How to Read a Trading Track Record (and Spot a Fake One)

Published 7 July 2026 · Artha Education · ~8 min read

Every Telegram tip channel claims "90% accuracy." Every Instagram trader posts screenshots of winning days. Almost none of it survives five minutes of honest scrutiny — and the skill of scrutinising a published record is worth more than any signal it advertises. Here's the checklist.

Why "accuracy %" alone is a meaningless number

The most common claim in Indian trading marketing — "90% accurate calls" — tells you nothing about profitability, because it says nothing about how much winners make versus how much losers lose. Two systems, both real:

System A is what "book small profits quickly, hold losers hoping" looks like — thrilling win rate, dying account. System B loses more often than it wins and prints money. (R is the risk unit from the 1% position-sizing rule: 1R = the amount risked per trade.) A record that shows accuracy without risk-reward is hiding the number that matters.

The seven things a real track record must show

1. Every trade — including losses and no-trade days. A record with no red is not a record; it's an advertisement. Real strategies lose 30–55% of the time, and honest publishers show days where the answer was "no valid setup, stayed out."
2. The call published before the outcome. Entry, stop and target must be timestamped ahead of the move, on a channel where posts can't be silently edited or deleted. "We told you about RELIANCE this morning" posted at 3:25 PM proves nothing.
3. A defined stop and target on every trade. Without a pre-stated exit, any trade can be declared a win eventually. "Target hit" is only meaningful if the target existed before entry — and if the stop that could have hit first is shown too.
4. Results net of costs. An NSE intraday round trip costs roughly 0.1–0.15% of turnover. Across 100 trades that's 10–15R — enough to flip many "profitable" gross records into losses. If costs aren't deducted, the record is gross fiction.
5. Results in R (or %), not screenshots of rupees. "₹47,000 profit today" is unverifiable and depends entirely on position size. "+1.5R" or "+0.8% of capital" is comparable, reproducible and honest about scale.
6. Sample size, stated plainly. 8 wins out of 10 trades is statistically almost meaningless — pure luck produces that regularly. Meaningful win-rate claims start around 30–50 trades, and even then an honest publisher quotes a range, not a point ("around 50–60%," not "58.3%").
7. A stated methodology. If the publisher can't explain why trades are taken — the setup, the filters, the exit logic — the record can't be evaluated, only believed. Belief is what the marketing wants; evaluation is what your capital deserves.

Red flags: the fake-record playbook

Two tactics deserve special mention because they're epidemic on Telegram:

The multi-channel trick

Run five channels. Post "buy" in two and "sell" in two, stay silent in one. Whatever the market does, at least one channel now has a "proven" record — delete the others and market the winner. The defence is rule #2 above: a public, unbroken, timestamped history in one place, gaps and all.

The survivor screenshot

Take 20 trades, screenshot the 7 winners, post them as "this week's calls." Each screenshot is real; the record is fake. The defence is rule #1: completeness. Ask one question — "show me every call from the last 30 days, including the losers" — and watch the conversation change.

A 60-second audit you can run on any service

Anything failing two or more rows isn't a track record. It's a funnel.

How Artha publishes its own record — judge it by the same rules

Artha's methodology auto-paper-trades its own signals every day and publishes the result at a public, no-login page. By the checklist above: every trade appears with entry, stop and target recorded when a confirming candle closes (never after the outcome); losses and no-trade days are published; 0.12% round-trip costs are deducted from every trade; results are shown in R; win rates are quoted with statistical bounds, not point estimates; and the methodology — opening range breakout with published filters — is documented for free on this site.

One more honest disclosure, in the spirit of rule #6: the public record is young — it began in July 2026, and a few days of trades prove nothing yet. That's exactly the point of publishing it daily and permanently: come back in three months and judge the full, unedited history by this article's own checklist.