Paper: arXiv 2405.13959

Authors: Prajwal Naga, Dinesh Balivada, Sharath Chandra Nirmala, Poornoday Tiruveedi

Abstract

This research paper aims to investigate the efficacy of decision trees in constructing intraday trading strategies using existing technical indicators for individual equities in the NIFTY50 index. Unlike conventional methods that rely on a fixed set of rules based on combinations of technical indicators developed by a human trader through their analysis, the proposed approach leverages decision trees to create unique trading rules for each stock, potentially enhancing trading performance and saving time. By extensively backtesting the strategy for each stock, a trader can determine whether to employ the rules generated by the decision tree for that specific stock. While this method does not guarantee success for every stock, decision treebased strategies outperform the simple buy-and-hold strategy for many stocks. The results highlight the proficiency of decision trees as a valuable tool for enhancing intraday trading performance on a stock-by-stock basis and could be of interest to traders seeking to improve their trading strategies.

Complexity vs Empirical Score

  • Math Complexity: 3.5/10
  • Empirical Rigor: 6.0/10 (capped from 7.5: transaction costs not modelled; see penalty flags)
  • Quadrant: Street Traders — empirical and applied, lower math complexity (effective empirical rigor capped from 7.5)

Why this score: The paper applies standard machine learning (decision trees) without advanced mathematical derivations, but demonstrates high empirical rigor through detailed backtesting methodology, data preprocessing, and performance metrics for NIFTY50 stocks.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Evaluate Decision Trees for<br>Intraday Trading Strategies"] --> B["Data Input<br>NIFTY50 Equities Data<br>+ Technical Indicators"]
  B --> C["Methodology<br>Train Decision Tree Models<br>per Stock"]
  C --> D["Computational Process<br>Backtest Generated Rules<br>vs Buy-and-Hold"]
  D --> E{"Outcomes"}
  E --> F["Success<br>Decision Tree Strategy<br>Outperforms Buy-and-Hold"]
  E --> G["Neutral<br>Buy-and-Hold<br>Remains Viable"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes using decision trees to generate unique intraday trading rules for each NIFTY50 stock, aiming to improve performance over simple buy‑and‑hold.

Source sentence
the proposed approach leverages decision trees to create unique trading rules for each stock, potentially enhancing trading performance and saving time.

Assumes 4

  • Risk‑free rate is assumed to average 7.2% market
    Source sentence
    The risk-free rate is assumed to average 7.2% [5].
  • KPIs are annualized using 252 trading days and 375 intraday intervals per day statistical
    Source sentence
    It is pertinent to mention that the annualizing of KPIs utilizes 252 × 375 = 94500 periods per year [4], with 252 representing trading days per year and 375 denoting trading intervals within a trading day.
  • Commissions and slippage are ignored in backtesting execution
    Source sentence
    It's important to note that commissions and slippage were not explicitly incorporated into the backtesting process.
  • Trading signals are executed one step later to mimic realistic delay execution
    Source sentence
    the buying and selling signals were adjusted forward by one step to accommodate the trading delay, thereby simulating a realistic trading scenario.

Data

Backtest sample: NIFTY50 stocks · equities · intraday

  • ICICI Breeze one‑minute intraday data · ICICI Breeze market-prices commercial
    2022 to 2024 · NIFTY50 constituents
    Source sentence
    The ICICI Breeze API was the primary source for data, providing one-minute time interval data from 2022 to 2024.
  • Yahoo Finance daily adjusted close data · Yahoo Finance market-prices public
    NIFTY50 constituents
    Source sentence
    Yahoo Finance's daily interval data, which includes adjusted close prices, is used to compute adjustment factors for each trading day.

Costs

Not modeled A trading or execution result is reported without costs.

Source sentence
It's important to note that commissions and slippage were not explicitly incorporated into the backtesting process.

Finds 3

Reported performance (Decision tree-based intraday trading strategy): Sharpe 2.36 (benchmark 5.81) · total return 34.2% · max drawdown 2.6%

  • The decision‑tree strategy outperforms the buy‑and‑hold benchmark in Sharpe ratio, volatility and maximum drawdown, though its total return is lower on the test set. performance
    Source sentence
    The strategy demonstrated promising results, outperforming the benchmark in terms of Sharpe ratio, volatility, and maximum drawdown. However, it's noteworthy that the total returns of the strategy were lower than the benchmark in the testing dataset.
  • On the training dataset the strategy's Sharpe ratio is 3.37 versus 0.1 for the benchmark; on the testing dataset 5.81 versus 2.36. performance
    Source sentence
    Sharpe Ratio 0.1 3.37 OVER 2.36 5.81 OVER
  • PSBBR (percentage of stocks beating benchmark in total return) drops 42.50% from training to testing, while PSBBS drops 28.57%. performance
    Source sentence
    PSBBR and PSBBS values drop from the training to the testing dataset, but the drop is far more significant for PSBBR at a 42.50% drop compared to a drop of 28.57% for PSBBS.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 12 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (3)

  1. 6Chapter Decision Trees (2016) — link
  2. CLASSIFICATION AND REGRESSION TREES WITH GINI INDEX (2020) — link
  3. Decision Tree Algorithm in Machine Learning (2022) — link

0 of 3 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.