What is scalping trading?

A scalp is a bet that the very next increment of price will arrive before the trade's invalidation does. Not the next move, the next increment: a few ticks on an index future, fifty paise on a liquid stock. The scalper holds no view on the session and is not forecasting the close; they are reading the pressure of bids and offers at the touch and the momentum of the next thirty seconds. Winners are booked the moment they exist. Losers are cut faster. Nothing is carried, which sounds like prudence and is really the definition of the style: flat dozens of times a day means paying the market's toll dozens of times a day.

Three numbers define the style and they are worth stating plainly, because most descriptions of scalping skip straight to the psychology. The holding period is seconds to a few minutes. The target is a few ticks, typically well under half a percent, which on a liquid ₹500 stock means aiming for something in the region of fifty paise to a rupee. The trade count is twenty to forty round trips in a session. It is the third number that does the damage, and it does it in a way the first two conceal, because frequency is not a description of style so much as a multiplier applied to every fixed cost the style incurs.

That is what separates scalping from its neighbours rather than any difference in charting. An intraday momentum trader who holds for hours and targets one to three percent pays exactly the same stack of levies, but the target is many multiples of the toll, so the toll behaves like a rounding error. The same is true of a delivery position held for weeks, as our guide to intraday versus delivery trading sets out in detail. Scalping is not a different kind of analysis. It is the same analysis compressed until the toll and the target are the same size, and that compression is the entire subject.

To see the problem properly you have to stop measuring cost in rupees and start measuring it in the unit the scalper actually trades in: ticks. A tick is the minimum price step, ₹0.05 for an NSE cash-segment stock priced at ₹250 or above. On an illustrative 400-share position in a ₹500 stock, one tick is worth ₹20. Convert the whole toll into that unit and something clarifying happens, because now the cost and the target are denominated in the same thing and can be compared directly rather than rhetorically.

Run the published rate model over that position and the round trip costs ₹118.28 in invoiced charges plus ₹40.00 of spread, ₹158.28 all in, which is 7.914 ticks. Now the awkward question answers itself. A symmetric scalp with a target of T ticks and a stop of T ticks, facing a toll of c ticks, breaks even when the gross hit rate reaches (T + c) divided by 2T. With no toll at all that expression is exactly one half, the coin flip everyone expects. With c at 7.914 ticks, a ten-tick target needs 89.6 percent of scalps to win, and a twenty-tick target still needs 69.8 percent. The entire distance between 50 percent and those figures is the toll, and none of it is psychology.

Scalping needs a high hit rate not because small losses are pleasant, but because the toll is charged per round trip regardless of outcome.

Push the formula one step further and it produces the hardest fact on this page. The required hit rate reaches 100 percent exactly when T equals c, because at that point a winner recovers precisely what the round trip cost and every loser is pure subtraction. Below that target, no hit rate whatsoever breaks even, not 95 percent, not 99, not a perfect record. On this position that wall sits at 7.9 ticks, which is roughly forty paise on a ₹500 stock, and it is uncomfortably close to where scalpers actually aim. Trading is not being made hard there. It has been made arithmetically impossible, and the chart below shows exactly where the door closes.

The 7.9 ticks is not an estimate; it is a sum of published rates, and it is worth walking because the shape of the bill is as instructive as the total. Take the illustrative position again: 400 shares of a stock near ₹500, about ₹2,00,000 deployed, bought and sold within the session. Brokerage at a representative Indian retail broker is the lower of about 0.03 percent or ₹20 per executed order, which on this size is the ₹20 cap, so ₹40 for the round trip. Securities transaction tax on intraday equity is 0.025 percent of the sell side, ₹50.10. Exchange transaction charges at 0.00307 percent of turnover on both sides come to ₹12.29. The SEBI turnover fee is ₹10 per crore, ₹0.40. Stamp duty is 0.003 percent on the buy side, ₹6.00. GST at 18 percent applies to brokerage plus exchange charges plus the SEBI fee, but not to STT or stamp duty, which are themselves taxes: ₹9.48.

That is ₹118.28 of invoiced cost, the part a contract note itemises. Then comes the line no contract note shows. A scalper who needs immediacy crosses the spread twice, giving up one tick on entry and one on exit, ₹0.05 × 400 × 2 = ₹40.00. The all-in toll is ₹158.28. Every rupee figure on this page is computed from the same published rate model that drives our cost estimator, and the rates are as notified as of 17 July 2026; verify them against your own broker's schedule and the current exchange circulars before relying on them, because Finance Acts and exchange notifications move these numbers.

Now read the table in the right column rather than the left, because the tick column is where the insight lives. The statutory toll in ticks does not depend on your position size at all. Every levy is a percentage of value, and a tick is worth the tick size multiplied by your quantity, so the quantity cancels out of the ratio entirely. What is left is a property of the instrument, its price divided by its tick size, and nothing you do about sizing touches it. Only brokerage, being a flat ₹20 cap, shrinks in tick terms as you scale up. That is why the toll falls from 12.6 ticks at 100 shares to 7.9 at 400 and only 5.6 at 20,000, converging on a floor of about 5.5 ticks that no amount of capital gets under. Scale is a real lever, but it is a small one, and it is exhausted quickly.

Of the eight lines in that table, one deserves separate treatment, and it is worth being precise about why rather than repeating the folklore. The spread is not the largest line on the illustrative position: at 2.000 ticks it sits just under the STT's 2.505. What makes it the line to understand first is a different pair of properties. It is charged first, at the instant of the fill, before any levy is computed on anything. And it is the only line that can grow without limit. Every other entry in that table is a notified rate you can look up this evening and will still be true tomorrow. The spread is a live quantity set by whoever happens to be resting orders when you press the button.

The mechanics are geometry, not luck. Every instrument has a bid, the highest price a buyer will pay, and an ask, the lowest a seller will accept, separated by at least one tick, as our guide to the bid-ask spread works through. A scalper needs immediacy and therefore cannot rest an order and wait: they buy by lifting the ask and sell by hitting the bid, surrendering the spread on both legs even when the price does not move at all. Resting a limit order at the bid avoids the toll but introduces adverse selection, because the order fills most readily at exactly the moment the price is coming through it. There is no third option, and that is the point.

What decides the size of that first charge is not the quoted spread but the depth behind it, which is why liquidity is a scalper's first filter rather than a nicety. If two thousand shares rest at the best ask, an illustrative 400-share market buy fills entirely at the touch and pays one tick above the mid. If only 120 rest there, the same order walks up through the next two levels and fills at a volume-weighted ₹500.10, two ticks above the mid instead of one. Same order, same intent, same quoted spread, twice the cost. Push that through the hit-rate formula and a ten-tick scalp goes from needing 89.6 percent to needing 99.8 percent, which is to say it stops existing. A scalper's instrument list is not a preference; it is the constraint that decides whether the arithmetic is open or closed.

Everything so far has priced a single round trip. The style's defining number is the twentieth to fortieth one, and this is where the levies stop being a bookkeeping detail and become the strategy's dominant term. The mechanism is almost insultingly simple: the toll is charged per round trip and the target does not grow when you trade more often, so turnover multiplies friction linearly while the edge per trade stays exactly where it was. Both quantities are straight lines in the number of trades. Straight lines that start together and diverge never meet again, which means no amount of frequency rescues a per-trade number that is already negative. Frequency is a multiplier, and a multiplier applied to a negative number does not help.

The levies that hurt most here are the ones linked to turnover rather than to profit, and this is the crucial distinction that our guide to trading taxation in India develops properly. Income tax is charged on what you make, so it takes a share of a good year and nothing from a bad one. STT, exchange charges, stamp duty and the SEBI fee are charged on what you transact, so they are indifferent to whether the trade worked. Consider what frequency does to the base they are applied to. The illustrative position is ₹2,00,000 of capital, but each round trip transacts about ₹4,00,400 of turnover, and twenty round trips a day across twenty-two sessions turns that ₹2,00,000 into roughly ₹17.6 crore of monthly turnover.

Apply the small rate to the large base and the result is the number that should end most scalping ambitions before the first order. Intraday STT is 0.025 percent, a rate so small it looks like a rounding error on any single trade, just ₹50.10. Across 440 round trips it is ₹22,044 a month, about 11 percent of the deployed capital, in that one levy alone. The full all-in toll comes to ₹69,642 a month, roughly 35 percent of the capital, paid every month simply to keep playing. And notice what the trader's own screen would show while this happens: a majority of green trades, disciplined losses, and an account that shrinks anyway, because the wall is invisible at the level of any single trade and undeniable only in aggregate.

Suppose the arithmetic clears. Suppose you have found an instrument deep enough to hold the spread at one tick and a target far enough above 7.9 ticks that the required hit rate is merely difficult rather than fictional. What the style then asks for is a short and unforgiving list, and it is worth separating the requirements you can buy from the one you cannot. Depth and a one-to-two-tick spread you can select. A computed cost stack, the table above rebuilt with your own numbers, you can produce in an evening. Order-entry routine drilled to keystroke speed you can practise, and you must, because at this horizon a two-second hesitation is not indecision, it is a repricing. Written per-trade and per-day limits you can set in advance, and they have to be mechanical, because forty discretionary decisions a day under a live profit and loss is the exact environment in which resolve degrades.

One requirement cannot be bought at any retail price, and it is better to state it plainly than to let a platform's marketing imply otherwise. A retail order travels from a device, across a public network, through a broker's risk checks, to the exchange: a journey measured in tens to hundreds of milliseconds. Professional participants run co-located servers inside the exchange's own data centre, with round trips measured in microseconds. That is a gap of several orders of magnitude, and no subscription, no faster laptop and no lighter charting package closes it. Better software makes a retail scalper faster than other retail traders; it does not make them fast.

The honest conclusion is narrower than "manual scalping is impossible" and more useful. It is that any strategy whose profit depends on reacting to a public event before someone else does is lost before it begins, because the someone else is a machine sitting in the building. That rules out reaction races to news, to quote changes, and to the appearance of size in the book, which is a large fraction of what short-horizon trading folklore actually describes. Whatever survives at retail must not depend on being first, and a scalper who cannot say clearly what their edge depends on instead has almost certainly not escaped the category.

There is one cost the rate model cannot price, and it does not appear on the contract note or in the tick table. Forty round trips a day is eighty order decisions, each made in seconds, each under a running profit and loss, and each an opportunity to depart from the plan. Across twenty-two sessions that is roughly 880 decisions a month. A trader working a slower horizon, taking perhaps two positions a week, makes something like eight. The two traders may be identically disciplined, identically prepared and identically calm. They are not identically exposed, because exposure to error is a function of how many chances you give it.

Put a number on it and the asymmetry becomes hard to argue with. Suppose both traders follow their own written rule 99 percent of the time, which would be exceptional discipline by any honest measure. The slower trader expects 0.08 deviations a month, which is to say roughly one a year. The scalper, at the same 99 percent, expects 8.8 a month. Same discipline, same person, 110 times the errors, purely because the style asked the question 110 times as often. To bring the scalper's monthly error count down to the slower trader's, they would need to follow their rule 99.991 percent of the time, which is not a standard of discipline that human beings meet under time pressure.

And this is where the psychological load rejoins the arithmetic rather than sitting beside it as a separate topic. In most styles a deviation costs you the difference between the plan and what you did. In scalping a deviation is usually another round trip, and another round trip is another ₹158.28. Nine unplanned trades in a month is about ₹1,400 of pure toll on the illustrative position, spent on trades the plan never sanctioned. The style does not merely tax you for trading. It taxes you again, at the same rate, for every occasion on which its own tempo made you trade when you should not have.

None of the arithmetic on this page is hypothetical at the population level, because the regulator publishes the outcomes. SEBI's September 2024 study of individual traders in the equity derivatives segment found that about 93% of individual traders in equity derivatives made net losses over FY22 to FY24, aggregate net losses exceeding ₹1.8 lakh crore (SEBI, September 2024). That is the segment where India's highest-frequency retail activity lives, and transaction costs are named in the study itself as a material drag on outcomes, including on the small minority who finished ahead.

It is worth being careful about what that figure does and does not establish. It is not evidence that short-horizon trading is impossible, nor that everyone who tries it is naive, and it is not a prediction about any individual. It is a base rate: the prior you are entitled to start from before you have any evidence about yourself specifically, and the number a reasonable person would want to know before committing capital to the most active end of that same population. Scalping does not merely sit inside this group. It sits at the point where the cost term repeats fastest per rupee of capital, which is the one mechanism this page has been able to price exactly.

And that is what makes the base rate legible rather than merely discouraging. There is no need to reach for a theory about retail traders being emotional or undisciplined to explain it, though both are sometimes true. The toll is charged on every round trip whether the trade was right or wrong; the required hit rate rises as the target shrinks; below a target of about 7.9 ticks nothing works at any hit rate; and frequency multiplies friction in a straight line. A population trading that way, at that frequency, against that stack, produces that distribution of outcomes. The data and the arithmetic are telling the same story, which is usually a sign that both are right.

The point of pricing a style honestly is not to talk anyone out of it; it is to replace a vague enthusiasm with a specific test that can be passed or failed. Treat scalping the way an engineer treats a specification. Some professionals do operate it, with co-located infrastructure, rebate-aware execution and cost stacks tuned to the rupee. None of those conditions describe a retail learner with a laptop, and the gap between those two situations is not effort or nerve. It is the toll, and the toll is measurable in advance, which is the one genuinely good piece of news on this page.

So the productive order runs opposite to the one most people follow. Compute the toll in ticks first, on the instrument you actually intend to trade, using your own broker's schedule. Set it against your target, and if the target is not a large multiple of the toll, stop there; the rest of the plan is decoration. Then read off the hit rate the formula demands and ask, without flattering yourself, whether you have any evidence you can reach it. Only after all three of those does chart-reading matter at all. Learn to read structure on timeframes where the toll is a rounding error, prove a positive expectancy there with a journal, and only then decide whether compressing the horizon, and multiplying the toll, buys you anything at all. That upstream sequence, arithmetic before analysis, is exactly what the method we teach is built to install.

The market charges by the transaction, and it charges the same whether the transaction was clever. That single sentence is the whole of scalping's economics, and every figure on this page is a restatement of it. The fewer transactions your skill needs, the more of the move you keep, which is a conclusion about arithmetic rather than about temperament, and it is available to anyone willing to do the multiplication before rather than after.

• The base rate for the population. Securities and Exchange Board of India, Analysis of Profit and Loss of Individual Traders Dealing in Equity F&O Segment (September 2024): about 93% of individual traders in equity derivatives made net losses over FY22 to FY24, aggregate net losses exceeding ₹1.8 lakh crore. The study also identifies transaction costs as a material drag on outcomes. sebi.gov.in
• The statutory levy schedule. Securities transaction tax on intraday equity at 0.025 percent of the sell side; exchange transaction charges on the cash segment; the SEBI turnover fee of ₹10 per crore; buy-side stamp duty at 0.003 percent for intraday; GST at 18 percent on brokerage plus exchange charges plus the SEBI fee, and not on STT or stamp duty. Summarised on NSE's investor reference for turnover fees, STT and other levies. Rates as of 17 July 2026 and subject to Finance Acts and exchange circulars; verify at source. nseindia.com
• The rate model behind every figure here. All rupee and tick figures on this page are computed from the same published round-trip cost model that drives our own cost estimator, applied to an illustrative 400-share intraday position in a ₹500 stock. The model itemises brokerage, STT, exchange transaction charges, the SEBI turnover fee, stamp duty and GST on their correct bases across both legs. Cost estimator
• Margin framework and the end of extra intraday leverage. SEBI circular SEBI/HO/MRD2/DCAP/CIR/P/2020/127 (20 July 2020) established the framework for verifying upfront collection of margins in the cash and derivatives segments, phased to full upfront margin by September 2021. As of 17 July 2026 that position stands; verify the current framework at source. sebi.gov.in
• On the tick size and what it does to the arithmetic. The NSE cash-segment tick is ₹0.05 for stocks priced at ₹250 and above. Because every statutory levy is a percentage of value and one tick is worth the tick size multiplied by your quantity, the quantity cancels: the statutory toll expressed in ticks is a property of the instrument's price-to-tick ratio, not of your position size. Only the flat brokerage component shrinks with scale, which is why the toll converges to a floor rather than falling towards zero.