Backtests without the flattering shortcuts
We test popular trading ideas on ten years of NIFTY 50 minute data, the way a careful trader would have to trade them, and publish the results, including the ones that fail.
It started with a long list of strategies
DataPointX began as a private research project: a decade of 1-minute data on every stock that has been in the NIFTY 50 since 2016, and a long list of strategies from courses, books, YouTube and academic papers to test.
Most of them looked convincing on a chart. Almost all of them lost money once three things were right: thecosts of every order, fills at prices you could actually get, and thestocks that were really in the index on each day.
That's useful to know, and it isn't what most trading content says. So we publish it: each strategy run exactly as taught, with the code, the costs and every period shown. With Ask, anyone can test their own idea the same way, in plain English.
Six rules behind every number
Costs come first
An edge smaller than the cost of a round trip isn't an edge. Every result is net of real charges, then stressed with extra slippage.
The right stocks
Only the stocks that were in the NIFTY 50 on each day. Testing today's list quietly adds years of winners.
Separate periods
2016–20, 2021–23 and 2024–26 are reported separately. One good period is a story, not a strategy.
Publish the failures
We don't hide losing results or tune settings until something looks good, and we say how many variants we tried.
Show the code
Every post includes the exact code that produced its numbers, so anyone can check or re-run it.
Research, not advice
No tips, calls or signals. We show what happened in the past and how to test an idea, never what to trade.
From a popular claim to a published result
- 1
Pick a popular claim
A setup that is widely taught or shared: a course, a PDF, a forum favourite or an academic paper.
- 2
Write exact rules
Entry, exit, stop, target and timing written as code, as close to the source as possible. Variants are decided before anything runs.
- 3
Run it on ten years of data
Every NIFTY 50 stock on the days it was in the index, next-bar fills, real costs, everything closed by 15:15.
- 4
Try to break it
Three periods, +3 bps slippage, a random-stock benchmark and per-year results, with automatic warnings.
- 5
Publish, whatever the result
The post, the charts, every variant's numbers and the code, with a Markdown version for tools.
Frequently asked
Do you give tips, calls or trade recommendations?
No. DataPointX publishes research on historical data and tools to test ideas. We are not a SEBI-registered adviser or research analyst, and nothing on the site tells you to buy or sell anything.
Why publish strategies that fail?
Because most popular intraday setups do fail once costs, fills and the right stock list are in place, and knowing that before you trade is worth more than another chart that looks good. A failed test, done properly, is a useful result.
Do you have strategies that work?
We keep our own trading research private, and we do not sell strategies. What we publish is what fails, why it fails, and the methods to check any idea yourself.
What data do you use?
1-minute and 5-minute bars for every stock that has been in the NIFTY 50 since January 2016 (51 million minute bars), daily bars, the index membership on each day, and public NSE and NSDL reports on F&O positions and FII flows. Prices are adjusted for splits and bonuses.
How do you count costs?
Typical Indian discount-broker charges: ₹20 per order, plus STT, NSE transaction charges, SEBI fees, stamp duty and GST, plus 1 bp of slippage on every fill. About 6.5 bps for a ₹5 lakh round trip. Every result is also shown with 3 extra bps of slippage per fill.
How accurate are answers from Ask?
Ask runs real backtests with the same engine and rules as our research, and shows the code it ran. The AI can still misread a question or choose an assumption you did not intend, so check the stated rules and the code before relying on a result.
Can I quote or share your research?
Yes. Quote numbers and charts with attribution and a link to the original post. Each post also has a plain Markdown version for tools and AI assistants.