The research Kyra is built on

Evidence at a glance

Full citations and context

Loss aversion → revenge trade detection

Kahneman, D., & Tversky, A. (1979). Prospect Theory: An Analysis of Decision under Risk. Econometrica, 47(2), 263–291.DOI: 10.2307/1914185

The foundational paper of behavioral economics. Kahneman and Tversky demonstrated that people weight losses roughly twice as heavily as equivalent gains — the same dollar amount feels significantly worse to lose than it does good to win. The finding earned Kahneman the 2002 Nobel Prize in Economics.

The trading application: loss aversion is the engine of the revenge trade. A trader who just lost $500 feels the loss more acutely than they would feel a $500 win, and the resulting motivation to "get it back" is what produces the revenge entry — typically larger, less planned, and worse-executed than the trade that produced the original loss.

The article-length explainer for users: How to stop revenge trading.

Disposition effect → premature-exit detection

Shefrin, H., & Statman, M. (1985). The Disposition to Sell Winners Too Early and Ride Losers Too Long. The Journal of Finance, 40(3), 777–790.DOI: 10.1111/j.1540-6261.1985.tb05002.x

Shefrin and Statman applied prospect theory to investor behavior and identified the disposition effect: investors systematically sell winning positions too early (to lock in the gain) and hold losing positions too long (to avoid realizing the loss). The pattern has been replicated across decades and asset classes.

The trading application: the disposition effect predicts that a trader with a discretionary exit will tend to take profits before their plan would, and tend to widen stops or let stops fail when the trade goes against them. Both errors are P&L-eroding even when the entry was correct.

Overconfidence after wins → oversizing detection

Barber, B. M., & Odean, T. (2001). Boys Will Be Boys: Gender, Overconfidence, and Common Stock Investment. The Quarterly Journal of Economics, 116(1), 261–292.DOI: 10.1162/003355301556400

Barber and Odean analyzed trading records from over 35,000 retail brokerage accounts and showed that traders who exhibited higher overconfidence (proxied by gender in this study, but the result generalizes) traded more frequently and earned lower returns. The mechanism: overconfidence leads to oversized positions and excessive trading frequency, which compound into worse risk-adjusted returns.

The trading application: a win sequence reliably elevates a trader's confidence in their next setup. The next position tends to be larger than the trader's risk discipline would normally produce — the "I'm hot, this is working" sizing creep. Over time, the oversized trades compound into outsized losses that erase the prior wins.

Self-awareness and trading performance → the entire Kyra thesis

Lo, A. W., Repin, D. V., & Steenbarger, B. N. (2005). Fear and Greed in Financial Markets: A Clinical Study of Day-Traders. American Economic Review, 95(2), 352–359.DOI: 10.1257/000282805774670095

Lo, Repin, and Steenbarger combined physiological measurement (skin conductance, heart rate) with trading-outcome data from professional day traders and found that traders with more intense emotional reactions to gains and losses had significantly lower P&L. Crucially, the study did not advocate suppressing emotion — it argued for awareness and structured handling of emotional states as they arose.

This is the closest published research to Kyra's product thesis. The paper makes the case that measuring emotion alongside outcome produces actionable insight; Kyra is the operational implementation of that case for retail traders.

The article-length explainer for users: How emotions affect trading performance.

Cortisol and risk-taking → after-loss caution

Coates, J. M., & Herbert, J. (2008). Endogenous steroids and financial risk taking on a London trading floor. Proceedings of the National Academy of Sciences, 105(16), 6167–6172.DOI: 10.1073/pnas.0704025105

Coates and Herbert measured salivary cortisol and testosterone in male professional traders over an 8-day window and found that cortisol levels rose with market volatility and that elevated cortisol corresponded to reduced risk-taking. The neurochemical signature is the same one observed in stress responses generally: cortisol elevation, narrowing of focus, conservative-decision bias.

The trading application: a trader who has just absorbed a significant loss is, neurochemically, in a state that biases against further risk. This is sometimes useful (it prevents the immediate revenge trade) and sometimes harmful (it makes the trader pass on the next setup that would have worked). The state is real; the question is how to handle it deliberately.

Time-of-day and decision quality → the afternoon-loss pattern

Danziger, S., Levav, J., & Avnaim-Pesso, L. (2011). Extraneous factors in judicial decisions. Proceedings of the National Academy of Sciences, 108(17), 6889–6892.DOI: 10.1073/pnas.1018033108

While not a trading study, this paper documented that judicial rulings followed a strong time-of-day pattern: favorable rulings dropped from ~65% at the start of a session to nearly 0% just before a break, then recovered after the break. The mechanism — decision fatigue — has since been documented in medical, financial, and consumer-judgment contexts.

The trading application: decision quality is not constant throughout a trading day. A trader who is sharp at 9:31am may be running on fumes at 2:31pm without consciously noticing. Time-of-day P&L patterns are often the most surprising patterns Kyra surfaces because traders rarely track them manually.

Sample size and inference → why we wait

Schönbrodt, F. D., & Perugini, M. (2013). At what sample size do correlations stabilize? Journal of Research in Personality, 47(5), 609–612.DOI: 10.1016/j.jrp.2013.05.009

The methodological backbone of why Kyra has tier labels. Schönbrodt and Perugini quantified how unstable correlation estimates are at small sample sizes (n < 250) and how the instability decays as n grows. Their finding generalizes to any rate estimate: small samples produce wide confidence intervals, and acting on small-sample point estimates as if they were precise leads to systematic overinterpretation.

The article-length explainer: How many trades before patterns emerge.

Why this matters for what Kyra is, and what it isn't

Three implications of building on research rather than on intuition:

1. The claims have ceilings. Kyra does not surface a pattern without enough data to back it. The literature above sets the floor — what can be claimed about a sample, at what size.
2. The patterns generalize. The behavioral mechanisms in the bibliography above are robust across decades, asset classes, and trader demographics. Kyra's specific patterns are personal to the user; the kinds of patterns Kyra looks for are the kinds the research has shown to matter.
3. The product cannot be replaced by an LLM. An LLM can describe the research above in plausible sentences. It cannot operate the inferential math against your trade history with sample-size honesty. The product is the math, not the description.

For an article-length introduction to how these mechanisms show up in retail trading, start with How emotions affect trading performance. For the statistical machinery Kyra uses on top of the research, see Pattern detection and The math behind Kyra.

If you find an additional behavioral-finance paper that should be in this bibliography, email feedback@neontigerlabs.com. The bibliography evolves as the research does.