--- title: "Does machine learning work for trading | RupeeCase" description: "Machine learning can find structure in market data, but a model that scores well on the data used to build it has not shown a tradable edge." source_url: "https://www.rupeecase.com/learn/questions/does-machine-learning-work-for-trading" --- A model earns a further look only after it passes the held back test with costs The order of checks that a machine learning trading idea passes before its score is read. A map of method, not a result. Source: RupeeCase method map; SEBI study on F&O outcomes (Aug 2026). As of 16 Oct 2026. Machine learning is a set of methods that fit a function to data. In trading the function maps features, such as recent returns, volume and open interest changes, to a forecast or a trade decision. The method is rarely the weak point. The label, the features and the test are. Three things decide whether a model deserves a further test. The label must be what a trade could earn after costs, not an arbitrary next day direction. The features must be known at the moment of the trade, which rules out any field that is revised later. And the test must run on data the model never saw while it was built. Failures tend to come from a few repeated mistakes. Overlapping labels make the sample look bigger than it is. Many variants tried on one history make the best one look lucky. Survivorship and look ahead bias flatter the result. A model that loses in a fresh period is telling the truth, even when its earlier score was high. The flow below is a method map, not a result. It shows the order in which a model passes through checks before its score is worth reading. Each arrow is a stage where a model can fail quietly. SEBI's Aug 2026 study of all individual equity derivatives traders in FY26 is a reminder of the stakes. It found 87.7% lost money after costs, and fewer lost before them. The gap between the two is the traders for whom costs decided the sign, and a model that ignores costs has not tested what a trader would keep. **Before you trade.** Derivatives trading carries a high risk of loss. SEBI's study "Profitability of Individual Traders in the Equity Derivatives Segment (FY25 to FY26)" (August 2026) found that 87.7% of individual traders in equity F&O made net losses in FY26, with aggregate net losses of Rs 91,685 crore. [Read the SEBI study](https://www.sebi.gov.in/reports-and-statistics/research/aug-2026/study-profitability-of-individual-traders-in-the-equity-derivatives-segment-fy25-fy26-_103835.html) Does machine learning work for trading?Machine learning can find structure in market data, but a model that scores well on the data used to build it has not shown a tradable edge. The test that counts is a held back period, with costs charged, that the model never saw while it was being built. Partly true at best. Machine learning can find structure in data, but a high score on the data used to build a model is not evidence of a tradable edge, and the claim fails without a clean out of sample test and costs. LGo deeper ## Where Learn teaches this. [Go deeperWhat a backtest is, and five biases that ruin it9 min, Intermediate](/learn/path-19/module-19-1-backtest-biases)[Go deeperBacktesting options on Indian data11 min, Advanced](/learn/path-19/module-19-6-backtesting-options) SSources ## Go to the source. 4 sources, SEBI the most cited The publisher of each source listed below. Source: Module undefined source list. 1. [SEBI, study on F&O profit and loss of individual traders (Aug 2026)](https://www.sebi.gov.in/sebi_data/attachdocs/aug-2026/1787233506209.pdf) SEBI 2. [SEBI, safer participation of retail investors in algorithmic trading (circular, 4 Feb 2025)](https://www.sebi.gov.in/legal/circulars/feb-2025/safer-participation-of-retail-investors-in-algorithmic-trading_91614.html) SEBI 3. [SEBI, norms for sharing and usage of price data for educational purposes (30 day lag, from 8 May 2026)](https://www.sebi.gov.in/legal/circulars/may-2026/norms-for-sharing-and-usage-of-price-data-for-educational-purposes_101293.html) SEBI 4. Lopez de Prado, M., Advances in Financial Machine Learning, Wiley, 2018 (labels, purged cross validation, backtest overfitting) Book ?Questions people ask ## Related questions. [QuestionHow do people build an algo with Python and broker APIs?Intermediate](/learn/questions/build-an-algo-with-python)[QuestionHow do I backtest an option strategy in India, or in Python?Intermediate](/learn/questions/backtest-option-strategy)[QuestionMust a backtest include STT and costs?Intermediate](/learn/questions/backtest-include-stt)[QuestionWhat are SEBI's algo rules (static IP, 10 orders per second, algo ID)?Intermediate](/learn/questions/sebi-algo-rules)[QuestionIs algo trading legal in India for retail?Beginner](/learn/questions/is-algo-trading-legal-in-india) [All 130 questions](/learn/questions/) Explore next ## Go deeper on this question. * [QuestionWhat is a Kalman filter pairs strategy?A short answer with a dated chart](/learn/questions/kalman-filter-pairs) * [QuestionMean reversion or momentum: what is the difference?A short answer with a dated chart](/learn/questions/mean-reversion-vs-momentum) * [ToolOption payoff builderLegs on a dated Nifty chain, breakevens and charges](/learn/tools/payoff-builder) * [GlossaryInvesting glossary, 230 termsPlain word definitions, and the app’s own](/learn/glossary) Community ## Join my WhatsApp group. Scan the code, fillFill one short form, and join my WhatsApp group. Slides, sources, factsheets and the rebalance logs live there, and members join a daily Google Meet with me. New to RupeeCase? [Getting started](/getting-started) [See the terminal](/terminal) [Sign up for RupeeCase](https://invest.rupeecase.com/signup) [Join the WhatsApp group](https://forms.gle/C6HMtM1Z2GfeocQV7) [One form, one group](https://forms.gle/C6HMtM1Z2GfeocQV7)Scan with your phone camera Education only, not investment advice, a research report or a recommendation to buy, sell or hold any security. RupeeCase (QC Alpha Technologies Pvt Ltd) is not registered with SEBI as an investment adviser or research analyst. Contracts and prices shown are for illustration only, come from NSE end of day files at least 30 days old, and are not recommendatory. Derivatives trading carries a high risk of loss.