eBroker

Telling the desk what to do

For when you have a view of your own, and want it executed with the same discipline as everything else.

The engine decides on its own all day. This is the other direction: you decide, and the desk does the parts you should not have to do by hand — watching the price, sizing the position, and getting you out.

An instruction says one thing, in your numbers:

Buy ITC at or below ₹122. Commit ₹40,000. Stop at ₹114. Target ₹189. Expires 30 September.

What comes back is not a confirmation. It is a read-back — the whole instruction as the desk understood it, in its own words, with the quantity worked out and the consequences spelled out. You confirm that, or you go back and change it. Nothing is armed until you do.

---

Step 1You say it, in your own numbers

Four things make an instruction. You supply the first two; the desk proposes a stop and you may change it.

Step 2What it refuses, and what it only warns about

The difference matters. A refusal means the instruction cannot be written as given. A warning means it is your call and you should know what you are agreeing to.

Refused, with the reason on the field:

• No stop. Always. The engine would reject the order anyway, so accepting it here would only mean finding out later, from a rejection you cannot act on.
• A stop on the wrong side of your entry. On a buy the stop sits below the entry — if you meant to sell into strength, that is a target, not a stop, and the desk says exactly that rather than silently swapping them.
• A target on the wrong side. Same reasoning, other direction.
• Money that will not buy one share at the price you named.
• More than your envelope allows, counting what your other instructions have already committed.
• A trigger far from where the stock actually trades. Refused with a reason, never quietly adjusted to something more reasonable.

Allowed, but you are told:

• It will fire immediately. Your trigger is already met, so this is not a patient limit — it is a buy at market, now. Some people mean that. You should know which one you are doing.
• A tight stop on a long hold. Flagged because it defeats the thing you asked for: an 8% stop on a twelve-month view fires in week three.

Step 3You confirm the desk's words, not your own

This is the safety mechanism, and it is the oldest one in the business: a broker repeats the order back before placing it. You are shown the whole instruction — instrument, side, trigger, quantity, money, stop in both rupees and percent, target, expiry — and you confirm that.

The read-back also states plainly what will not happen, because those are the assumptions that cost people money:

• It will not average down if the price falls further.
• It will not re-enter after the stop takes you out.
• A target is not a guarantee. If the stock gaps straight through it, you get the price that exists, not the price you wrote.
• The stop is enforced by us on every cycle. It is not a resting order at the exchange, and if the desk is down it is not being watched.

Step 4It waits, then it acts

Your instruction sits armed. Every scan, the desk checks the live price against your trigger. When it is met, the order goes in at a price that will actually fill, and from that moment the position is managed exactly like every other position on the desk — the same stop machinery, the same trailing, the same exit discipline.

If nothing happens before the expiry you set, the instruction lapses. It does not sit there for a year waiting to surprise you.

---

Every instruction carries a stop. There is no setting for this

You may move your stop. You may make it wide. You may not remove it, and the screen says so rather than silently refusing.

A position without a stop is undefined risk, and undefined risk is the one thing this desk does not carry — not for the engine's trades and not for yours. The risk gateway would reject it even if the screen let you through, which is the correct number of places for a rule like that to live.

A long hold gets a wide stop, deliberately. If you tell the desk you are holding something for a year, it proposes a disaster stop rather than a trading stop. A tight stop on a twelve-month view fires in week three and defeats the thing you asked for. The read-back shows it in rupees and as a percentage, so “eight percent” is never an abstraction.

What your instruction is still subject to

Your word goes inside your own envelope. It does not go around the account's protection, and that is not a limitation to apologise for — it is what makes the feature safe to have.

• Your self-directed envelope. A capital limit for instructions specifically, separate from what the automated sleeves may use. Outside it, the instruction is refused with the reason.
• The single-name limit. Concentrating the account in one stock is capped, and an instruction is exactly the case that limit was written for.
• The account-wide deployment cap and the daily loss limit. If the desk has stopped for the day, your instruction does not restart it.
• Whether the account can get out again. Checked when you submit, not when the stop fires. On an account that cannot sell delivery stock through the API, a buy instruction would create a position whose stop can never execute — so it is refused at the door, with that explanation, rather than accepted into a trap.

Your own controls, on your own desk

The same panel that shows your positions lets you act on them. You can stop new entries across every sleeve, resume them, tighten your daily loss limit, and change how your capital is split between segments. Those are yours because it is your money — they are not behind a request to us, and they do not wait for anyone to be awake.

Two things are deliberately not on that panel. Nothing there can stop your exits. Pausing, halting and offboarding all leave every open position's stop-loss running on every cycle — the control simply does not exist, rather than existing and being discouraged. And the loss limit moves one way: you may tighten it below what was agreed, never loosen it past.

Each of these is a control on your desk. There is no shared button anywhere on this site that could act on somebody else's.

Getting out

Your exits are the ones you set: the stop, the target, and the date you said to hold until. All three are enforced by the desk on every cycle, using the same machinery the automated sleeves have run on for months.

Closing a position on demand — a “sell it now” button — is not built yet. When it is, it will apply only to positions your own instructions created. Positions the engine opened will keep their own exit discipline and will not be closeable that way, because one impatient afternoon should not be able to unwind the exact positions the discipline exists to protect.

---

Honest limits

Read next: how the engine decides on its own, or how to get a desk connected. The full risk position is set out in the risk disclosure.