base rate before reading a streak. On our own arithmetic the same twelve-win run reads as 49.8, 74.8 or 89.9 percent confidence depending only on how common skill is, and nobody in the room knows that figure.

Count how many wins would actually settle it. From a 25 percent base rate, nineteen consecutive wins are needed for 90 percent confidence, which most professionals never get the opportunities for.

Expect the exceptional performer in a group of fifty. On our own simulation about one person in fifty produces an eight-win run from nothing, and they will have a sincere account of their method.

Make the question analytic. A title we could not obtain claims people detect response-outcome independence analytically and fail to naturalistically, which points at writing the counterfactual down rather than perceiving it.

Separate this from overconfidence. Overprecision is corrected by widening an interval; illusory control is corrected by asking what would have happened without you.

Hold the mechanism loosely. A study inside the meta-analysis's own reference list is titled to say involvement rather than illusion drives the behaviour.

Do not overcorrect into fatalism. This literature is about chance tasks by design, and control over most business decisions is real.

The Limits Of This Analysis

Several caveats matter. This article discusses research on judgment and is not investment, trading or risk management advice; nothing here is a recommendation about any strategy or security, and the applications are our own reasoning and untested. Everything is verified to August 2026. We did not obtain the 1975 paper, and every statement about it here comes from later sources quoting or describing it, which for a paper of this influence is a serious limitation. We did not obtain the 1975 companion paper on outcome sequence, nor the earlier 1996 meta-analysis. We obtained the 2013 meta-analysis's abstract, parts of its introduction and its reference list, and not its analyses, so we report no moderator results and no breakdown by inducing factor. We obtained the traders study's abstract verbatim from four independent sources and not the paper, so we report no correlations and no effect sizes for either of its two outcome measures, and we did not obtain the computer task it used. We did not obtain the study challenging the mechanism or the study on naturalistic conditions, and report both from titles and citations in another paper's reference list. One source is a practitioner website and one an information systems teaching wiki, both flagged at every use. All arithmetic is ours and uses invented parameters throughout, including a skilled win rate of 60 percent against 50, three base rates, and a twenty-attempt window, none of which any source states. The conversion of the meta-analytic effect to a correlation and to an ordering probability is our own and assumes normality that no source establishes. And an association between a measured bias and remuneration does not establish that reducing the bias would raise remuneration, which nothing in the abstract claims and which this article does not assert.

Related Reading

No. 61 One Percent Feels Like Five The half of prospect theory that explains insurance, lotteries and litigation. Computed with the published parameters, a one percent chance is acted on... Behavioural Finance The Interval Nobody Wants To Say The neighbouring error, which needs a different correction entirely. Behavioural Finance Less Than One Percent The translation method used here to convert an effect size into a decision.

Frequently Asked Questions

What is the illusion of control?

An academic source quotes the founding definition as an expectancy of a personal success probability inappropriately higher than the objective probability would warrant. The research is mostly on chance tasks, because you can only demonstrate an inappropriately high expectancy where the true probability is known.

What causes it?

Four cues typical of skill situations: competition, choice, familiarity and involvement, with foreknowledge and outcome sequence added by later work. None carries information about whether you control the result, and together they describe ordinary commercial work rather than any laboratory artifice.

How strong is the evidence?

Unusually good for this series. A meta-analytic effect of 0.62 with a confidence interval of .49 to .75, above this publication's own median of 0.41 across fifty articles. Plus a field study of 107 traders in four organisations where the bias was inversely related to both manager ratings and total remuneration.

How much does a winning streak prove?

Less than anyone treats it as. On our own arithmetic, assuming skilled people win 60 percent and unskilled 50 percent, twelve consecutive wins from a 25 percent base rate leaves you 74.8 percent confident. Reaching 90 percent takes nineteen consecutive wins, which most professionals never get the opportunities for.

Will someone always have a streak?

Yes. On our own simulation, where nobody is skilled and everyone wins half the time, about 2.7 percent of people produce a run of eight or more wins in twenty attempts. In a team of fifty that is roughly one person, who will have a genuine experience of control and a sincere account of their method.

Is the mechanism settled?

No. A study inside the meta-analysis's own reference list is titled to claim that active involvement, not illusory control, increases risk taking. And the meta-analysts themselves warn that large heterogeneity in dependent measures suggests the underlying phenomena can be of a different nature.

Does this mean I control nothing?

No, and we want that reading closed off. This literature studies chance tasks by design, because the objective probability has to be known. Most business decisions are not chance tasks and control over them is real. The claim is narrow: four specific cues inflate perceived control without carrying information about it.

References

1. Academic preprint evaluating cognitive biases in language models, section on the illusion of control, quoting the founding definition as an expectancy of a personal success probability inappropriately higher than the objective probability would warrant, attributed to Langer (1975); recording that people tend to overestimate their ability to control events, citing Thompson (1999); and recording that Langer, who named the illusion of control, reports that factors typical of skill situations such as competition, choice, familiarity and involvement can cause individuals to feel inappropriately confident. Together with a separate information systems teaching wiki entry confirming the citation as Langer, E. J., The Illusion of Control, Journal of Personality and Social Psychology, 32(2), 1975; recording Langer, E. J. and Roth, J., Heads I Win, Tails It's Chance: The Illusion of Control as a Function of the Sequence of Outcomes in a Purely Chance Task, Journal of Personality and Social Psychology, 32(6), 1975, pp. 951–955; and recording Presson, P. K. and Benassi, V. A., Illusion of Control: A Meta-Analytic Review, Journal of Social Behavior and Personality, 11(3), 1996, pp. 493–510. Note: an academic preprint and a teaching wiki, the latter flagged as non-peer-reviewed. We did not obtain the 1975 paper and every statement about it here comes from sources quoting or describing it. Note that this source gives the 1996 second author's initial as V. A. where reference 5 gives V. S. arxiv.org
2. Stefan, S., & David, D. (2013). Recent developments in the experimental investigation of the illusion of control. A meta-analytic review. Journal of Applied Social Psychology, 43(2), 377–386, DOI 10.1111/j.1559-1816.2013.01007.x. Repository copy reproducing the abstract: on the aim being to estimate the effect sizes associated with different factors thought to induce the illusion of control considering different conceptualizations of the phenomenon; on the overall weighted mean effect size D having been .62, with the 95 percent confidence interval about the mean running from .49 to .75; on experiments generally tending to obtain moderate to large effect sizes and the illusion of control being constantly present as an effect of several situational and psychological factors; and on the large heterogeneity of the studies in terms of dependent measures nevertheless suggesting that the underlying phenomena can be of a different nature. Note: a repository copy. We obtained the abstract in full and not the paper's analyses, so we report no moderator results and no breakdown by inducing factor. academia.edu
3. Fenton-O'Creevy, M., Nicholson, N., Soane, E., & Willman, P. (2003). Trading on illusions: Unrealistic perceptions of control and trading performance. Journal of Occupational and Organizational Psychology, 76(1), 53–68, DOI 10.1348/096317903321208880. Publisher record reproducing the abstract in full: on the paper examining the impact of illusory control beliefs on the performance of traders in financial instruments; on the authors arguing that the task and environment faced by traders are conducive to the development of illusions of control and that individual propensity to illusion of control will be inversely related to trader performance; on data from 107 traders in four organizations, gathered using an innovative computer task designed to assess illusion of control in the field, having shown individual differences in this bias to have a significant, inverse association with performance as measured by managers' ratings of trader performance and by total remuneration; on the authors concluding that at least in this context illusion of control is maladaptive and that it is productive to take an individual difference approach; and on implications for debates about the costs and benefits of positive illusions being discussed. Note: the publisher's record. We obtained the abstract in full and not the paper, so we report no correlations or effect sizes for either outcome measure, and we did not obtain the computer task used. bpspsychub.onlinelibrary.wiley.com
4. The lead author's own institutional research repository record for the same paper, confirming the citation as Fenton-O'Creevy, M., Nicholson, N., Soane, E. and Willman, P. (2003), Journal of Occupational and Organizational Psychology, 76(1), pp. 53–68, with the same DOI, and reproducing the abstract's opening. Note: the authors' own institutional repository, used as an independent confirmation of the citation details and abstract. oro.open.ac.uk
5. Repository copy of the 2013 meta-analysis carrying its reference list and introductory passages, recording the definition of the illusion of control as an individual's tendency to overestimate their ability to control the positive outcome of an event; recording that studies following the original experiments shared a common characteristic, namely that even though participants' behavior was not the true cause of the result they still believed they were in control of the outcome; and listing among its included studies Martinez, F., Bonnefon, J. F., and Hoskens, J. (2009), Active involvement, not illusory control, increases risk taking in a gambling game, The Quarterly Journal of Experimental Psychology, 62, 1063–1071; Morris, M. W., and Sim, D. L. H. (1998), Journal of Experimental Social Psychology, 34, 494–512; Presson, P. K., and Benassi, V. S. (1996); and Reuven-Magril, O., Dar, R., and Liberman, N. (2008), Journal of Abnormal Psychology, 117, 334–341. Note: a repository copy carrying the reference list. We obtained none of the studies listed. Note that this source gives the 1996 second author's initial as V. S. where reference 1 gives V. A. researchgate.net
6. Bibliographic service record for the 1975 paper, confirming the citation as Langer, Ellen J., The illusion of control, Journal of Personality and Social Psychology, 1975, volume 32, pages 311–328. Note: a bibliographic record used for independent confirmation of the citation. We did not obtain the paper. semanticscholar.org
7. Second repository copy of the 2013 meta-analysis carrying further introductory text, recording the corresponding author's affiliation at the Department of Clinical Psychology and Psychotherapy, Babes-Bolyai University, Cluj-Napoca, Romania; on some features of chance situations making them resemble skill situations, these elements including familiarity with the task, choice, involvement, competition, or foreknowledge; on people experiencing the illusion of control in chance situations because they interpret the presence of skill-related factors as cues that one is actually being in control; and on the sequence of desired outcomes influencing perceptions of controllability, such that if a person is expecting a series of particular outcomes, a larger number of successes at the beginning of a task favor the illusion of control. Note: a second repository copy, used for the mechanism and the sequence effect. We did not obtain the 1975 companion paper on outcome sequence. academia.edu
8. Publisher record for the 2013 meta-analysis carrying its reference list, identifying Matute, H. (1996), Illusion of control: Detecting response-outcome independence in analytic but not in naturalistic conditions, Psychological Science, 7, 289–293; Matute, H. (1995), Quarterly Journal of Experimental Psychology, 48B, 142–157; Lefcourt, H. M. (1973), The function of the illusions of control and freedom, American Psychologist, 28, 417–425; and Grou, B., and Tabak, B. M. (2008), Ambiguity aversion and illusion of control: Experimental evidence in an emerging market, The Journal of Behavioral Finance, 9, 22–29. Note: a publisher record carrying a reference list; citations and titles only. We obtained none of the works named, and the boundary condition reported in this article rests on a title rather than on findings. onlinelibrary.wiley.com
9. Repository page for a recent paper on trading psychology, carrying a reference list which dates the Fenton-O'Creevy and colleagues study to 2010 in two separate entries while giving the correct DOI 10.1348/096317903321208880 for a paper published in 2003. Note: a repository page reproducing a reference list. Recorded as a bibliographic variant: the year is displaced by seven years while the DOI is correct. researchgate.net
10. Practitioner investment website article on the illusion of control, citing Langer (1975), The Illusion of Control, Journal of Personality and Social Psychology, 32(2), 311–328; and citing the traders study with the lead author's surname rendered as "Fento-O'Creevy", missing a letter, while giving the journal, volume, issue and page range correctly as Journal of Occupational and Organizational Psychology, 76(1), 53–68. Note: a practitioner website, not an academic source, flagged at every use. Cited only to record the name variant; nothing in this article relies on it for any substantive claim. alphaarchitect.com

This article discusses research on judgment and is not investment, trading or risk management advice; nothing here is a recommendation about any strategy or security. The 1975 founding paper was not obtained and all statements about it come from later sources describing it. Neither the meta-analysis nor the traders study was obtained beyond its abstract, so no moderator results, correlations or effect sizes from either are reported. Two sources are non-academic and flagged at every use. All arithmetic is the authors' own and uses invented parameters throughout. An association between a measured bias and remuneration does not establish that reducing the bias would raise remuneration.