How Do AI Intraday Signals Work? Entry, Target and Stop-Loss Explained

What components make up an AI intraday signal?

Every complete intraday signal consists of four essential components that transform raw market data into an actionable trade setup. The direction indicates whether the opportunity is a BUY or SELL. The entry price specifies where to initiate the position, typically near the current market level but adjusted to avoid chasing momentum spikes. The target price defines the profit objective, usually calculated as a multiple of the risk being taken. The stop-loss price sets the maximum acceptable loss, protecting capital when the trade thesis proves incorrect.

These four elements work together as a complete risk framework. Without a defined entry, traders chase prices and enter at unfavourable levels. Without a target, profit-taking becomes emotional rather than systematic. Without a stop-loss, a single losing trade can eliminate gains from many winning positions. The combination creates a repeatable process that removes guesswork from position management.

The mathematical relationship between these levels determines the signal's risk-reward ratio. A trade with entry at ₹100, stop-loss at ₹98 and target at ₹104 carries a 1:2 risk-reward ratio because the potential gain (₹4) is twice the potential loss (₹2). This structure allows profitability even when fewer than half of all signals reach their targets, provided traders consistently honour both the stop-loss and target levels.

How does the stock selection process work?

AI signal systems begin by defining a fixed universe of securities to scan. Consistency in this universe ensures that results remain comparable across different market conditions. Systems that scan the entire NSE equity segment often generate signals in illiquid stocks where the bid-ask spread alone erases any theoretical edge. A focused approach using indices such as NIFTY 500 provides approximately 500 stocks with sufficient liquidity for retail traders to enter and exit positions at prices close to the signal levels.

Liquidity screening happens before any technical analysis. The system filters for minimum average daily turnover, typically several crores, and examines the five-day average volume. Stocks that suddenly appear on the scanner due to a one-day volume spike often represent event-driven moves (results, news, block deals) that do not suit systematic intraday strategies. The filtering removes these outliers, leaving only those securities trading with predictable liquidity patterns.

Scanning occurs continuously during market hours from 9:15 AM to 3:30 PM IST. The system refreshes data every few minutes, recalculating indicator values as new price and volume information arrives from the exchange. This real-time approach allows the AI to detect developing setups as they form, rather than relying on end-of-day analysis that misses intraday opportunities. The continuous scanning also means a stock can appear in the morning session, disappear by midday if conditions change, then reappear in the afternoon if the setup reforms.

Which indicators does the AI combine?

No single technical indicator provides reliable signals across all market conditions. Trend indicators work well in directional markets but generate false signals during consolidation. Oscillators excel at identifying overbought or oversold extremes but lag during strong trends. Volume indicators confirm the strength of moves but cannot predict direction alone. AI signal systems address these limitations by scoring multiple indicators simultaneously and requiring agreement across several categories before generating a trade setup.

Trend indicators such as Exponential Moving Averages (EMA) form the foundation. A common pairing uses the 9-period and 21-period EMAs. When the faster 9-period EMA crosses above the slower 21-period EMA, the system scores the stock positively for bullish setups. The reverse crossover suggests bearish conditions. The system also examines whether price is trading above or below both EMAs, assigning higher scores to stocks that show clear directional alignment.

Momentum oscillators add a second dimension. The Relative Strength Index (RSI) measures the magnitude of recent price changes on a 0-100 scale. Values above 70 suggest overbought conditions, while readings below 30 indicate oversold levels. The Moving Average Convergence Divergence (MACD) compares two EMAs of price and plots the difference as a histogram. When MACD crosses above its signal line, the momentum is turning positive. The system looks for RSI moving through key levels (such as rising above 50 in a bullish setup) while MACD confirms the same direction.

Volume analysis separates genuine moves from weak price drifts. The system compares current volume against the 20-day average volume. A bullish price move on volume 50% above average carries more conviction than the same price move on below-average volume. Volume-Weighted Average Price (VWAP) provides an additional reference point. VWAP calculates the average price weighted by volume throughout the session. Stocks trading above VWAP are in above-average demand, while those below VWAP face selling pressure. A bullish signal becomes stronger when price is both rising and positioned above VWAP.

Price action patterns complete the picture. The system identifies candle formations such as engulfing patterns, inside bars and breakouts from short-term ranges. These patterns provide context for the indicator readings. A bullish crossover of moving averages becomes more significant when it occurs alongside a bullish engulfing candle that breaks above a prior swing high.

The scoring system assigns numerical values to each indicator. A perfect alignment might score 10, while a marginal setup scores 4 or 5. The AI only generates a signal when the combined score exceeds a preset threshold, typically 7 or 8 out of 10. This multi-indicator requirement filters out ambiguous setups where some indicators suggest bullish conditions while others remain neutral or bearish.

How are entry levels calculated?

The entry price represents a compromise between two competing goals: entering near the current price to avoid missing the move, and waiting for a pullback to improve the risk-reward ratio. Chasing a stock that has already moved sharply often results in entering at a local peak, followed by an immediate pullback that triggers the stop-loss. Waiting too long for a perfect pullback means missing the trade entirely as price runs toward the target without offering an entry opportunity.

Many AI systems place the entry level at or just beyond the current price, often rounded to the nearest ₹0.05 or ₹0.10 for ease of order entry. For a bullish signal, the entry might be set 0.2% to 0.5% above the current market price, anticipating that the stock will consolidate briefly then resume its upward move. This approach ensures the trader enters only if the bullish momentum continues, avoiding situations where price reverses immediately after the signal appears.

Some systems use VWAP as an entry reference. In a bullish setup, the entry might be placed at VWAP plus a small buffer, on the theory that institutional buying pressure exists above VWAP. Price dipping to VWAP provides an attractive entry point where both technical indicators align and the intraday average price supports the position. This VWAP-based approach works particularly well in stocks with strong intraday trends where price repeatedly tests and bounces from VWAP throughout the session.

The entry level should never be so far from the current price that it becomes unlikely to trigger. Placing a bullish entry 2% above the current level in a stock that typically moves 1% per hour means waiting several hours for the entry, by which time market conditions may have changed entirely. The optimal entry sits close enough to trigger within the next 15-30 minutes if the setup remains valid.

How are stop-loss levels determined?

The stop-loss represents the price level where the trade thesis is proven wrong and the position should be exited to prevent further losses. Proper stop-loss placement balances two risks: setting it too tight results in premature exits from trades that would ultimately succeed, while setting it too loose allows losses to grow unnecessarily large when the setup genuinely fails.

Average True Range (ATR) forms the mathematical basis for most AI-generated stop-losses. ATR measures the average distance between the high and low prices over a specified period, typically 14 days. A stock with ATR of ₹20 typically moves ₹20 between its intraday high and low. The stop-loss is then set at a fraction of ATR below the entry for a bullish trade, commonly 0.5× to 1.0× ATR. This approach ensures the stop-loss adapts to each stock's volatility profile.

For example, a volatile stock with ATR of ₹30 and entry at ₹1,000 might receive a stop-loss at ₹970 (1.0× ATR below entry). A stable stock with ATR of ₹10 and entry at ₹500 might get a stop-loss at ₹495 (0.5× ATR below entry). Both stops give the trade room to move according to the stock's typical behaviour, while the absolute rupee risk remains proportional to volatility.

Support and resistance levels provide additional context. If the ATR-based calculation places the stop-loss at ₹485 but a clear support level exists at ₹480, the system may adjust the stop to ₹479.50, just below support. This adjustment recognises that other traders are likely watching the same support level. Price may test ₹480 without breaking it, and a stop at ₹485 would exit prematurely. Conversely, if support breaks, price often accelerates lower, making ₹479.50 an appropriate exit point.

The stop-loss is never moved closer to the entry once the trade is live. Tightening stops turns a carefully calculated risk level into an arbitrary one, typically triggered by fear as price moves against the position. The only acceptable stop-loss adjustment is a trailing stop that moves in the direction of profit after the trade has moved favourably, but this occurs only after price has reached certain milestones, not in response to adverse movement.

How are target levels established?

The target price represents the profit objective where traders should exit the position and realise gains. Unlike stop-losses, which are rooted in volatility measures, targets derive primarily from the risk-reward ratio the system is designed to achieve. A 1:2 risk-reward ratio means the target is set at a distance twice as large as the stop-loss. If entry is ₹1,000 and stop-loss is ₹980 (risk of ₹20), the target is placed at ₹1,040 (reward of ₹40).

This mathematical approach ensures consistency across all signals. Every trade offers the same theoretical reward relative to its risk, making it possible to evaluate the system's performance fairly. A signal system that mixes 1:1 ratios with 1:3 ratios across different stocks creates an inconsistent performance profile where gains and losses do not balance predictably. Fixed risk-reward ratios allow traders to calculate the required win rate for breakeven. At 1:2, the system breaks even with a 33% win rate (before costs). At 1:1.5, breakeven requires 40%. At 1:3, breakeven drops to 25%.

Resistance levels and technical targets sometimes modify the risk-reward calculation. If the ATR-based stop and fixed multiple suggest a target of ₹1,050, but strong resistance exists at ₹1,045 based on prior swing highs, the system may lower the target to ₹1,044. Placing a target above a known resistance level reduces the probability of the target being hit, as price often stalls at resistance and reverses. The adjustment maintains realistic expectations while preserving the general risk-reward framework.

Some AI systems use multiple targets, exiting a portion of the position at the first target (perhaps 1:1.5 risk-reward) and holding the remainder for a larger gain (2:1 or beyond). This tiered approach increases the percentage of trades that produce some profit, improving the psychological experience of trading the signals, while still capturing larger moves when they occur. The published signal shows both levels, allowing traders to choose their preferred approach.

Worked example: BHARTIARTL intraday setup

The following example uses actual NSE daily closing data for BHARTIARTL to demonstrate how an AI signal system would process real market conditions. These are official exchange closing prices, not intraday levels, but they illustrate the mechanics of entry, stop-loss and target calculation.

On 2026-10-15, BHARTIARTL closed at ₹1,854.00 after a strong session (open ₹1,795.10, high ₹1,871.10, low ₹1,795.10). Volume of 11,315,521 shares was more than double the previous day's 5,185,554 shares. This combination of a large up move on expanding volume would trigger positive scores on both momentum and volume indicators.

Assume the signal system scanned the stock near the close of 2026-10-15 and prepared a bullish signal for 2026-10-16. The 14-period ATR for BHARTIARTL was approximately ₹35 based on recent daily ranges. The system sets the entry at ₹1,866.00 (slightly above the previous high to confirm continuation), the stop-loss at ₹1,830.00 (roughly 1.0× ATR below entry, or ₹36 risk), and the target at ₹1,938.00 (2.0× the risk distance above entry, or ₹72 reward). The risk-reward ratio is 1:2.

On 2026-10-16, BHARTIARTL opened at ₹1,866.00, exactly at the signal's entry level. Price rallied during the session, reaching a high of ₹1,909.70 and closing at ₹1,905.00. The high of ₹1,909.70 did not reach the target of ₹1,938.00, so the position would remain open at the end of 2026-10-16 if this were an intraday signal extended overnight (though most intraday systems close all positions by 3:20 PM IST).

On 2026-10-19, the stock opened higher at ₹1,913.80 and touched ₹1,929.60 intraday, still short of the ₹1,938.00 target. Price then reversed sharply to close at ₹1,875.20. The low of ₹1,863.90 came within ₹33 of the stop-loss at ₹1,830.00 but did not trigger it.

By 2026-10-23, BHARTIARTL closed at ₹1,892.10 after touching an intraday low of ₹1,860.70. The position remained open but had moved little from the entry level despite high volatility. On 2026-10-27, the stock finally rallied strongly, opening at ₹1,916.40, breaking above the prior high, and closing at ₹1,978.80 with an intraday high of ₹1,978.80. This high exceeded the target of ₹1,938.00 by ₹40.80, delivering the intended 1:2 risk-reward gain.

In an actual intraday system, this trade would have been closed on 2026-10-16 at 3:20 PM IST regardless of the target being reached. The example demonstrates how the signal levels (entry ₹1,866.00, stop ₹1,830.00, target ₹1,938.00) relate to real market movement. Price required several sessions to reach the target, illustrating why intraday systems often see fewer signals reach their full targets compared with swing trading systems that hold positions overnight.

How should traders measure signal performance?

Transparent measurement separates credible signal systems from marketing claims. Any provider can publish a selection of successful trades while omitting the losses. Honest performance tracking requires several elements. First, every signal generated must appear in the results, including those that hit stop-losses or expired without reaching the target. Second, the measurement process must honour the stop-loss level, not assume the trader held through it. Third, costs such as brokerage, STT (Securities Transaction Tax), and GST must be included in the net return calculation.

Replay testing applies the signal rules to historical data, processing each time period in sequence as if trading in real time. The system does not know future prices when it generates a signal. It measures whether the subsequent price action hit the stop-loss or target first, and records the appropriate gain or loss. This candle-by-candle approach mirrors real trading, where the worst intraday price matters, not just the closing price.

The key metrics include win rate (percentage of signals that hit the target before the stop), average gain per winning trade, average loss per losing trade, and overall profit factor (total gains divided by total losses). A system with a 40% win rate, average gain of ₹8,000 per winner, and average loss of ₹4,000 per loser produces a profit factor of 1.33× before costs ((0.40 × ₹8,000) / (0.60 × ₹4,000) = 3,200 / 2,400). After subtracting transaction costs, the net profit factor might drop to 1.15×, still profitable but far from the headline ₹8,000 average winner.

SEBI guidelines require registered investment advisors and research analysts to disclose past performance, but many social media tipsters and unregistered services make claims without evidence. Traders should demand verifiable records showing every signal over a continuous period of at least three months, including dates, stocks, entry prices, outcomes, and costs.

Common mistakes when using intraday signals

Ignoring the stop-loss represents the most damaging mistake. Traders receive a signal with a defined stop, watch the stock move against the entry, and decide to hold because they believe price will recover. This behaviour converts a calculated ₹2,000 risk into an open-ended ₹10,000 or ₹20,000 loss when the stock continues falling. The purpose of a stop-loss is to accept small losses so that capital remains available for the next signal. Overriding the stop defeats the entire risk management framework.

Oversizing positions creates a related problem. A trader with ₹2,00,000 capital might risk ₹10,000 (5%) on a single intraday signal. A string of three or four losing trades, which happens regularly even in profitable systems, reduces capital by 15-20%. The psychological pressure becomes intense, leading to panicked exits on subsequent trades or desperate attempts to recover losses by doubling position sizes. Proper position sizing limits risk on each trade to 1-2% of capital, ensuring that a normal losing streak does not impair trading ability.

Cherry-picking signals sounds appealing but usually backfires. A trader receives five signals in a morning session and decides to take only the two that look strongest based on personal chart reading. This approach introduces subjective judgment that the AI system has not been tested on. If the trader's judgment were superior to the AI's, they would not need the signals in the first place. The system's edge, if it exists, comes from the aggregate performance of all signals, not from a hand-picked subset.

Modifying the target or stop-loss levels similarly breaks the system. A trader sees the target approaching and decides to hold for extra gain, then watches price reverse and the profit disappear. Or the stock hits the target in the first hour, but the trader adjusts the target higher, believing the momentum will continue. These modifications discard the tested risk-reward ratio. Performance measurement assumes consistent application of the published levels. Ad hoc adjustments mean the trader is running a different, untested strategy.

Trading without verifying the chart creates execution errors. Signals can contain data errors, or market conditions may have changed in the seconds between signal generation and order entry. Spending 30 seconds to confirm the stock is trading near the entry level, volume is adequate, and no major news event has occurred prevents costly mistakes. This verification does not mean second-guessing the signal's direction, only confirming that the setup still exists.

What role does risk-reward play in profitability?

The risk-reward ratio determines the minimum win rate required for a trading system to be profitable. At 1:1 risk-reward, half of all trades must win just to break even before costs, and more than half must win to be profitable after brokerage and taxes. This threshold is difficult to achieve consistently because market noise and false breakouts cause many trades to fail regardless of the quality of the setup.

A 1:2 risk-reward ratio shifts the mathematics in the trader's favour. The breakeven win rate drops to 33% before costs. A system that wins 40% of its trades generates a meaningful profit even after accounting for brokerage, STT, and slippage. This cushion matters during inevitable losing streaks. A system running at 1:1 that encounters ten trades where only four win loses capital. The same ten trades at 1:2, with four winners and six losers, produces a small net gain.

Wider risk-reward ratios such as 1:3 lower the breakeven win rate further (25%) but reduce the probability of each individual trade reaching its target. Targets placed far from the entry level are hit less often because price must move through more potential reversal points. The optimal ratio balances the frequency of winners against the size of gains. Many institutional traders use 1:1.5 to 1:2 as a practical middle ground for intraday setups.

The risk-reward ratio also defines position sizing. If a trader decides to risk ₹2,000 per trade and the signal shows ₹10 distance from entry to stop-loss, the appropriate position size is 200 shares (₹2,000 / ₹10). At 1:2 risk-reward, the target delivers ₹4,000 per winning trade. A 40% win rate across ten trades yields four winners (₹16,000 total) and six losers (₹12,000 total), netting ₹4,000 before costs. This systematic approach removes guesswork from position sizing and ensures consistent risk across different signals.

What should traders verify before executing a signal?

Chart confirmation forms the first verification step. Open the stock's chart on a trading platform and confirm the current price is near the signal's entry level. If the signal shows entry at ₹500 but the stock is already trading at ₹510, the setup may have already played out. Check that the visual pattern (trend, support/resistance levels) aligns with the signal's direction. A bullish signal in a stock that is clearly in a downtrend on the 15-minute or hourly chart warrants extra caution.

News and corporate announcements require checking. Visit the NSE website or the stock's page on the exchange portal and review any announcements from the past 24 hours. An unexpected earnings release, management change, or regulatory action can invalidate a technical signal. A bullish signal on a stock that just announced a large share dilution or missed earnings estimates may face selling pressure that overwhelms the technical setup.

Liquidity verification ensures the stock can be traded at the signal's levels. Check the order book for spread width. A stock with a ₹2 spread between bid and ask prices gives away ₹2 per share before the trade even begins, eroding the risk-reward ratio. Confirm that the 5-day average volume is at least several lakhs of shares. Signals in stocks averaging 50,000 shares per day carry high execution risk, as a single large order can move the price sharply.

Position sizing comes next. Calculate the number of shares based on personal risk tolerance and the distance from entry to stop-loss. Use the formula: shares = (amount willing to risk) / (entry price - stop-loss price). If planning to risk ₹3,000 and the signal shows entry ₹1,000 with stop ₹980, the position size is 150 shares (₹3,000 / ₹20). Verify this quantity fits within margin limits and does not exceed position concentration rules.

Order type selection matters for accurate execution. Market orders guarantee a fill but may cause slippage, especially in volatile markets. Limit orders at the signal's entry level guarantee price but risk missing the trade if price moves before the order fills. Many traders use limit orders for entry, accepting that they may miss a small percentage of signals, in exchange for avoiding poor entry prices that reduce the risk-reward ratio.