Smarter Tools, Smarter Investors?

Financial Literacy and Financial Behavior in Investment Analytics and Technology Applications

Introduction

Problem and motivation

Investment analytics and technology applications influence what investors can measure, which alternatives they compare, and how they interpret uncertainty. This project asks whether smarter investment tools create better informed investors. Its central problem is the possible gap between analytical capability and human understanding: a platform can calculate a risk measure correctly while its user misunderstands what that measure means.

The scope includes portfolio analysis, brokerage dashboards, automated advice, and AI supported investment research. Financial literacy and behavior form the explanatory link between technology and decisions. Literacy supplies concepts such as diversification, compounding, and risk. Behavior concerns the choices people make when information, confidence, and financial goals interact. The project therefore examines financial literacy through its application to investment analytics rather than as a standalone budgeting topic.

Research questions and objectives

The overarching question is how investment analytics and technology applications interact with financial literacy and investor behavior to shape investment decisions. Supporting questions ask which analytical measures help investors compare alternatives, how interfaces communicate those measures, and whether baseline literacy changes the effect of additional risk information. The immediate objectives are to review relevant evidence, demonstrate a transparent portfolio calculation, analyze a published survey as preliminary context, and specify a feasible technology evaluation.

Data and defined sources

The empirical test dataset is the undergraduate subset of Amelina Sjam’s published 2025 Indonesian survey, available from Mendeley Data at https://doi.org/10.17632/k269sxcw6h.1 under CC BY 4.0. The original workbook is included unchanged. These are published secondary survey records, not interviews or responses collected for this project. The source contains 952 records. Selecting freshmen, sophomores, juniors, and seniors produces 817 records, excluding 50 graduate records and 85 records without an academic year.

The derivative CSV retains the source columns and adds two binary variables. stock_participation maps the source participation category to 1 for participation and 0 otherwise. saving_above_consumption maps savings above consumption to 1 and the two other observed categories to 0. Missing responses remain missing. The source participation field includes stocks and mutual funds. The basic and advanced literacy indices are retained as supplied; their construction and direction need clarification before interpreting a higher score as greater knowledge.

Variable Observed N Missing N Use
Basic literacy index 814 3 Knowledge indicator
Advanced literacy index 814 3 Knowledge indicator
Market participation 699 118 Investment participation
Saving above consumption 720 97 Financial behavior context

The included test_sample.csv presents the first ten undergraduate records restricted to the four analytical variables. It is a reproducible excerpt for inspecting coding and missingness, not an additional sample. All reported statistics use the full undergraduate subset with the relevant available observations. The survey contains no direct measures of app adoption, AI use, dashboard comprehension, or realized portfolio performance, so it cannot estimate technology effects.

Preliminary data analysis

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 participation result describes engagement with investments. The saving result supplies context about financial capacity and behavior, but does not measure investment skill or returns.

Comparison group Complete N Mean basic index
Market nonparticipants 589 0.120
Market participants 107 −0.135
Saving at or below consumption 606 0.061
Saving above consumption 111 0.132

The unadjusted participant minus nonparticipant difference is −0.255 index units. The corresponding saving-group difference is +0.071. The comparisons use 696 and 717 complete records, respectively, because both the outcome and basic index must be present. They are descriptive associations, not causal estimates. Missingness, the sampling process, financial resources, and index coding may influence the patterns. These results do not demonstrate that literacy reduces participation or that an investing app improves decisions.

Worked investment analytics example

A separate constructed example illustrates why investment technology should communicate more than endpoint returns. Both portfolios start at 100, with no external cash flows. The four observations are teaching inputs rather than historical prices or respondent outcomes. No annualization is appropriate because no calendar interval is assigned.

Portfolio Observed value path Cumulative return Maximum drawdown
A 100, 120, 90, 108 8.0% 25.0%
B 100, 108, 100, 106 6.0% 7.4%

Cumulative return equals final value divided by initial value minus one. Maximum drawdown is the largest value of one minus current value divided by the highest value observed up to that point. Portfolio A falls from 120 to 90, so its maximum drawdown is 25%. Portfolio B falls from 108 to 100, producing 7.41%. A has the higher endpoint return, while B has the smaller observed drawdown. These measures describe different features of the same paths and do not establish which portfolio suits every investor. Fees are excluded from this example.

Proposed technology evaluation

A future experiment would randomly assign participants to a return-first screen or an expanded screen displaying the same portfolios with paths, drawdowns, and plain-language definitions. Both screens would use identical fees and investment scenarios. A short baseline literacy assessment would precede exposure. The primary outcome would be a predefined comprehension score measuring whether participants can identify the larger drawdown, explain the return-risk trade-off, and apply the stated investment horizon. Allocation choice and confidence would be secondary outcomes.

The primary estimate would compare mean comprehension between assigned groups, reporting an uncertainty interval. A prespecified interaction with baseline literacy could assess whether the effect differs by knowledge if sample size permits. Recruitment, consent, scoring rules, and a sample-size justification would be established before collection. No participants have been recruited for this proposed experiment, and no experiment results are claimed. Adding fees or AI explanations as separate treatments would require a larger design so their effects can be distinguished.

Literature review

Portfolio construction and diversification

Markowitz (1952) formalized the trade-off between expected portfolio return and variance. The risk of a portfolio depends on asset weights and on the covariance among holdings.

A portfolio may contain several funds with overlapping positions. A holdings view can reveal shared exposure to a sector or to the same large companies.

Scenario analysis asks how an allocation would behave under a specified shock. A simple comparison can show the role of each asset class without assigning probabilities to the scenarios.

Machine learning and generative AI

Machine learning can estimate patterns in historical data. Gu, Kelly, and Xiu (2020) compare methods for estimating equity risk premiums and report gains from models that capture nonlinear interactions in their study.

Generative AI serves a different role. It can summarize an earnings report or draft questions for further research. An analyst can check the summary against the filing and trace any figures to the original table.

Prediction systems should be evaluated on later, unseen data. For generated summaries, key checks include source traceability, correct figures, and clear distinction between reported facts and interpretation.

Robo-advisers and automated rules

A robo-adviser may use a questionnaire about goals, time horizon, and risk tolerance as an input to a portfolio recommendation. Services and questionnaires vary across providers.

Rebalancing rules can restore portfolio weights after market movements. A rule can operate on a schedule or when weights move beyond a threshold.

The quality of the output depends in part on the information collected. An annual expense, near-term cash need, or changing risk capacity may affect whether a suggested allocation fits the user. Source: SEC Investor.gov.

Financial literacy as an analytical skill

Interest and compounding help investors interpret growth over time. Inflation helps distinguish purchasing-power gains from nominal gains. Knowledge of diversification helps explain why portfolio risk depends on how holdings move together.

These concepts matter when two funds look similar on a platform. An investor may need to compare fee assumptions, benchmark choice, the time window, and how much the funds overlap.

Lusardi and Mitchell (2014) frame financial knowledge as human capital. Their framework also allows investment experience to build knowledge over time.

Investor behavior and interface choices

An investor may react strongly to a recent loss, extrapolate a strong recent return, or follow what peers appear to be buying. These are plausible mechanisms for interpreting financial decisions.

A mobile app may emphasize daily price changes. A dashboard can instead show performance against a chosen benchmark and the investor’s stated horizon. These displays make different comparisons easier to see.

A useful research question is whether showing longer horizons, fees, and downside scenarios changes comprehension or choices when the underlying investment options are held constant.

Financial literacy and investment participation

Van Rooij, Lusardi, and Alessie (2011) examine financial literacy and stock market participation. Their work supports treating knowledge as relevant to engagement with investment markets. However, participation is only one outcome: it does not by itself establish diversification quality, suitable risk exposure, or the effective use of an analytical tool. This distinction motivates a project that measures comprehension directly instead of assuming that entering the market demonstrates informed decision making.

Data quality and evaluation of analytical tools

An analyst starts with a decision, such as comparing two allocations, then selects data that can answer it. Inputs may include prices, distributions, fees, holdings, and an appropriate benchmark.

The analyst aligns dates and currencies and checks missing observations. A return series should account for distributions where possible, and comparisons should use the same horizon and return conventions.

A spreadsheet can make assumptions easy to review. R or Python can repeat calculations across many assets or scenarios. A dashboard can share results alongside definitions and data dates.

A backtest applies an investment rule to historical data. One example compares a monthly rebalancing rule with a buy-and-hold baseline over the same period.

Using future information, excluding firms that disappeared, or selecting the best rule after many trials can inflate apparent performance. Fees, trading costs, and assumptions about execution also affect the result.

A credible test specifies the rules first, holds out a later test period, and records model changes. Human review is especially useful when the model’s inputs or market conditions change.

Synthesis and research contribution

The reviewed literature addresses different levels of the investment process. Portfolio theory explains the quantities an analyst seeks to evaluate. Machine-learning research studies predictive methods. Financial-literacy research examines the knowledge investors bring to financial decisions. Automated advice introduces a practical setting where a system translates inputs into portfolio guidance. Combining these perspectives suggests evaluating both analytical validity and user understanding.

The project proposes a connection among these strands rather than claiming a proven relationship between app use and investment outcomes. Technology determines which information is available and prominent. Financial literacy may affect interpretation. Confidence, attention, and the investment horizon may affect the subsequent choice. The proposed dashboard experiment tests one manageable part of that broader framework. The published survey supplies preliminary behavioral context, while the worked example demonstrates the analytical distinction that the interface would communicate.

The contribution is a framework for evaluating investment technology through decision comprehension. Its scope remains broad enough to discuss AI, automated advice, and portfolio dashboards, while its empirical extension has a specific and measurable outcome. The completed preliminary analysis supports feasibility and transparency; evidence of an actual technology effect requires additional data collection.

Reference

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

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

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

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

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

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