How we measure track records, in plain words
8 min read · data: NSE daily data for liquid company stocks, Nov 2023 – Nov 2026 · descriptive study, not advice
Every ready scan, every order block list and the daily Stock in Focus on tietiy carries a track record. We publish it whether it flatters us or not. This page explains exactly how those records are built, so you can judge them yourself. Our Scorecard shows all of them in one place.
A track record describes what happened to many past cases. It does not say what any one stock will do.
Step 1: find every past case
For a scan like New 52-week high, we go back through our stored NSE data, day by day, and find every time a stock matched the rule. Two filters apply:
- Liquid company stocks only. The stock’s average daily traded value over the last 20 sessions must be at least ₹1 crore, and its price at least ₹20. ETFs and index funds are left out. Very thin stocks make returns look bigger and less real than they are.
- No looking ahead. A case only uses information that was known at that day’s close. Prices are adjusted for splits and bonuses so a 1:2 split does not look like a 50% crash.
Step 2: measure the next 10 trading days
For each case we measure the stock’s return from that day’s close to the close 10 trading days later. Ten days is about two calendar weeks. It is long enough to matter and short enough to have many cases.
One honest note: you only see the list after the close, so in practice you could not have traded at that exact closing price. That makes our numbers, if anything, slightly kinder than reality.
Step 3: compare with the average stock, not with zero
This is the most important step, and the one most backtests skip.
If the whole market rose 3% in those 10 days, a stock that rose 3% did nothing special. So for each case we subtract the average 10-day return of all liquid stocks over the same dates. What is left is the excess: how much better or worse the case did than an ordinary stock at the same time.
Without this step, any list looks great in a rising market and terrible in a falling one. We learned this the hard way. In our own research, comparing with zero instead of the market average produced four “edges” that disappeared once we fixed it. Our Hinglish study Base rate aur drift benchmark tells that story.
We also report how often the cases rose at all, next to how often the average stock rose. The average stock rose over 10 days only about 45–48% of the time in our window, not 50%: a few big winners pull the average up while most stocks lag it slightly.
Step 4: count the cost of trading
A delivery trade (buying and later selling) costs roughly 0.4% in total: securities transaction tax on both sides, exchange and broker charges, and slippage. We do not subtract this from the figures. Instead we use it as a bar: an average excess smaller than 0.4 percentage points is no edge at all, because costs would eat it.
Step 5: run a luck check
Even random lists beat the average now and then. To see whether a gap is bigger than luck, we use a block bootstrap: we re-sample the past dates in blocks of 10 sessions thousands of times and see how often a gap this large appears by chance. Sampling whole blocks of dates matters, because cases on the same day, or on overlapping 10-day windows, are not independent.
This gives a p-value (how likely a gap this size is from luck alone) and a 90% range for the average excess.
Step 6: give the badge
The badge follows a fixed rule, written in our code and tested:
| Badge | Rule |
|---|---|
| ◷ Too new to judge | Under 250 sessions of history, or fewer than 30 cases |
| ▲ Beat the average stock | Excess above +0.4 points and luck check p < 0.05 |
| ▼ Trailed the average stock | Excess below −0.4 points and p < 0.05 |
| ≈ Mixed history | Everything else: too small to beat costs, or could be luck |
| ◎ Context, not a signal | Calendar or status lists (results this week, F&O ban, circuits); not scored |
What the records say today
As of 18 November 2026, among our ten scored price scans:
- ▲ Beat: New 52-week high (+0.78 points, 14,558 cases).
- ▼ Trailed: Delivery spike (−0.91 points) and Gap down 3% or more (−0.74 points).
- ≈ Mixed: the other seven, including Volume 3× normal, Gap up 3%, Closed above 20-day high, and Price above 200-day average.
So about one in three scored lists shows a clear difference from the average stock, and even those differences are small next to how much single stocks swing. Most popular chart conditions, measured honestly, look a lot like the average stock. We think that is worth knowing.
What a badge does not mean
- Not a forecast. “Beat” means the past cases, on average, did a little better. Many single cases still fell.
- Not advice. We do not tell anyone to trade. A badge is a summary of history, shown so that lists are not read with false confidence.
- Not permanent. Records update every evening. A badge can change when new cases arrive. Each record page shows its window and number of cases.
The limitation
Our data starts in late 2023, so most records cover two to three years. That includes rising and falling phases, but not every kind of market. Costs are an estimate, and taxes are not included. Base rates describe the past. They do not predict the future.
Related: How to use the Scan builder · What a 52-week high means