Why Your Three P&L Numbers Never Match

Three documents that were never trying to agree

The mistake that starts the confusion is treating the three figures as three attempts at the same measurement. They are not. Each document was designed for a purpose, and each purpose forces a different definition of what counts, when it counts, and what gets subtracted first. Once you can state the three questions, the disagreement stops looking like a defect and starts looking like information.

The platform figure answers an economic question: what are my positions worth right now compared with what I paid for them. It is a mark to market number. It includes positions you have not closed, because their current value is exactly what the question is about. It is usually measured from the average cost the platform holds for each position, which is a lifetime cost rather than a period one, and it is typically shown gross of most of the charge stack, because the display is about market movement rather than about cost.

The ledger figure answers a cash question: what actually moved through the account. It is a settled cash record, and cash moves only on settlement, which is a day or more after the trade. It is scrupulous about charges, because every single one of them was debited, and it is completely blind to open positions, because an open position has moved no cash. It also contains a great deal that has nothing to do with trading at all: your own fund transfers, maintenance fees, pledge charges, dividends arriving from companies.

The taxable computation answers a statutory question: what income arose in this period, under which head, after which deductions. It counts realised transfers, not marks. It runs over a fixed period that you cannot choose. It refuses some costs that the ledger cheerfully debited, and it splits the result into heads that are taxed under different rules and whose losses cannot be pooled. It is also, strictly speaking, not one number at all, which is a point worth holding on to.

Notice the last cell. The third figure is a set of separate computations that happen to belong to the same person. Capital gains sit in one place, speculative business income in another, ordinary business income in a third, and dividends in a fourth. They are taxed under different rules and their losses are ring fenced from each other. Adding them together, as this page does in order to close the reconciliation, is a presentational device and nothing more. No return treats them as one figure.

The constructed year, stated in full so it can be checked

The trader below does not exist. The year is built rather than observed, deliberately, because a real account would either be missing several of the effects worth demonstrating or would contain identifying detail that has no place on an educational page. Every figure that follows is derived from this single construction, and the construction is stated here in enough detail to be reproduced or disputed.

The period is 1 April 2025 to 31 March 2026. The trader runs three activities side by side, which is entirely ordinary and is also the root of most reconciliation trouble. There is an equity delivery book of eight holdings, five of which were closed during the year and three of which were still open at the period end. There is an equity intraday book of 64 round trips, with a combined traded value of 1,82,06,379 rupees. And there is a derivatives book on an index, 18 option positions and 4 futures positions closed during the year, with one further futures position of 75 units left open across the year end at an entry of 23,640 and a closing mark of 24,508.

Two of the delivery lots were bought before the year began, so their cost basis is carried forward from an earlier period. One lot was sold on 31 March 2026, the final trading day, which puts its settlement in the following period. Its mirror image also exists: a sale made on 28 March of the previous year, whose cash arrived on 1 April 2025 and therefore lands in this year's ledger. Both boundaries are present on purpose, because a reconciliation that walks only one end of the period will double count or lose a trade and will not be able to tell which.

Two holdings paid dividends during the year, 2,850 rupees on one and 12,400 rupees on the other. The charge model applies the statutory rates that were in force during the period, together with an illustrative allowance for brokerage and for the derivative exchange charges, and the full schedule with its sources is set out in our companion piece on the real cost of an Indian trade, which this page deliberately does not restate. Total charges across every trade in the year come to 13,992 rupees, of which securities transaction tax is 7,679, brokerage is 3,118 and goods and services tax is 859. Account level charges, meaning demat maintenance and pledge requests, add a further 566 rupees that no trade level profit and loss figure will ever contain.

The three totals, and the bridge between them

Computed on this construction, the three views produce three different answers. The platform figure is 5,69,241 rupees. The ledger figure is 4,65,349 rupees. The taxable computation, with all heads added together for reconciliation purposes only, is 4,12,087 rupees. The gap between the first and the second is 1,03,892 rupees, and a further 53,262 rupees separates the second from the third.

What follows is the whole bridge. It is the centre of this page, and it is worth reading as a list of causes rather than as arithmetic, because the arithmetic is trivial and the causes are the part that transfers to your own records.

Walking the reconciliation, cause by cause

Unrealised positions are the largest single cause, and they are split in two here for a reason. The three open delivery lots carry 95,150 rupees of unrealised gain in total. The platform counts all of it. The ledger counts none of it, because no cash has moved. But the reconciliation separates 50,400 rupees that accrued during the year from 44,750 rupees that accrued before 1 April and is merely being displayed again. This second amount is the quieter problem. Even a trader who correctly excludes unrealised gains from a cash figure will often assume that whatever unrealised movement the app shows belongs to the period on screen. It does not. The app is showing you the distance from your original purchase price, however many years ago that was.

The settlement boundary contributes twice, once at each end. The 31 March sale of 800 units for 1,89,520 rupees produced 38,720 rupees of gain that the trading statement records and the ledger does not, because the money arrived in April. The prior year sale contributes 30,590 rupees in the opposite direction: cash in this ledger, trade in last year's records. These two items are the classic source of a trader concluding that the ledger is missing a trade. It is not missing anything; it is recording a different event.

Charges are the cause everyone can name and almost nobody quantifies. The 13,992 rupees of trade charges and 566 rupees of account level charges are visible in the ledger and largely absent from the display. Against a platform figure of 5,69,241 rupees they look small, and in aggregate they are, but they are not distributed evenly. On the intraday book they are decisive, and the section below shows why.

Dividends enter the ledger and leave the trading result entirely. The two holdings paid 15,250 rupees between them. 1,240 rupees was deducted before payment, so 14,010 rupees reached the account. The ledger therefore shows the net amount, while the computation counts the gross amount as income with credit for the tax already deducted. Only one of the two payments attracted a deduction, because the relief is tested separately for each paying company, and one payment fell below the relevant amount while the other did not. Under the Income-tax Act, 2025, the deduction obligation on dividends sits at section 393(1), and the relief from deduction is cumulative rather than a simple threshold: it requires an individual shareholder, payment by a mode other than cash, and an aggregate for the tax year that does not exceed ten thousand rupees. Confirm the current position before relying on it.

The open futures position is the most interesting item in the entire bridge. Futures settle their mark to market in cash every day. The 65,100 rupees of movement on the open position has therefore already passed through the ledger even though the position has not been closed. The cash is real and settled; the position is not. This single item accounts for a large part of the distance between the ledger figure and the taxable one, and it is the item most likely to be handled inconsistently, because the trader can see the cash and reasonably assumes that cash means realised.

What happens to that item in a tax computation is not a question this page can answer for you, and it is worth being precise about why. For an ordinary investor, a gain arises on transfer, so an open position produces nothing to tax. Where securities are held as stock in trade of a business, a different discipline applies. The Income-tax Act, 2025 provides at section 276 that business income is computed on the cash or mercantile system regularly employed, subject to income computation and disclosure standards, and at section 277 that securities held as inventory are valued under those standards. Section 32(1)(h) allows a marked to market loss only as those standards compute it, and a companion provision disallows such losses otherwise. The shape of the rule is therefore asymmetric: it can recognise an unrealised loss and it does not recognise an unrealised gain. Which of these applies to your activity depends on classification, and that is a question for a chartered accountant rather than a page.

Two costs move between the ledger and the computation. Securities transaction tax of 7,679 rupees was debited in full, but it is not allowable in computing capital gains, so the portion attributable to the delivery sales is added back in the bridge. It is allowable where the income is business income. Under the Income-tax Act, 2025 the disallowance for capital gains sits at section 72(3) and the allowance for business income at section 32(1)(k). The corresponding provisions existed under the Income-tax Act, 1961 for the year being reconciled. The practical consequence is that the same rupee of tax is deductible in one part of your own records and not in another, which is precisely the kind of asymmetry a single blended profit figure destroys.

The boundary effect, and the rule that survives not knowing the answer

The period edge deserves its own treatment because it is where an honest reconciliation can still go wrong. The mechanics are not in dispute. A trade is executed on one day and the cash settles on a later one, so the trading statement and the funds ledger will always disagree about any trade in flight across the boundary. That disagreement is not an error, and its size is exactly the value of those trades. The settlement cycle itself, including what a holiday does to it, is covered in detail in our guide to T plus one settlement, and there is no need to repeat it here.

What is genuinely open is which of the two dates the tax computation should follow for an exchange traded transfer. The commonly cited position is that the date of the broker's note governs, and there is a long standing departmental circular usually quoted for it. This page will not state that position as verified, because the source could not be read at first hand while this was written, and a citation that cannot be opened is not a citation. If you need the answer for a filing, it is a question for a chartered accountant, and it is one worth asking explicitly rather than assuming.

Here is the useful part: you do not need the answer in order to reconcile correctly. What you need is consistency across both edges of the period. Whichever basis you adopt, apply it to the opening boundary and the closing boundary in the same way. If you count the 31 March sale as belonging to this year, you must also exclude last year's 28 March sale, even though its cash is sitting in this year's ledger looking like income. The failure mode is not choosing the wrong basis. The failure mode is choosing one basis at the far end of the period and the other at the near end, which counts one trade twice and drops another entirely, and produces a discrepancy that no amount of re checking the arithmetic will explain.

The reconciliation on this page uses the trade date, which is stated in the method note as an assumption rather than a conclusion. On the constructed year the two boundary items are 38,325 rupees and 30,131 rupees net of charges, moving in opposite directions. If you adopted the settlement basis instead, both would flip, and the taxable figure would change by the difference between them. That is a large enough amount to be worth deciding deliberately.

The gap does not come from everywhere

Aggregate reconciliations hide where the trouble actually lives. Splitting the same bridge by segment produces a result that is more useful than the total, and slightly surprising.

The equity delivery segment shows a platform figure of 1,81,095 rupees against a ledger figure of 74,989 rupees, a gap of 1,06,106 rupees. That is where nearly all the trouble is, and the reason is structural: delivery positions sit open across the boundary, they carry a cost basis that may be years old, and their settlement lags the trade.

The derivatives segment shows a gap of only 4,999 rupees on a platform figure of 3,84,610 rupees. That gap is the charge stack, exactly, and nothing else. The reason is that futures settle their mark in cash every day and options are realised on square off, so there is very little for the two documents to disagree about. This is the opposite of what most traders expect, because derivatives feel like the complicated segment. For reconciliation purposes they are the easy one. What derivatives do produce is a tax boundary problem rather than a cash boundary problem, which is the open position item discussed above.

The intraday segment is small in absolute terms and instructive out of all proportion. The platform shows the book slightly ahead, at 3,536 rupees of gross gain across 64 round trips. The ledger shows it behind, at a net loss of 2,695 rupees. The entire difference is 6,231 rupees of charges, which exceeded the gross result. A trader reading only the display would conclude the intraday book was marginally profitable and worth continuing. The ledger says the opposite, and the ledger is the one that ties to the bank. If you take one operational habit from this page, take the habit of reading the intraday book from the ledger, because it is the activity where the charge stack is most likely to be the whole story. The mechanics of keeping trade level records that make this visible are covered in our trade journal standard and grader.

The turnover question, where this gets expensive

There is a fourth number, and it causes more unnecessary alarm than the other three combined. Turnover, in the sense that matters for whether an audit requirement is triggered, is not the value of what you traded. It is a purpose built figure, and the distance between it and the notional value of your contracts is not a detail. It is two orders of magnitude.

On the constructed derivatives book the notional value of the contracts is 14,65,00,320 rupees. This is the number a trader arrives at by multiplying lot size by index level by number of positions, and it is the number that produces a sleepless night, because it sounds enormous and it is. It is also irrelevant to the test. A single index option lot carries a notional running into lakhs, and a moderately active year will therefore run into crores of notional on capital that never exceeded a few lakh rupees.

The figure the test actually reads is built from the differences on squared off transactions, taking each profit and each loss as a positive amount and adding them up. On this book that comes to 4,87,865 rupees. A trade that made ten thousand contributes ten thousand, and a trade that lost eight thousand also contributes eight thousand. It is a measure of how much you churned, not of how much you controlled, and it sits far below the notional for exactly that reason. The professional guidance that governs this computation is the accounting institute's guidance note on tax audit, and it makes a point worth repeating: the method exists solely to determine whether the audit requirement applies, and it does not change how the income itself is classified or taxed.

There is a second component that is widely misreported, including in guides that are otherwise careful. The current edition of that guidance also brings option premium received on sales into turnover, with a rule against double counting where the premium has already been captured in the net profit figure. On this book the premium received on option sales is 8,62,935 rupees, so a premium inclusive reading gives 13,50,801 rupees against 4,87,865 rupees on differences alone, a factor of 2.8 on the same trading. That is a large enough divergence that the method has to be settled before the figure is used for anything, and it is not a question you should answer from a blog post, this one included. Establish which reading your accounts support, and be able to show how the profit figure was built.

As to the threshold itself: the audit requirement under the Income-tax Act, 2025 sits at section 63, and the table there sets a business turnover figure of one crore rupees, substituted by ten crore rupees where cash receipts do not exceed five percent of receipts and cash payments do not exceed five percent of payments. The equivalent provision under the Income-tax Act, 1961, which governs the year being reconciled here, was section 44AB with the same two figures and the same twin five percent tests. Trading settles almost entirely through the banking system, so the cash tests are usually satisfied comfortably. Thresholds change with almost every Finance Act, and this page states them as read from the gazetted text rather than as advice: verify the current position at source before you rely on it. The presumptive scheme is a separate doorway with its own ceilings and its own continuity conditions, and the deeper treatment of both, along with the audit machinery, belongs to our guide on how futures and options are taxed in India.

The practical point for a reconciliation is narrower than any of this. Turnover is not a fourth answer to the question of how you did. It is a size measurement used for one specific administrative test, and it will never reconcile to your profit and loss figure because it is not trying to. Putting it in the same column as your result is a category error, and it is the reason so many traders believe their records are inconsistent when they are simply mixing units.

Speculative and non speculative, and why the split reaches the reconciliation

The last structural reason the third figure refuses to be one number is that the statute sorts the same trader's activity into boxes that never mix. Equity intraday, bought and sold within the session without delivery, is speculative business. Exchange traded derivatives, economically much closer to intraday than to investing, are carved out of the speculative definition and taxed as ordinary business income. Equity delivery, where shares are actually received, is usually capital gains. One trader, three regimes, running at the same time on the same screen.

The carve out is worth understanding precisely, because its conditions bear directly on your documents. Under the Income-tax Act, 2025, the definition of a speculative transaction at section 2(31) excludes a specified derivative transaction, and section 2(33) defines that term with conditions attached: the transaction must be carried out through a registered broker or intermediary, electronically on the screen based system of a recognised stock exchange, and supported by a time stamped contract note carrying the unique client identity number and the permanent account number. Under the Income-tax Act, 1961, which governs the constructed year, the same carve out lived in proviso (d) to section 43(5), covering an eligible transaction in derivatives carried out on a recognised stock exchange.

Read that list of conditions again as a records requirement rather than as law. The contract note is not paperwork you keep in case someone asks. It is the document that evidences the character of the income. This is why the reconciliation procedure below puts contract notes among the four documents you pull rather than treating them as an optional cross check, and it is why a reconciliation built only from a downloaded profit and loss statement is structurally incomplete.

On the constructed year the split matters in a way the blended figure conceals. The delivery book produced 70,887 rupees of long term capital gain and 14,700 rupees of short term. The intraday book produced a speculative loss of 2,695 rupees. The derivatives book produced 3,13,945 rupees of non speculative business income. Dividends contributed 15,250 rupees under a fourth head entirely. Anyone adding those into a single figure has produced a number that is arithmetically correct and structurally meaningless, because the speculative loss cannot be set against the derivatives profit. Losses of the two kinds are ring fenced from each other, they carry forward for different numbers of years, and the right to carry them forward at all depends on filing within the due date. The set off and carry forward rules are the subject of their own treatment in our guides to capital gains for the active trader and to futures and options taxation, and this page defers to them rather than restating them.

The procedure, in the order that works

Everything above is diagnosis. This is the part to keep. The procedure is deliberately mechanical, and its value comes almost entirely from the order, because each step can only be checked once the one before it has been settled.

Fix the period and the basis before you export anything. One April to thirty first March, and one decision about whether you are reading trade dates or settlement dates. Write the decision down. Every platform lets you choose a window, and the default is almost never the tax year, so an export taken casually will already be measuring something else before you have started.

Pull all four documents, not one. Most reconciliations fail here, because the profit and loss statement is easiest to download and appears to contain everything. It does not contain the cash, so it cannot show a timing difference. Timing differences are visible only where two records disagree.

Tie the ledger to the bank before you look at a single trade. Opening balance plus credits minus debits must equal the closing balance, and the fund transfers in the ledger must match your bank statement. This step catches nothing about trading and everything about whether the ledger you are working from is complete. Skipping it means every later difference is ambiguous, because you cannot tell a missing trade from a missing page.

Strip out what is not trading. Your own transfers, interest, dividends, maintenance charges, pledge fees. None belongs in a trading result, several belong in the tax computation under other heads, and all of them will otherwise sit in the difference column looking like an unexplained gap.

List every open position and the cost the platform is holding for it. This is the step that produces the largest single reconciling item, and it is also the one where you discover whether the platform's average cost matches your own record, which corporate actions and partial sales routinely disturb.

Walk both boundaries, on the same basis. Trades at the start of the period and trades at the end. This is the discipline point from the boundary section, and it is worth a line in your working papers rather than a mental note.

Split by head, then stop. Delivery, intraday, derivatives, and anything under other heads. Three heads with three sets of rules. The reconciliation ends here, and what happens next is a return, which is not the same exercise and not one to do from a web page.

The three checks that catch most discrepancies are worth stating separately, because they are quick and they are diagnostic. Does the ledger close to the bank. Does every contract note appear in the ledger. Does each open position appear in exactly one place and no other. If all three pass and a difference remains, the difference is almost certainly one of the named items on this page, and you can go looking for it by name rather than by re adding columns.

What a genuine error looks like

If most mismatches are not errors, the fair question is how you recognise one when it happens. Expected differences and genuine errors have different shapes, and the difference is easy to test.

An expected difference has three properties. It has a name, meaning you can say which of the twelve causes it is. It has a direction that follows from that cause, so an unrealised gain always makes the platform figure larger and never smaller. And it has a matching entry somewhere else, so a trade absent from this period's ledger appears in the next period's, and a dividend that arrived net has a deduction certificate behind the missing part. If you can supply all three, you have found a difference and you are finished with it.

A genuine error supplies none of the three. The characteristic shapes are worth memorising. A contract note with no corresponding ledger entry, which means a trade was executed and the cash never moved through the account you are looking at. A debit in the ledger with no document behind it, which is either a charge you did not know about or something that needs an explanation from the intermediary. A quantity in the holdings statement that does not match the demat account, which is a settlement or corporate action problem and not an accounting one. A charge that is a different order of magnitude from comparable trades in the same segment. And an opening balance that does not equal the previous period's closing balance, which invalidates everything downstream of it and should stop the exercise until it is resolved.

There is a fourth category that sits between the two, and it is the most common source of a real filing problem: an item handled inconsistently. Two boundary trades treated on different bases. An account level charge deducted in one head and ignored in another. A dividend counted net in one place and gross in another. Each of these is individually defensible and jointly wrong, and none of them shows up as an obviously broken number. The defence is the same as everywhere else in the exercise, which is to write down the basis before you start and then apply it without exception, including in the places where applying it is inconvenient.

The section numbers on your saved checklist have changed

One further point belongs on a page about reconciliation, because reconciliation is a documents and references exercise before it is an arithmetic one. The Income-tax Act, 2025 received assent on 21 August 2025 and came into force on 1 April 2026, replacing the Income-tax Act, 1961. Most of the concepts survived the change and were renumbered rather than abolished, but every section number a trader learned over the last decade now points at a repealed provision.

This has a very practical consequence. Any checklist, spreadsheet template or working paper that hard codes a section reference is now citing something that no longer exists, and a great deal of published material has not caught up. The renumbering also means that a reconciliation performed today may straddle two Acts, because the year being reconciled and the year you are working in are not necessarily governed by the same statute. The examples throughout this page therefore cite both where the point matters: the provision that governs the constructed year, and the provision that carries the same rule forward.

The terminology moved too. The concept of a previous year has gone, replaced by a tax year, defined at section 3(1) of the new Act as the twelve month period of the financial year commencing on 1 April. The dates are unchanged. The word is not, and a search of your own notes for the old term will now miss things.

None of this changes a single number in the reconciliation above, which is rather the point. The mechanisms that make three records disagree are properties of what the records are for, not of the tax code of the moment. Rates, thresholds and section numbers change, and must be verified against the current text before they are relied on. The structure of the disagreement does not.

What the three numbers are actually for

It would be reasonable to finish a page like this believing that trading records are hopelessly unreliable. That is not the conclusion. The conclusion is that each of the three figures answers a narrower question than it appears to, and knowing which question each answers turns three sources of anxiety into three genuinely useful instruments.

The ledger tells you whether the account grew. It ties to your bank, it contains every cost, and it cannot be argued with, so when the question is whether the activity is worth doing it is the only one of the three that answers. The intraday book in the constructed year is the demonstration: a display showing a small gain, a ledger showing a loss, and cost as the whole of the difference.

The platform figure tells you what your capital is currently worth, which is what position sizing needs. It is right for a forward looking question and wrong for a backward looking one, and its worst failure mode is being read as a period result when it is a valuation at a point in time carrying years of accumulated movement. The taxable computation is the only one that belongs on a return, and its virtue is that it forces the split by head the other two blur. Producing it is how you find out that your intraday losses and your derivatives profits are not in the same pool, which is worth knowing regardless of the filing.

Run properly, the reconciliation takes an afternoon once a year and produces something more valuable than a reconciled number. It produces the knowledge that your records are complete, which is a different and better thing than the belief that they agree. Records that agree are usually records that are only being read one way. The habit of pulling four documents, fixing a basis, and naming every difference is a small operational discipline, and it is the sort of unglamorous machinery that separates an activity being run as a business from one being run on a screen. If that machinery is the interesting part rather than the tedious part, it is a fair description of the method we teach.

Every figure comes from a single deterministic construction, seeded so that it reproduces identically on each run. The trader, the holdings and the trades do not exist and are not modelled on any account. The activity was constructed so that each reconciling item appears at least once, which is the only way a worked reconciliation can demonstrate them, and the structural properties claimed in the prose are asserted in the code rather than described in a comment.

The charge model applies the statutory rates in force during 1 April 2025 to 31 March 2026, together with an illustrative allowance for brokerage, for derivative exchange transaction charges and for depository charges, each of which is commercial rather than statutory and varies between intermediaries. The rates in force from 1 April 2026 are deliberately not used, because they do not apply to the period being reconciled. The full charge schedule and its primary sources are set out on the companion page linked above.

The reconciliation uses the trade date as the basis for the taxable computation. This is stated as an assumption, not as a conclusion, for the reason given in the boundary section. The bridge is verified in two directions: the platform to ledger walk must close to the independently computed ledger figure, and the ledger to taxable walk must reach a head by head computation built separately from the same trade list. Both closures are asserted, and the run fails if either misses by more than a rupee of rounding.

Statutory references were read from the gazetted text of the Income-tax Act, 2025 and from the current edition of the accounting institute's guidance note on tax audit. Where a source could not be read at first hand it is not cited, and the page says so in the place where the citation would have gone. All results are illustrative and simulated. They are not a track record, not a forecast, and not an indication of what any account would produce.

• The real cost of a trade, worked to the paisa →
• How futures and options are taxed in India →
• Capital gains for the active trader →
• What T plus one settlement actually means →
• The trade journal standard and grader →