"Just close my positions when I hit my loss limit. I know I won't do it myself."

It is an honest request. The person asking has watched themselves blow through a limit they set that morning, and they have concluded, correctly, that they cannot be trusted at 1:20 PM with a red P&L.

We are not going to build it. Not in v1, not later, not as an opt-in toggle. LossGuardian is read-only: it cannot place, modify or cancel an order on your account, and the API permissions it holds do not allow it to. That is a design decision, not a limitation we are working around.

Here is the reasoning, because "we just won't" is not an argument.

Auto-squareoff removes the rep you need

Discipline is not a personality trait you either have or do not have. It is the accumulated result of a specific event repeating: you are shown that you have crossed your own line, you feel the pull to keep going, and you close the position anyway.

That last clause is the whole thing. It is the rep.

If software closes the position, the event still happens. You still crossed the line, you still felt the pull. But the rep does not. You did not do the hard part. Do that a hundred times and you have not built a hundred reps of discipline. You have built a hundred reps of being rescued, which is a different skill, and one that only works while the rescuer is present.

Now take the app away. Your laptop is at the service centre; you are trading from a phone; the API token expired at 6 AM and you did not notice. Every trader who outsourced the decision is, that day, exactly the trader they were before they installed anything.

The goal is not to have a good day while the software is running. The goal is to become someone whose default behaviour at 1:20 PM is different. Those are not the same product.

The dependency is quiet and it compounds

Watch what happens to your limits once something else enforces them.

When you are the one who has to close the position, the limit is a real constraint and you set it like one. When software will close it for you, the limit becomes a setting. Settings get adjusted. The number drifts up, ₹5,000 becoming ₹8,000, because "the app will catch me anyway, I just want room to breathe." The limit stops being the point at which you stop and becomes the point at which you are stopped, which are psychologically opposite things.

Six weeks in, the trader with auto-squareoff has a wider limit than the one they started with and no independent ability to honour it. That is a worse position than the one they were in before, and it took a feature request to get there.

Software should not have that power over your account

Set the psychology aside for a moment and look at it as an engineering problem.

To close a position, an application needs order-placement permission on your broker account. That permission does not come with a qualifier. It cannot be granted as "may exit, may not enter": the broker's API does not model intent, it models capability. Whatever can square off your NIFTY position at 1:20 PM can also, if it is wrong, place an order you did not want at 9:20 AM.

So ask what has to hold for that to be safe:

• The P&L calculation has to be right, on every product type, including the ones where the broker's own numbers lag by a few seconds.
• The connection has to be healthy. A stale feed showing a loss that is not there is now an order, not a false alarm.
• The logic has to be right on the day the market gaps, when the exchange feed hiccups, and when your broker's API returns a shape nobody tested against.
• Our deploy has to be right. Every time.

Software gets things wrong. Ours will get things wrong. The difference between a read-only tool and one holding trade permissions is the difference between a bug that shows you a wrong number and a bug that takes a position off at the worst possible price and posts the loss to your account. One of those you notice and report. The other one you cannot undo.

We would rather be a tool that occasionally shows a wrong number than a tool that occasionally trades.

There is a security argument on top of the safety one. A read-only token is worth very little to anyone who steals it. A token that can trade is worth taking. The smallest permission that does the job is the one worth holding, and the job here does not require trading. You can read exactly what LossGuardian stores and where on the security page.

What we do instead

We make the moment impossible to miss, and impossible to pretend you did not see.

When you cross a limit you set, LossGuardian puts a hard-stop alert in front of you that you cannot wave away with a click. It states the number, it states the rule you wrote, and it stays. It is a warning designed to be inconvenient, because a warning you can dismiss in half a second is a warning you have already learned to dismiss.

It also catches the thing a P&L number alone will not. Revenge trading does not announce itself as a big loss. It announces itself as a sequence: a loss, then a re-entry ninety seconds later, then position sizes creeping up. By the time the rupee figure looks alarming, the tilt started twenty minutes earlier. Timing and frequency are the signal; the loss is the lagging indicator.

Then you close the trade. On your broker's own platform, with your own hands. The rep counts.

"But I really cannot stop myself"

Then the limit is not the problem, and neither is the software.

If you genuinely cannot close a position after being told, in unmissable terms, that you are past your own line, no amount of automation fixes that. It just moves where the failure shows up. The trader who cannot honour a limit with a tool enforcing it is the trader who widens the limit, trades a second account, or turns the tool off. All three happen, and all three are worse than an honest failure you can see.

The honest version is useful. It gives you something to work with: a day you can look at, a pattern you can name, a number you can bring down next week. Being quietly saved teaches you nothing, and the record looks fine right up until the day it does not.

The trade-off, stated plainly

You can have a tool that protects today's capital by taking the wheel, or a tool that builds the trader who protects every future day's capital. Over one session the first one looks better. Over a year it is not close.

We picked the second one. If you want the first, we are the wrong product, and we would rather say so than sell you a dependency.

The rules are yours. We make sure you cannot pretend you did not see them.

Related reading: how LossGuardian compares to other Indian risk tools on permissions, pricing and what they are allowed to do: LossGuardian vs other tools. If you trade on Zerodha, the connection is an official broker-issued integration: LossGuardian for Zerodha.

This is not bad luck clustering. It is a sequence, and it runs in roughly the same order every time.

The sequence

9:15 to 10:30. The plan is intact. You are rested, the plan is fresh, and you have not lost anything yet. Almost nobody revenge trades in the first hour. This is worth noticing, because it tells you that discipline is not a fixed quantity you either possess or do not. At 9:20 AM you have plenty of it.

The first real loss. Not a scratch, but a loss with a number attached that you did not want to see. The immediate feeling is not "I was wrong." It is urgency. Something needs to be done about this, and it needs to be done now.

The re-entry. This is the hinge. The trade you take within a few minutes of a loss is almost never the trade your plan describes. It is the same underlying, often the same direction, taken because you want the last one back. Ask a trader afterwards what the setup was and they will describe the previous trade, not this one.

The size creep. The next lot is bigger. It has to be, because the hole is bigger, and a normal-sized win no longer gets you back to flat. This is the point at which the day changes character. Up to here you were losing money. From here you are trying to win money back, which is a different game with a different risk profile, and one you never planned for.

12:30 to 2:00. The drift. Lunchtime volumes thin out, moves get choppy, and setups that would have been obvious at 10 AM are ambiguous now. Meanwhile you have been staring at a screen for four hours and your capacity for a hard "no" is at its lowest point of the day. Bad conditions meeting a depleted decider.

The last hour. Either you got it back, which teaches you the wrong lesson and guarantees a repeat, or you did not, and the day ends at a number that is a multiple of the loss you were actually upset about.

Notice what the sequence does not require: a big loss. The starting loss is usually ordinary. What makes the day expensive is everything that came after it.

Why the afternoon specifically

Three things stack up, and only one of them is about the market.

The gap between decision and consequence has closed. In the morning, a trading rule is abstract: a sentence you wrote about a hypothetical loss. By 1 PM the loss is real, it is on your screen with a minus sign, and the rule is now asking you to accept it permanently. Those are very different asks, and the person being asked at 1 PM is not the person who agreed at 9.

You have spent the day making decisions. Every skipped setup, every hold, every exit is a small act of self-control, and by the afternoon you have made hundreds. The capacity to refuse yourself something is finite over a session, and the afternoon is where it runs out. Nobody breaks their rules while fresh.

Loss hurts more than the equivalent gain feels good. This is the oldest finding in behavioural economics and it does not care that you know about it. A ₹10,000 loss creates more pressure to act than a ₹10,000 gain creates pressure to stop. So the losing day generates urgency and the winning day does not, which is exactly backwards from what would be useful.

Why a rupee limit alone does not catch it

Most traders' only guardrail is a daily loss number. It is necessary, and it is not sufficient, for one structural reason: a rupee limit is a lagging indicator.

Look at the sequence again. The tilt began at the re-entry: the trade taken ninety seconds after a loss, for the wrong reason. At that moment your P&L was still well inside your limit. Nothing fired. The limit only speaks up after the size creep has already done most of the damage, which is to say, after the decision that mattered was already made.

The signal that is actually early is not the money. It is the timing:

• Time between trades. A re-entry seconds after a loss is a different animal from one taken after ten minutes and a fresh look at the chart. The gap is measurable, and it collapses when you are on tilt.
• Trade frequency versus your own baseline. Fourteen trades on a day you normally take five is information, whatever the P&L says.
• Position size relative to your average. Size creep is arithmetic. It shows up in the data long before it shows up in the damage.
• Direction of the sequence. Loss, re-entry, loss, bigger re-entry is a pattern with a name, and it can be recognised while it is running.

Every one of those is visible in your order history in real time. That is the whole design idea behind LossGuardian: watch the behaviour, not just the balance, because the behaviour is what is early.

What to actually do about it

Write a cooldown rule and make it the primary rule. Not "I will not lose more than X"; that is your backstop. The front-line rule is a time rule: after a losing trade, no new position for N minutes. Ten is a reasonable starting point. The cooldown is aimed precisely at the re-entry, which is the hinge of the whole sequence, and it is far easier to honour than an open-ended instruction to calm down.

Cap the number of trades, not just the rupees. A hard count, "six trades, then I am done for the day", is unambiguous in a way that a rupee figure is not. You cannot argue yourself past a count.

Decide about the afternoon in the morning. If you have a loss on the board at 1 PM, the decision about whether to keep trading should already have been made, at 9, by someone who was not upset. "If I am down at 1 PM, I stop" is a rule the morning version of you will happily sign and the afternoon version cannot renegotiate.

Make the alert impossible to dismiss. A notification you can click away in half a second will be clicked away in half a second, because you have trained yourself on years of banner ads to do exactly that. The alert has to cost you something to get past, and it has to state the rule in your own words so that continuing is an explicit act rather than a reflex.

Review the sequence, not the total. At the end of a bad day, the number tells you almost nothing. The trade list tells you everything: where the first real loss was, how many seconds until the next entry, when size changed. That is the autopsy worth doing, and it is why LossGuardian keeps a trade-by-trade drill-down rather than just a daily figure.

The part that does not automate

You can be warned. You can be shown the sequence as it forms, with the clock and the sizes and your own rule quoted back at you. What you cannot outsource is the next thirty seconds, where you close the position and stand up.

That is deliberate on our part: we will never square off a position for you, because the trader who is rescued at 1:30 PM never becomes the trader who does not need rescuing. But being warned, clearly, at the re-entry rather than at the wreckage, changes the odds a great deal. Most people do not fail the test at 1:30. They never get told they are taking it.

Next: the practical build, how to set a daily loss limit you cannot talk yourself out of. If you trade options on Dhan, the integration is broker-issued and read-only: LossGuardian for Dhan. And here is how we compare with other risk tools.

The failure is rarely willpower in the way people assume. It is that the limit was never built to survive contact with a losing afternoon. It was a round number picked in a good mood, stored in the trader's head, with no definition of what counts and no consequence for passing it. Of course it broke.

A limit that holds has four properties. It is calculated, it is written, it is unambiguous, and it is placed where you cannot avoid it. Here is how to build one.

1. Get the number from your account, not from your mood

The number people usually pick is whatever loss feels tolerable. That is the wrong input, because your tolerance moves: it is generous on a Monday and nonexistent after three red days.

Start from arithmetic instead.

Step one: decide what a bad month is allowed to cost. Take your trading capital and pick the drawdown at which you would genuinely stop and reassess. Not the figure that wipes you out; the figure at which you have a problem. For most people that is somewhere between 6% and 10% of capital. On ₹5,00,000, call it 8%, or ₹40,000.

Step two: divide across the month's losing days. In a month of roughly 20 sessions, plan for six to eight losers. That is not pessimism, it is a normal distribution of outcomes. ₹40,000 across seven bad days is a daily limit of about ₹5,700. Call it ₹5,000, because round numbers are easier to hold in your head under stress.

Step three: sanity-check it against your average win. If your daily limit is four times your typical good day, it is not a limit, it is decoration: one bad day would erase four good ones and you would still be inside your rule. A daily loss limit ought to sit at or below one to one and a half times your average winning day. If it does not, the limit is too wide, or the strategy has a risk-reward problem the limit is hiding.

That is the number. Notice what it did not come from: how confident you feel, what you lost yesterday, or what someone on YouTube trades.

If you would rather not do the arithmetic by hand, the daily loss limit calculator runs these three steps on your own figures and writes the rule out for you. It works entirely in your browser, and nothing you type is sent anywhere.

2. Write it down as a sentence, not a figure

"₹5,000" is not a rule. It is a quantity, and a quantity can be reinterpreted at 1:30 PM by a motivated person. A rule is a sentence with no room in it.

Write this, in your own words, somewhere permanent:

If my realised plus unrealised loss for the day reaches ₹5,000, I close everything and stop trading for the day. Not a smaller size. Not one more setup. Stop.

Then close the loopholes explicitly, because every one of these is a real argument a real trader has made with themselves:

• Realised or unrealised? Both. Counting only booked losses means holding a losing position is how you stay under your limit, which is precisely the wrong incentive.
• Per account or across accounts? Across. If you trade two brokers, the limit is the sum. A limit you can evade by opening a second app is not a limit. This is the single most common way a rule quietly dies, and it is exactly why a risk tool has to see every broker you use rather than one at a time.
• What about a hedge? It counts. Every open leg counts.
• Does a green afternoon reset it? No. Once you have stopped for the day, you have stopped, even if the market hands you an obvious setup at 2 PM. Especially then.
• Can it be raised mid-session? Never. It can be reviewed on a weekend, with a month of data in front of you, and changed for next week. Not today.

That last one is the load-bearing clause. Most limits do not get broken; they get amended, in the moment, by the person least qualified to amend them.

3. Add the two limits that fire earlier

A rupee limit is a backstop, and it is a late one: by the time you reach it, the sequence that got you there has been running for an hour. Pair it with two rules that trip before the money does:

A trade count. "Six trades, then I am done." A count is beautifully unarguable. You cannot be halfway past it, cannot round it down, cannot decide that this one does not really count.

A cooldown after a loss. "No new position for ten minutes after a losing trade." This one is aimed at the exact hinge of a bad day: the re-entry taken ninety seconds after a loss, for the wrong reason. Ten minutes is usually enough for the urgency to fade to the point where you can look at a chart rather than a grievance. There is more on why the afternoon breaks people in why discipline fails at 1:30 PM.

Together these three cover different failure modes: the count catches overtrading, the cooldown catches tilt, and the rupee limit catches the day that goes wrong quickly.

4. Put it somewhere you cannot avoid meeting it

Here is where most of this advice usually stops, and where it usually fails.

A limit written in a journal is checked when you feel like checking it, which is never the moment it matters. A limit in your head is renegotiated silently. A mental note plus a broker P&L screen means you have to be the one who notices, and at 1:30 PM on a losing day, you are the least reliable observer available.

The limit has to be enforced by something that is not you, at the moment you cross it, in a form you cannot dismiss with a reflex click. That is the entire job LossGuardian does: it watches your positions across your brokers, and when you cross a line you wrote, it puts a hard-stop alert in front of you that states your own rule back at you and does not go away.

What it will not do is close the position. It is read-only: it cannot place, modify or cancel an order, by design, and that is a decision we are not going to reverse. The closing is yours. That is what makes the rep count.

It works the same way on every broker we support, and the price does not change between them: Zerodha, Fyers, Dhan, Upstox, Angel One and Groww. If you trade two of them, both are counted against the same limit, which is the point.

5. Review it monthly, with data

At the end of each month, sit down with the record and ask three questions:

1. How many days did I hit the limit? More than three or four out of twenty and the number is too tight for your strategy. You are being stopped out of normal variance, which is its own kind of damage.
2. How many days did I pass it? Any answer other than zero is the thing to fix, and it is a process problem, not a number problem. Do not widen the limit to make the breaches disappear.
3. What did the breach days look like? Not the total; the sequence. Where was the first loss, how long until the next entry, when did size change. The pattern is almost always the same one, and once you can name yours you can watch for it.

Then adjust once, deliberately, for next month. On a weekend, with the data in front of you, by the version of you that is not down ₹5,000.

That is the whole method. The number is arithmetic, the rule is a sentence with the loopholes closed, the early warnings are time and count rather than money, and the enforcement lives outside your head. None of it is complicated. It is just built to be difficult to argue with, which is the only property that matters at 1:30 PM.

See also: how LossGuardian compares with other Indian risk management tools, and what it stores and where on the security page.