We Published Every Trade for 30 Days, Before the Outcome. Here Are the Results.

Published 21 August 2026 · Artha Education · ~9 min read

Anyone can post a winning trade after the market closes. The only claim that costs anything to make is the one published before the outcome exists — where the timestamp, not the marketing, decides whether you were right. For 30 trading days we published every single setup our system produced, in public, ahead of the result. Losses stayed up. Days with no trade were posted too. Nothing was edited or deleted. This is the full score.

The rule we set on day one

On 12 July 2026 we started what we called the Glass Box Challenge, with three conditions written down before the first trade:

1. Every setup gets published before it resolves. Posted to a public Telegram channel on a 15-minute delay from when our system recorded it — hours before the trade would close either way. The post says which stock and which direction, and states plainly that the result will be published tonight, win or lose.
2. Every result gets published, in R. At 15:35 IST the outcome of each trade posts to the same channel — including the losses, including the days the answer was "no valid setup, we stayed out."
3. Nothing is ever deleted or edited. No quiet pause on a bad week. A public channel keeps its own history; the archive is the proof.

That third rule is the one that makes the first two mean anything. A track record you can only see the current version of is a brochure. A channel you can scroll backwards through is evidence.

The final score

Thirty trading days, 12 July to 21 August 2026. All results are R-multiples — the profit or loss as a multiple of the amount risked on that trade — and all are net of a 0.12% round-trip cost charge covering brokerage, STT, exchange fees and roughly 0.05% slippage.

Translated into money at a fixed ₹1,000 risked per trade — the 1% rule on a ₹1,00,000 account — that is {RUPEES} across the whole challenge, on paper. We publish the R figure first and the rupee figure second on purpose: the rupee number depends entirely on position size, which is why screenshot-based "₹47,000 today" claims tell you nothing.

What these numbers do not prove

We would rather say this ourselves than have it said for us.

{SIGNALS} trades is a small sample. It is enough to notice a pattern and nowhere near enough to establish one. A win rate measured over a few dozen trades carries a confidence interval wide enough to drive a truck through — the honest statement is a range, not a decimal. Our own signal scoring refuses to trust a per-stock win rate until it has at least 10 trades behind it, and even then it uses the pessimistic end of the interval. Apply the same scepticism here.

Thirty days is one market regime. This window was calm — India VIX spent most of it in the low teens, and breakout strategies flatter themselves in low-volatility trending conditions. The same methodology in a choppy, gap-heavy month would produce a different table, and eventually will. A 30-day record's real function is to prove the process is honest, not to prove the edge is durable.

These are paper trades. Our system records entries at the breakout level only after a candle closes beyond it, settles against real 5-minute candles, charges costs, and resolves same-candle ties in favour of the stop — the conservative choice every time. That is a serious simulation, and it is still not the same as a real fill in a fast-moving market. See paper trading before real money for what a simulation can and cannot teach you.

The no-trade days matter more than the wins

On {NOTRADE} of the 30 days, no setup cleared the filter bar — so nothing was published except a note saying exactly that.

Those are the days a tip channel cannot afford to have. A service that has to justify a subscription by delivering daily calls will manufacture a setup on a day that does not have one, and those manufactured trades are where retail accounts actually die. Sitting out is not a gap in the record — it is the part of the record that shows the filters are load-bearing. If you take nothing else from this article, take the habit: a day with no valid setup is a successful trading day.

Costs ate {COSTR}R — and that is the normal amount

Across {SIGNALS} trades, the 0.12% round-trip charge consumed {COSTR}R of gross profit. That is the line item most published records simply omit, and omitting it is how a break-even strategy gets advertised as a profitable one.

The arithmetic is unforgiving at higher frequencies. An intraday round trip on NSE costs roughly 0.1–0.15% of turnover once brokerage, STT, exchange charges, stamp duty, GST and realistic slippage are counted. A trader taking five trades a day is paying somewhere near 0.5–0.75% of turnover daily — which is why most intraday traders lose money even with a coin-flip win rate. Any record that does not show its cost assumption should be read as gross, and therefore as fiction.

How to verify all of this yourself

You should not take our summary at face value — that would defeat the point of running the challenge. Three independent checks:

1. Scroll the channel backwards. Every setup post sits above its result post, timestamped, all the way back to 12 July. Compare the count of setup posts against the count of results; they reconcile because they have to.
2. Check the live scoreboard. arthaalgo.in/?track is public, needs no login, and is generated from the same recorded trades — not from a spreadsheet we maintain by hand. It will keep updating after this article stops being current.
3. Read the methodology and disagree with it. The setup is a straightforward opening range breakout with a scored filter stack: volume confirmation, a body close beyond the range, prior-day level confluence, VWAP alignment, market regime, time of day, and a chase guard that refuses a trade already extended toward its target. If you think a filter is wrong, you can now argue about it against 30 days of published outcomes instead of against an opinion.

What we are doing with the result

The record stays up permanently, whatever it says. The methodology gets changed only on evidence, one variable at a time — we log the score contribution of every filter on every trade specifically so that a future decision to re-weight one is a measurement rather than a hunch, and the sample needed for that is measured in hundreds of trades, not dozens.

The thing worth copying here is not our filter stack. It is the format. If you keep a trading journal, publish the next twenty setups somewhere with a timestamp — a public channel, a group chat, anywhere the entry cannot be quietly revised — and grade them in R against a cost assumption. Most traders discover their remembered record and their actual record are different people's records. That discovery is worth more than any signal.