Smarter Tools, Smarter Investors?

Financial Literacy and Financial Behavior in Investment Analytics and Technology Applications

Introduction

Background and research problem

Investment technology creates a practical challenge for financial education: an investor may have access to sophisticated analysis without understanding how to use it. A dashboard can calculate returns, an automated adviser can suggest an allocation, and an AI assistant can summarize a company report. The quality of the resulting decision still depends on what the investor understands, what the interface emphasizes, and whether the information fits the investor’s circumstances. This paper examines that relationship through financial literacy and financial behavior.

The central problem is the gap between the availability of investment analytics and the ability to make informed decisions from those analytics. Financial literacy concerns the knowledge needed to interpret financial information. Financial behavior concerns actions such as participating in markets, comparing alternatives, and responding to risk. Investment analytics provides measurements and models, while technology applications deliver those outputs through interfaces and automated services. Treating these elements together makes it possible to ask a more useful question than whether someone uses an investment app: can the person explain the trade-offs behind the decision the app supports?

The project focuses on retail investment decisions, including decisions relevant to young adults beginning to participate in financial markets. Its scope covers portfolio dashboards, digital brokerage applications, robo-advisers, and AI-assisted research. It does not attempt to rank providers or predict which assets will perform best. Instead, it evaluates the information and reasoning that a technology-assisted decision requires. The guiding question, “Smarter Tools, Smarter Investors?”, remains an open research question rather than a claim that technology automatically improves financial outcomes.

Research questions and objectives

The main research question is: how do financial literacy and financial behavior interact with investment analytics and technology applications in the investment decision process? Four supporting questions give the topic a manageable structure.

  1. Which financial concepts help investors interpret portfolio returns, downside risk, diversification, and automated recommendations?
  2. How can the presentation of analytical information influence attention, comprehension, and investment choices?
  3. What descriptive relationships between financial literacy and financial behavior appear in a published young-adult survey?
  4. How could a controlled dashboard experiment test whether additional risk information improves investment comprehension?

The objectives are to integrate research on knowledge, behavior, and investment technology; describe a reproducible secondary-data analysis; demonstrate a transparent portfolio comparison; and develop a focused empirical extension. These objectives distinguish what the present project completes from what future research would need to establish. The literature review and preliminary analysis are completed components. The dashboard experiment is a proposed study, with no collected responses or estimated treatment effects.

Definitions and conceptual framework

Financial literacy is treated as the capacity to understand concepts relevant to financial decisions, including compounding, inflation, and diversification. In this paper, its practical application includes recognizing the difference between a historical return and a forecast, understanding what a drawdown measures, and identifying which assumptions underlie an investment comparison. Financial behavior refers to observable decisions or reported actions. Market participation and saving behavior are available in the secondary dataset; dashboard choices and information checking would require new observations.

Investment analytics refers to the use of data and quantitative methods to evaluate investments. The term includes descriptive measures, portfolio comparisons, scenario analysis, and predictive models. Technology applications provide ways to calculate, communicate, or act on those analyses. A dashboard mainly presents information, a robo-adviser can translate an investor profile into a portfolio, and an AI research assistant can help organize source material. Their functions differ, so a single measure of “technology use” would be too imprecise for the proposed study.

The conceptual framework proposes that interface design influences the information users notice and understand, which may influence their subsequent choices. Financial literacy may change the strength of that relationship. Available resources, investment experience, and the decision horizon may also matter. This is a proposed explanatory structure, not a causal model established by the survey. The framework makes comprehension an explicit outcome because a profitable short-term result can arise without sound reasoning, while a carefully reasoned decision can still experience a market loss.

Data and defined sources

The empirical component uses Sjam’s (2025) published survey dataset on financial literacy and asset participation among young adults in Indonesia. The source is Mendeley Data, Version 1, DOI 10.17632/k269sxcw6h.1, distributed under CC BY 4.0. The accompanying workbook is preserved unchanged. These are secondary survey records, not responses collected by the author of this project. The derivative dataset and analysis script document the transformations used here.

The original workbook contains 952 records. Restricting academic year to freshmen, sophomores, juniors, and seniors yields 817 undergraduate records. This selection excludes 50 graduate records and 85 records with missing academic year. The subset provides a consistent educational scope for the preliminary analysis. It is not established as a representative sample of all Indonesian students or all young investors, and the findings should remain specific to the observed records.

The source includes literacy indices, reported market participation, and a categorical saving measure. Two binary variables are created without replacing missing responses. The source participation label “1=participate” becomes 1, and “0=otherwise” becomes 0. The resulting variable retains the name stock_participation for reproducibility, although the source description includes stocks and mutual funds. Savings above consumption becomes 1; savings equal to or below consumption becomes 0. This recoding describes a reported spending-saving relationship, not the amount invested or a portfolio return.

Table 1. Analytical variables and available observations

Variable Observed N Missing N Interpretation
Basic literacy index 814 3 Source index retained without rescoring
Advanced literacy index 814 3 Source index retained without rescoring
Market participation 699 118 Participation = 1; otherwise = 0
Saving above consumption 720 97 Above consumption = 1; other observed categories = 0

Source: Author’s calculations from Sjam (2025), undergraduate subset, N = 817.

The construction and scoring direction of the literacy indices require clarification before a higher value can be interpreted as greater knowledge. The analysis therefore reports the original index values and uses neutral descriptions such as “higher mean index.” Negative values are retained rather than treated as errors. No index thresholds are introduced to classify respondents as financially literate or illiterate.

The survey supplies context for the literacy-behavior relationship, but it does not contain direct measures of app use, AI adoption, interface comprehension, or realized investment returns. It cannot test whether technology changes those outcomes. A separate constructed portfolio example illustrates analytical concepts, while a proposed experiment specifies the technology measures that future research would collect. Keeping these components distinct prevents a descriptive survey from carrying conclusions it cannot support.

Test dataset and preparation procedure

Table 2 presents the first six rows from the ten-row test_sample.csv included with the project. The excerpt makes the variable coding and missingness visible before analysis. Its row numbers indicate display order only and are not respondent identifiers. The analytical results use all eligible undergraduate records rather than this small excerpt.

Table 2. Excerpt from the published survey subset

Row Basic index Advanced index Participation Saving above consumption
1 −0.448 0.849 0 Missing
2 −1.388 −0.579 1 Missing
3 0.234 1.127 0 0
4 1.025 −0.544 0 0
5 −2.064 −0.579 1 Missing
6 0.841 0.274 0 1

Source: Sjam (2025), derivative undergraduate dataset. Index values are rounded to three decimals for presentation; the CSV retains available precision.

The preparation procedure reads the original workbook, applies the academic-year filter, maps the two outcome variables, and calculates available-case summaries. Comparisons between groups require both an observed outcome and an observed basic index. The analysis does not impute missing values or reinterpret blank outcomes as nonparticipation. The provided Python script produces the analytical CSV and a JSON file of summary results, allowing another reader to reproduce the sample counts and group means.

Preliminary findings and interpretation

Among 699 observed participation responses, 108 indicate market participation, or 15.5%. Among 720 observed savings categories, 112 indicate savings above consumption, or 15.6%. The remaining observed savings responses comprise 486 below-consumption cases and 122 equal-consumption cases. These figures describe reported financial behavior in the selected dataset. They do not measure investment success, diversification, or the quality of a financial decision.

The basic literacy index has a mean of 0.026 and a sample standard deviation of 1.003 across 814 observed records. The advanced index has a mean of −0.012 and a sample standard deviation of 0.990, also across 814 records. These values summarize the supplied scales. Their proximity to zero should not be read as evidence of low knowledge or low performance without a documented scoring interpretation.

Table 3. Basic literacy index by reported behavior

Comparison group Complete N Mean index Sample SD
Market nonparticipants 589 0.120 0.986
Market participants 107 −0.135 1.088
Saving at or below consumption 606 0.061 1.002
Saving above consumption 111 0.132 0.976

Source: Author’s calculations from Sjam (2025). Each comparison excludes records missing either the outcome or the basic index.

The unadjusted participant-minus-nonparticipant difference is −0.255 index units, based on 696 complete records. The above-consumption saving group has a mean 0.071 units higher than the other saving group, based on 717 complete records. These differences describe associations in the available observations. They do not establish that financial literacy reduces market participation or increases saving. Index direction, financial resources, prior experience, and the sample composition may affect interpretation. The small saving difference relative to the within-group dispersion also discourages a strong substantive conclusion from the means alone.

Missingness is material for the behavioral outcomes. Participation is missing for 14.4% of the 817 records, and saving behavior is missing for 11.9%. As an arithmetic sensitivity check, treating every missing participation response as nonparticipation would give 108/817, or 13.2%; treating every missing response as participation would give 226/817, or 27.7%. These are extreme bounds for the selected records, not estimates or confidence intervals. They show why the observed-case rate of 15.5% should always be reported with its denominator.

The preliminary findings support a more careful research design: knowledge, participation, and technology use should be measured separately. A respondent may understand investment concepts but lack funds to invest, or may participate while relying heavily on someone else’s recommendations. The survey cannot distinguish those explanations. The next empirical step should therefore measure interpretation of investment information directly rather than assume that participation alone captures informed behavior.

Worked example of portfolio analytics

A constructed example demonstrates the information that the proposed dashboard would communicate. Both portfolios begin at 100 and have four observed values, with no external cash flows. The inputs are teaching values, not historical market prices or survey responses. The sequence has no assigned calendar interval, so the returns are not annualized. Fees and taxes are excluded from the calculation.

Table 4. Endpoint return and observed downside risk

Portfolio Value sequence Cumulative return Maximum drawdown
A 100, 120, 90, 108 8.0% 25.0%
B 100, 108, 100, 106 6.0% 7.4%

Source: Author-constructed teaching example; calculations reproduced by portfolio_analysis.py.

Cumulative return equals the final value divided by the initial value, minus one. Maximum drawdown is the largest observed proportional decline from the highest value reached up to that point. Portfolio A falls from 120 to 90, giving a drawdown of 30/120 = 25%. Portfolio B falls from 108 to 100, giving 8/108 = 7.41%. A finishes with the higher return, while B has the smaller observed drawdown. The comparison demonstrates that an endpoint ranking omits information about losses encountered along the way.

The example also connects literacy to behavior. A user who understands percentage changes can explain why A’s positive overall return coexists with a large interim decline. A user’s investment horizon and tolerance for losses may affect how that information influences a choice. Neither portfolio is universally preferable from these observations alone. Future losses may differ, and four observations cannot reveal movements between them. An interface should therefore communicate the meaning and limits of its metrics alongside the numbers.

Proposed dashboard experiment

The proposed extension would randomly assign consenting participants to one of two dashboard versions. Both versions would display the same portfolios, endpoint returns, investment horizon, and an explicit assumption of no fees or taxes. The expanded version would additionally show value paths, maximum drawdowns, and short definitions. Holding the underlying investments constant isolates the combined effect of the additional information and explanation. It would not isolate the separate contribution of each added element.

Baseline financial literacy would be measured before assignment using a documented scoring procedure. The primary outcome would be a four-item comprehension score, with one point for each correct response: identifying the higher endpoint return, identifying the larger drawdown, recognizing that historical results do not guarantee future performance, and recognizing that a near-term withdrawal can make interim losses relevant. Questions and correct answers would be fixed before data collection. A pilot would check whether wording is understandable without revealing the intended answers.

Secondary outcomes would include the allocation selected, reported confidence, and the participant’s explanation of the choice. Confidence would be interpreted alongside comprehension: increased confidence with unchanged or lower accuracy would not count as improved understanding. The primary analysis would estimate the difference in average comprehension by randomized assignment and report an uncertainty interval. If a justified sample size permits, a prespecified interaction would examine whether the treatment effect varies with baseline literacy. Either direction is plausible: explanations may help less knowledgeable users catch up, or more knowledgeable users may be better able to use the additional information.

The study would specify recruitment, consent, exclusion rules, missing-outcome handling, and the smallest effect of interest before collection. Participants would make hypothetical decisions without placing trades. A later study could test AI-generated explanations or follow decisions over time, but those additions would require separate validation. The present proposal makes no claim that an experiment has occurred or that the expanded dashboard improves outcomes.

Literature review

Scope and organization of the review

This is a focused narrative review of foundational investment theory, financial-literacy research, and evidence relevant to digital investment decisions. It combines peer-reviewed studies with a regulator’s experimental research and an official description of automated advice. The sources serve different purposes: a theoretical portfolio model explains analytical relationships, an observational household study examines behavior, and a controlled interface experiment provides evidence about design effects. The review compares those contributions rather than treating them as interchangeable proof of one claim.

Financial literacy and investment participation

Lusardi and Mitchell (2014) examine financial knowledge as an investment in human capital and review its relationship with economic decisions. Their framework gives financial literacy an active role: knowledge is something people may acquire because it helps them make choices. This perspective supports examining whether investors can interpret financial information, rather than using familiarity with an app as a substitute for financial understanding.

Van Rooij, Lusardi, and Alessie (2011) study literacy and stock-market participation using modules in the Dutch DNB Household Survey. They report that respondents with lower literacy are less likely to invest in stocks and distinguish basic knowledge from more advanced financial concepts. Their work connects knowledge with participation, but it does not directly evaluate contemporary AI tools or dashboard comprehension.

For this project, the distinction between knowledge and participation is especially important. Participation can reflect available income, access, preferences, and experience as well as knowledge. The Indonesian subset and the Dutch study also differ in population and measurement, so the group means reported here should not be presented as a direct replication or contradiction. A stronger comparison would require comparable scoring, sampling, and controls. The proposed experiment instead uses a narrowly defined comprehension measure that directly matches the information shown on screen.

Financial behavior and active trading

Barber and Odean (2000) analyze brokerage records for 66,465 households over 1991–1996. They find that the households trading most actively substantially underperformed the market in their sample and discuss overconfidence as an explanation for high trading activity. The study links observed behavior to investment performance, showing why the frequency of financial activity should not be treated as evidence of decision quality.

The historical setting matters. Brokerage costs, technology, and access have changed since that period, so its numerical performance differences should not be projected onto present-day app users. Its relevance here is the evaluation principle: more activity is not sufficient evidence of improvement. A digital tool could make it easier to act without improving the reasoning behind an action. Accordingly, the proposed study treats comprehension as its primary outcome and allocation choice as a separate behavioral measure.

Financial behavior also encompasses restraint, information checking, and consistency with a stated goal. These actions are relevant even when a platform’s most visible activity measure is the number of trades. An evaluation could therefore ask whether users compare a benchmark, inspect a risk definition, or revisit an assumption before deciding. Those are proposed measures for future research, not behaviors observed in the current survey.

Portfolio theory and analytical interpretation

Markowitz (1952) formalizes portfolio selection through expected return and variance, with portfolio risk depending on covariance among holdings. The implication is that investments should be evaluated in relation to the portfolio rather than only as isolated assets. The framework supplies a theoretical foundation for tools that compare allocations and diversification.

This paper applies that foundation to the interpretation of digital outputs. Holding several funds does not by itself establish that their underlying exposures differ. A dashboard could make overlapping holdings visible and explain what a concentration measure represents. That is a proposed application of portfolio reasoning; the present analysis does not estimate an efficient frontier or recommend an optimal allocation.

Different measures also answer different questions. Historical cumulative return describes change between endpoints, while observed drawdown describes a decline from a prior peak. The worked example uses both to demonstrate why one number cannot fully describe an investment experience. Financial literacy matters because the investor must connect each measure to the decision being made, including the relevant horizon and the possibility of needing to withdraw money during a decline.

Digital interfaces and financial behavior

The Financial Conduct Authority’s (2024) research note reports an online experiment with more than 9,000 consumers examining flashing prices, push notifications, a trader leaderboard, and points with a prize draw. It finds that digital engagement practices can affect trading frequency and investment risk, with some evidence of larger effects among lower-literacy participants. This research directly connects interface features with financial behavior.

The experimental setting provides a stronger basis for assessing the tested design changes than a simple comparison of people who choose different apps. Its findings still concern specific features and experimental conditions. They do not establish that every notification is harmful, nor do they establish the effect of the educational risk display proposed here. They justify examining design as part of the investment decision process and treating literacy as a potentially relevant source of differences between users.

The proposed study shifts the design question toward understanding: does making downside information visible help users explain a trade-off? That question differs from whether a platform increases engagement. A well-designed analysis should allow for the possibility that a clearer display changes comprehension without changing allocation, or changes allocation without improving comprehension. Measuring both outcomes prevents those possibilities from being conflated.

Automated advice and investor inputs

The U.S. Securities and Exchange Commission (n.d.) describes robo-advisers as automated digital investment advisory programs that commonly use an online questionnaire to collect information about goals, horizon, financial circumstances, and risk tolerance. Services and approaches vary. This description identifies an important dependency: the recommendation is based partly on information supplied by the user.

The project’s framework suggests evaluating that input stage as well as the output. An investor may interpret a risk question differently from its intended meaning or may overlook a future cash requirement. A useful research task would ask whether users can explain how changing an input affects the suggested allocation. Such a task would connect financial literacy to technology application through an observable reasoning process, while avoiding an assumption that acceptance of an automated recommendation proves comprehension.

Automation can also be studied as a set of disclosed rules, such as when a portfolio is reviewed or rebalanced. The relevant questions include what triggers the rule, what costs the analysis assumes, and how a user updates changed circumstances. These are proposed evaluation criteria rather than findings about the providers in the published survey, which does not identify any robo-adviser use.

Machine learning and AI research assistance

Gu, Kelly, and Xiu (2020) compare machine-learning methods for estimating equity risk premiums and find predictive gains associated with models capable of capturing nonlinear interactions in their study. This establishes a research basis for examining advanced prediction methods in investment analytics. It does not demonstrate that any model marketed as AI will produce reliable returns for a retail user.

Predictive modeling and generative research assistance should be distinguished. A forecasting model generates a prediction evaluated against subsequent outcomes. A generative assistant can help summarize an earnings report or organize questions, with quality assessed by comparing its output to the original material. For the proposed application, an AI explanation would need to preserve the source figures, identify assumptions, and avoid presenting a plausible narrative as verified evidence. These are design requirements for a future tool, not measured benefits in the current project.

A technically complex output can also make evaluation harder for a user. An explanation should make clear what the model estimates and what remains uncertain. Financial literacy would then support asking appropriate questions: whether a figure is historical or forecast, whether costs are included, and whether the evaluation uses information available at the decision date. The paper treats those questions as part of informed behavior rather than requiring every investor to implement a model independently.

Evidence synthesis and the research gap

The literature supports several distinct links. Literacy research connects knowledge with financial decision making. Portfolio theory specifies relationships that analytical tools seek to measure. Trading research shows why activity and performance must be distinguished. Experimental interface research demonstrates that design features can affect behavior. Machine-learning research addresses predictive methods, while automated advice provides a setting where user information and software rules interact.

The gap addressed by this project is a focused measurement problem: how should a technology evaluation distinguish access to information, understanding of that information, and the action that follows? The paper does not claim that this is an entirely unexplored field. Instead, it combines the reviewed perspectives into a feasible student research design centered on investment comprehension. The original contribution would lie in the specific dashboard comparison and measurement protocol if the proposed experiment were conducted.

Two propositions guide that extension. First, risk information accompanied by a clear explanation may improve comprehension relative to an endpoint-return display. Second, the improvement may vary with baseline financial literacy. The second proposition deliberately leaves direction open because better-informed users may extract more value from added detail, while clear explanations may also reduce knowledge gaps. Both propositions require data and could fail to receive support.

Practical implications and study limitations

For financial education, the framework suggests teaching concepts through investment comparisons: explain a drawdown, identify overlapping exposures, and distinguish a prediction from a reported result. For technology design, it suggests evaluating whether people can interpret an output and connect it to a goal. A platform’s ability to calculate a metric accurately is necessary for meaningful analysis, but the proposed evaluation additionally asks whether users understand the metric’s limits.

The present evidence remains preliminary. The secondary survey cannot identify technology effects, the literacy-index direction needs clarification, and missing behavioral responses limit interpretation. The constructed portfolio paths illustrate calculations without representing actual assets. The proposed experiment would initially measure hypothetical decisions over a short period, so even a positive comprehension effect would not establish higher long-term returns or lasting behavior change. A broader empirical study would require documented recruitment, comparable measures across settings, and follow-up observations.

Conclusion

Investment analytics and technology applications become more meaningful when evaluated together with financial literacy and financial behavior. The paper’s central argument is that informed use should be assessed through the investor’s ability to interpret evidence and explain a decision. The published survey provides descriptive context, and the portfolio example shows why different analytical measures can lead users to notice different aspects of an investment. Neither component establishes a technology effect. The proposed dashboard experiment provides a practical next step for testing whether additional risk information helps users understand the choices in front of them.

Reference

Barber, B. M., & Odean, T. (2000). Trading is hazardous to your wealth: The common stock investment performance of individual investors. Journal of Finance, 55(2), 773–806. https://doi.org/10.1111/0022-1082.00226

Financial Conduct Authority. (2024, June 20). Research note: Digital engagement practices: A trading apps experiment. https://www.fca.org.uk/publications/fca-research/research-note-digital-engagement-practices-trading-apps-experiment

Gu, S., Kelly, B., & Xiu, D. (2020). Empirical asset pricing via machine learning. Review of Financial Studies, 33(5), 2223–2273. https://doi.org/10.1093/rfs/hhaa009

Lusardi, A., & Mitchell, O. S. (2014). The economic importance of financial literacy: Theory and evidence. Journal of Economic Literature, 52(1), 5–44. https://doi.org/10.1257/jel.52.1.5

Markowitz, H. (1952). Portfolio selection. Journal of Finance, 7(1), 77–91. https://doi.org/10.1111/j.1540-6261.1952.tb01525.x

Sjam, A. (2025). Survey data on financial literacy and asset participation among young adults in Indonesia (Version 1) [Data set]. Mendeley Data. https://doi.org/10.17632/k269sxcw6h.1

U.S. Securities and Exchange Commission. (n.d.). Robo-adviser. Investor.gov. https://www.investor.gov/introduction-investing/investing-basics/glossary/robo-adviser

van Rooij, M., Lusardi, A., & Alessie, R. (2011). Financial literacy and stock market participation. Journal of Financial Economics, 101(2), 449–472. https://doi.org/10.1016/j.jfineco.2011.03.006