Lesson 02 · Trading

Short-term trading: day trading, scalping, swing & positional

By Robin CB · AI market analyst & investor

Short-term trading is negative-sum after costs: one trader's gain is another's loss, minus brokerage, taxes, and fees. SEBI's data on Indian individuals is unusually complete — and brutal. Read the base rates before anything else.

Last updated 31 Oct 2026 · Rates and rules change frequently — verify with SEBI/NSE and your broker before trading

The honest base rates

Equity F&O — SEBI's profitability studies

Study (SEBI)	Period / sample	Loss-makers	Aggregate net loss	Avg loss / trader
Jan 2023	FY22	89%	— (avg loss ₹1.1 lakh)	₹1.1 lakh
Sep 2024	FY22–FY24; >1 crore individuals	93%	> ₹1.8 lakh crore	~₹2 lakh (loss-makers)
Jul 2025	FY25; ~96 lakh traders, top 13 brokers	91%	₹1,05,603 crore	~₹1.1 lakh
Aug 2026	FY25–FY26; top 15 brokers (~90% of individuals)	87.7% in FY26	₹91,685 crore in FY26	₹1,16,654 in FY26

Only ~1% of individuals earned more than ₹1 lakh profit after costs over FY22–FY24. The top 3.5% of loss-makers (about 4 lakh traders) lost ₹28 lakh on average.
Costs are a large share of the damage. Over FY22–FY24 individuals paid roughly ₹50,000 crore in transaction costs (~51% brokerage, ~20% exchange fees). In FY26, ~₹24,800 crore of costs pushed ~4.4 lakh gross winners into net losses (secondary reporting).
Experience barely helps. In FY26, new traders lost at 87.81% and regular traders at 87.69% — virtually identical — while regular traders' turnover was ~45.7× their portfolio vs ~19.2× for new traders.
The other side is concentrated and algorithmic. Across FY25–FY26, prop traders grossed ~+₹90,437 crore and FPIs ~+₹44,980 crore, while individuals lost ~−₹1,70,125 crore gross; 99% of FPI/prop gross profits in FY26 came from algo-classified entities.
Losers keep trading. Over 75% of loss-making F&O traders kept trading in subsequent years (Sep 2024 study).

Source disagreement — shown, not resolved

The July 2025 study's FY25 aggregate loss (₹1,05,603 crore) differs from the figure in SEBI's later analysis tabled in Parliament (₹1,11,788 crore) — likely sample/revision differences (top 13 vs top 15 brokers). Treat the exact FY25 total as approximate; the ~91% loss rate is consistent. "Active traders" in FY26 is also reported as both 78.6 lakh (Parliament table) and 87.5 lakh (a different "active" definition) — use the Parliament-table figures when quoting.

Intraday equity (cash segment) — SEBI, July 2024

SEBI's peer-reviewed study of individual intraday traders (top-10 brokers, ~86% of individual clients; FY19, FY22, FY23), released 24 July 2024:

71% made a net loss in FY23, up from 65% in FY19. 80% of very frequent traders (>500 trades/year) lost money. 76% of traders under 30 lost money (their share of intraday traders rose from 18% in FY19 to 48% in FY23).
Costs added 57% on top of loss-makers' trading losses; profit-makers spent 19% of profits on costs. Frequent traders paid costs equal to 72% of their losses.
Participation surged: 15 lakh intraday traders in FY19 → 69 lakh in FY23; average trade size fell from ₹76,205 to ₹35,333. 54% of traders with three years' experience were still loss-makers.

International corroboration: Barber & Odean (2000, Journal of Finance) found the most active US households earned 11.4%/yr vs 17.9% for the market — and before costs, all turnover groups earned nearly identical gross returns (18.5–18.7%). The gap was costs, not stock-picking skill. Taiwan and Brazilian studies found the same pattern: fewer than 1% of day traders showed predictable positive abnormal returns after fees (Taiwan), and 97% of participants persisting >300 days lost money (Brazil).

Interpretation

The base rate for discretionary short-term trading by individuals is roughly "7–9 out of 10 lose, and the more you trade, the worse the odds." Any strategy or mentor claim must be judged against this base rate, not its best examples. A claim of consistent profitability needs audited, cost-inclusive, multi-year evidence — which retail educators almost never provide.

Mechanics

Intraday vs delivery

Intraday (MIS/CO): positions must be squared off the same day; the broker auto-squares-off, typically starting ~3:15–3:20 pm (broker-specific RMS cut-off — verify with your broker). No delivery, no DP charge, no overnight risk — but no chance to recover from a bad close either. Delivery (CNC): shares enter your demat on T+1 settlement. Most brokers allow MIS→CNC conversion with full margin before cut-off — never use conversion as a stop-loss plan; it turns a small intraday loss into an overnight position with gap risk.

Margin and the peak-margin regime

SEBI's peak-margin framework (circular July 2020, phased in through Aug/Sep 2021) requires 100% of prescribed margin upfront, monitored via at least four random intraday snapshots; the highest snapshot is the peak margin. Consequences:

Equity intraday leverage is capped at ~5× (minimum 20% margin) for eligible stocks; volatile/illiquid stocks demand more. The 20×–40× leverage of the pre-2020 era is gone.
F&O: SPAN + exposure margin upfront; the MIS (intraday) margin benefit in equity derivatives has been largely removed — in many futures contracts MIS and NRML (overnight) margins are now the same.
Options buyers pay the full premium upfront (mandatory from Feb 2025). Option sellers post SPAN + exposure margins that rise with volatility. An additional 2% Extreme Loss Margin applies to short index options on expiry day (from Nov 2024).

Leverage is no longer the binding constraint; costs, spreads, and skill are. Anyone selling "10× intraday leverage" in India in 2026 is misinformed or operating outside the framework.

Order types — and where each fails

Order	Behaviour	Failure mode
Market	Executes at best available price	Pays spread + slippage; in a gap, fills far from last price
Limit	Executes only at your price or better	No fill in a fast move
Stop-loss limit (SL)	Trigger → limit order	Gaps through your limit → no fill, stays exposed
Stop-loss market (SL-M)	Trigger → market order	Exit attempted, price unguaranteed — slippage in gaps/circuits
GTT	Broker-held conditional order	Broker-system dependent; verify behaviour in volatility

Bracket Orders (BO) were disabled by Zerodha in March 2020 and lost their margin advantage under peak-margin rules. Cover Orders (CO) remain available selectively by broker/segment. Availability is broker-specific and changes; verify before building any plan around them.

The auction / short-delivery trap

If you sell shares you don't hold and don't square off intraday, the exchange runs an auction on T+1 (2:00–2:45 pm) within a ±20% band. If the auction fails, the trade is closed out in cash — at the higher of the highest traded price from trade date to auction date, or 20% above the auction-day settlement price. Rules: never leave an intraday short open into the close; never short near an upper circuit; never sell BTST shares unless you understand the settlement chain. Price bands (2%/5%/10%/20% by category) freeze trading at the band — restrict short-term trading to liquid names (high daily turnover, tight spreads, large free float).

Day trading & scalping — why costs kill most scalpers

Scalping — holding seconds to minutes, targeting 1–3 ticks, many trades per day — is a market-microstructure game, not a prediction game. Every market order crosses the bid–ask spread; limit orders fill only when the market trades through them, precisely when price is moving against the side you wanted. And per SEBI's FY26 study summary, 99% of FPI/prop gross profits came from algo entities — a discretionary scalper competes for the same ticks against co-located, microsecond-latency systems.

The cost math — worked example

Rates as reported October 2026; STT futures/options rates reflect the April 2026 revision — verify on your broker's charges page. Buy 100 shares @ ₹1,000 (₹1,00,000), sell @ ₹1,005 (₹1,00,500). Gross profit ₹500 (+0.5%).

Item	Calculation	Amount
Brokerage	0.03% = ₹30 → capped ₹20, each side	₹40.00
STT	0.025% × ₹1,00,500 (sell)	₹25.13
Exchange charges	0.00307% × ₹2,00,500	₹6.16
SEBI fee	₹10/crore × ₹2,00,500	₹0.20
Stamp duty	0.003% × ₹1,00,000 (buy)	₹3.00
GST	18% × (40 + 6.16 + 0.20)	₹8.34
Total costs		₹82.83
Net profit	500 − 82.83	₹417.17

Breakeven: ₹82.83 ÷ 100 shares ≈ ₹0.83/share (~0.083%) before slippage. Shrink the target to a scalper's 0.1% (₹1/share, ₹100 gross): costs take ₹83, leaving ₹17 — and one adverse tick of slippage (₹0.05–0.20/share = ₹5–20) erases it. At 10 such trades a day you need a sustained win rate above ~55–60% after slippage just to stand still. Options scalping is worse: STT of 0.15% on premium + 0.03553% exchange charge + ₹20 brokerage per leg, and expiry-day (0DTE) options add theta decay measured in minutes. No public study isolates retail 0DTE scalping outcomes — the aggregate F&O loss data is the closest proxy. UNVERIFIED as a precise claim.

Swing & positional trading (days to weeks)

Swing: holding 2–10 trading days. Positional: weeks to a few months, usually trend-following on daily/weekly charts. Overnight exposure is the defining feature — and overnight gaps from results, guidance, macro data, and corporate actions can exceed any stop-loss: stops trigger at the next available price, not your stop price. Mitigations: smaller size, avoiding earnings/event dates unless the gap is the thesis, wider stops for event-prone names, and position limits such that a single gap cannot exceed ~1–2% of capital.

The strongest academic support for trend-based trading is at multi-month horizons: Jegadeesh & Titman (1993, Journal of Finance) found strategies buying past 3–12-month winners and selling past losers earned ~1% per month gross of costs (the 6/6 strategy, 1965–89 US data) — a portfolio effect over months, not a day-trading signal, replicated internationally. Caveats: gross-of-cost paper returns; momentum crashes exist; Indian evidence is thinner and regime-dependent. Short-horizon predictability is far weaker and contested. Interpretation: if your edge is "trend," the evidence favours weeks–months holding periods over minutes.

Technical analysis — the honest evidence

VWAP: the institutional execution benchmark. No evidence that "price above VWAP = bullish" predicts returns. A fair-value reference for judging fills, not a proven edge.
EMA / MA crossovers: mixed. A 2015–25 NIFTY-50 study (IJEDR, low-tier journal — treat cautiously) reported SMA crossovers at 52.3% "accuracy", Sharpe 0.34, below buy-and-hold's 0.45. Use EMAs for trend definition, not standalone signals.
RSI (14) / MACD: Chong & Ng (2008) and Chong, Ng & Liew (2014) found effects that are market- and parameter-specific and fragile; none beat buy-and-hold on the Nikkei, and results survived a 1% round-trip cost only in some markets. RSI "overbought" stays overbought in strong trends — the classic trap.
ATR: no directional content by design. Legitimate use is sizing and stops: stops as multiples of ATR (e.g. 1.5–2×) so stops respect the stock's normal noise. Risk practice, not prediction.
Bollinger Bands: one Nifty study reported a Bollinger+RSI combination outperforming buy-and-hold — single-study, low-tier-journal, likely data-mined. Do not treat as established.

Chart patterns — Bulkowski, read carefully

Bulkowski's Encyclopedia of Chart Patterns is the largest pattern database. Head-and-shoulders tops (bull market, >2,800 patterns): 19% break-even failure rate (81% moved at least 5% in the expected direction), but only 51% reached the measured-move price target, with pullbacks 50–64% of the time. Head-and-shoulders bottoms: 11% failure rate, 71% meeting target. "81% success" is not an 81% profitable-trade rate — it is a 5%-move threshold on daily charts, before costs, slippage, stops, and sizing, with patterns identified in hindsight. Osler (1998) tested head-and-shoulders trading directly on 100 US stocks: patterns occurred ~once per stock per year, trading them averaged −0.24% per round trip before costs. Patterns are context only — never standalone signals, never on intraday charts where these statistics do not apply.

Risk management — the only part with hard math

Position sizing

Fixed fractional risk: risk a fixed small percentage of capital per trade (commonly 0.5–2%; 1% is the usual starting point). Position size is derived, not chosen:

Shares = (Capital × Risk%) ÷ (Entry − Stop-loss)

Example: ₹5,00,000 capital, 1% risk = ₹5,000. Entry ₹1,000, stop ₹980 → ₹20/share risk → 250 shares. If the stop must be wider, the size shrinks. This single rule prevents most blow-ups.

R-multiples and expectancy

Express every trade in units of initial risk (R). Expectancy E = (Win% × Avg Win) − (Loss% × Avg Loss). Example: 40% win rate, avg win 2R, avg loss 1R → E = (0.40 × 2) − (0.60 × 1) = +0.2R per trade. A 40% win rate is fine if winners are twice losers — and a 60% win rate loses money if winners are half the losers. Win rate alone is meaningless. You need 50–100+ logged trades under the same rules before an expectancy estimate is even roughly meaningful.

Drawdown math

10% drawdown needs 11% to recover. 25% needs 33%. 50% needs 100%. This asymmetry is why survival dominates optimisation. Daily loss limits (e.g. stop after −2% of capital in a day, or after 3 consecutive losses) are circuit breakers against tilt.

Kelly criterion — powerful, dangerous, usually misused

Kelly maximises long-run geometric growth: f* = (b·p − q) / b, where p = win probability, q = 1−p, b = win/loss payoff ratio. Example: p = 0.55, b = 1.5 → f* = 25% of capital per trade — mathematically optimal and unusable in practice. A 5-point error in win rate can change the fraction 3× or more; full Kelly carries roughly a 1-in-3 chance of a 50% drawdown. The standard compromise is quarter- to half-Kelly; most retail traders sensibly risk 0.5–2%, far below Kelly. Kelly's real value is as a ceiling: sizing above Kelly is provably growth-destroying even with a genuine edge.

Psychology & process

Barber & Odean (2000): heavy traders weren't worse pickers — they paid more costs. Mechanisms: overconfidence; the disposition effect (Odean 1998 — winners sold outperformed losers kept by 3.4 points over the following year); revenge trading after a loss; decision fatigue from dozens of decisions per day. The process professionals use:

Written plan before the open: setups, entry/exit rules, max risk per trade, max daily loss, max trades per day. If it's not written, it's a mood, not a plan.
Trading journal: every trade logged (setup, entry, exit, R-multiple, screenshot); weekly review scores "did I follow the plan?" separately from "did I make money?"
Review routine: weekly — expectancy by setup, rule violations; monthly — within historical drawdown bounds?; quarterly — should this strategy be traded at all?
Pre-commitment devices: hard daily loss limit, position-size caps, no-trading lists (no first-5-minutes trades, no averaging down).
Paper trading first: 2–3 months simulated with real-time data and full cost accounting — it tests whether you can follow the process without paying tuition to the market.

Taxation — high level only; confirm with a CA

Item	Intraday equity (cash)	F&O (futures & options)
Classification	Speculative business income, u/s 43(5)	Non-speculative business income
Tax rate	Slab rates	Slab rates
Loss set-off	Only against speculative income	Against any business income (except salary)
Loss carry-forward	4 assessment years (return filed on time)	8 assessment years
ITR form	ITR-3 (Schedule BP)	ITR-3
Presumptive taxation (44AD)	Excluded — speculative business cannot use 44AD	Generally routed to ITR-3; 44AD use in trading is contested and risky

Turnover for audit purposes is the absolute sum of profits and losses (plus premium for options) — not gross traded value. Audit (44AB): turnover above ₹10 crore where cash receipts/payments are each ≤5%; ₹1 crore otherwise. Do not assume STT is deductible — its treatment has been litigated. Consult a chartered accountant before filing.

What SEBI changed in 2024–25

SEBI's 1 October 2024 circular imposed six measures on index derivatives, phased in Nov 2024–Apr 2025: (1) minimum contract value ₹15–20 lakh (from ₹5–10 lakh) — Nifty lot 25 → 75 (verify current lot sizes on NSE); (2) one weekly expiry per exchange — Nifty on NSE, Sensex on BSE; (3) upfront collection of option premium from buyers (Feb 2025); (4) removal of calendar-spread margin benefit on expiry day (Feb 2025); (5) additional 2% Extreme Loss Margin on short options on expiry day (Nov 2024); (6) intraday monitoring of position limits (Apr 2025; at least 4 snapshots/day). Plus mandatory risk disclosure — brokers must show at login that 9 out of 10 individual F&O traders lost money (FY22 basis). From 1 April 2026, STT rose to 0.05% on futures sales and 0.15% on option premium/exercise. Verify all of the above on sebi.gov.in / nseindia.com before trading — rules in this segment change frequently.

Bottom line

1. Base rates first: ~71% of intraday equity traders and ~88–93% of F&O traders lose money net of costs; experience barely improves the odds. 2. Costs are the strategy for sub-day holding periods. 3. Risk is the only controllable edge. 4. Longer horizons have better evidence. 5. Paper-trade and backtest with full costs before risking capital.