Why a trading journal works
You won't read “scientifically proven” on this page, because that would be a lie: no study in the world has tested trading journals for returns. What exists is something better. 25 years of behavioral finance research document which behavioral errors cost traders money, and psychology has experimentally measured what self-monitoring does to behavior. GlanWick builds on exactly this research. Here it is, with sources, honestly including what it does not show.
The problem, in numbers
Barber and Odean analyzed 66,465 households at a US broker (1991–1996): the heaviest traders earned 11.4% a year while the market returned 17.9%. The average household did better at 16.4%, but turned over 75% of its portfolio a year, and every turn costs.
In Taiwan, where the complete trading history of all individual investors exists, active trading costs individuals an aggregate 3.8 percentage points of return a year. The losses equal 2.2% of GDP. And among day traders there, less than 1% manage to earn positive abnormal returns net of fees, predictably and reliably.
That is the starting point any serious trading tool has to answer to. The interesting question is why so many keep going, so expensively, for so long.
Why your own patterns stay invisible
Research knows the mechanisms that hide patterns from their owner. A journal measures three of them directly.
The disposition effect
Traders realize gains about 50% more often than losses (Odean 1998, 10,000 accounts). The expensive part: the winners they sold beat the losers they kept by 3.4 percentage points on average over the following year. From the inside, the pattern feels like discipline.
In the journalThe analytics put holding time and outcome side by side: how long your winners stay open versus your losers. That's where the effect becomes visible as your own number.
Loss aversion
Losses weigh roughly 2.25 times as much as equal-sized gains (median estimate, Tversky & Kahneman 1992; the qualitative foundation is Prospect Theory, 1979). Doubling position size after two red trades is this asymmetry acting itself out, unnoticed.
In the journalPsychology tags and the losing-streak analytics show what you actually do after losses: size, frequency, setup fidelity. GlanWick proactively warns on losing streaks and overtrading.
Overconfidence
Men trade 45% more than women and pay for it with 2.65 percentage points of net return a year (Barber & Odean 2001, 35,000+ households). Overconfidence shows up as turnover, and turnover feels like diligence.
In the journalTrade frequency, fees and outcome sit in one view. Whether your turnover carries you or costs you is answered by your own statistics.
Context blindness
A trade is opened in one market environment but, without a record, judged in the memory of another. What happened around your entry (events, news, volatility) can't be reconstructed weeks later.
In GlanWickThe Market Context Engine records the context at entry: events, technicals, risks. The trade gets judged against the environment it actually happened in.
What recording demonstrably does
The strongest evidence for the journaling principle comes from psychology. A meta-analysis of 138 randomized experiments with 19,951 participants (Harkin et al. 2016) shows: people who systematically monitor their progress reach their goals measurably more often (d = 0.40). The effect grows when progress is physically recorded. In fairness: none of these experiments was a trading experiment.
And expertise research supplies the second half. Ericsson, Krampe and Tesch-Römer (1993) put it verbatim: without adequate feedback, efficient learning is impossible and improvement minimal, even for highly motivated subjects. A market doesn't give that feedback by itself; it rewards bad decisions often enough to cement them. A journal is where results become feedback.
So the chain is: documented, expensive behavioral errors, plus demonstrated effects of self-monitoring on behavior, plus feedback as a precondition of learning. We claim no more than that, and it is no less.
Unlimited journal, free forever. No credit card.
Sources
Every number on this page was checked against the primary sources on 2 September 2026. Each entry states what the study shows, and what it does not.
Shows: 66,465 US households, 1991–1996: the heaviest traders earn 11.4% a year while the market returns 17.9%.
Does not show: It does not show that trading less, or more thoughtfully, makes any individual profitable.
Shows: The complete trading history of all individual investors in Taiwan: an aggregate performance penalty of 3.8 percentage points a year, losses equal to 2.2% of GDP.
Does not show: It does not show that any intervention moves an individual trader to the winning side.
Shows: Taiwan day traders, 1992–2006: less than 1% predictably and reliably earn positive abnormal returns net of fees.
Does not show: It does not show what the persistence of the few profitable traders is based on.
Shows: 10,000 accounts, 1987–1993: gains are realized about 50% more often than losses, and the winners investors sell go on to beat the losers they keep by 3.4 percentage points on average over the following year.
Does not show: It does not show that merely knowing about the effect removes it.
Shows: The value function is steeper for losses than for gains: losses hurt more than equal-sized gains please.
Does not show: The often-quoted factor of 2.25 is not in this paper; it comes from the 1992 follow-up.
Shows: Median loss-aversion estimate: λ = 2.25. Losses weigh roughly 2.25 times as much as equal-sized gains.
Does not show: Lab experiments with wide individual variation; no evidence of any trading benefit.
Shows: 35,000+ households: men trade 45% more than women, and trading costs them 2.65 percentage points of net return a year (women: 1.72).
Does not show: Overconfidence is measured only indirectly via gender; the study does not show that self-reflection lowers the cost.
Shows: Meta-analysis of 138 randomized experiments (N = 19,951): monitoring progress improves goal attainment (d = 0.40), more so when progress is physically recorded.
Does not show: None of the 138 experiments concerns trading; d = 0.40 applies to goal attainment in general, not to returns.
Shows: Without informative feedback, efficient learning is impossible and improvement minimal, even for highly motivated subjects (verbatim, p. 367).
Does not show: There is no “10,000-hour rule” in it (that's Gladwell's popularization); the subjects were musicians, not traders.
Note: GlanWick is a financial information service and does not provide investment advice. This page describes research findings and product mechanics for informational and educational purposes; it is not a recommendation to act. Broker connections are strictly read-only, and every order inside the software is a simulation. Trading involves substantial risk, up to total loss.