Learn about signal generation and rules

6.1.4 · Stock-Market › Algorithmic & Quant Trading

Overview

Signal generation ek systematic process hai jisme ye identify kiya jaata hai ki trade mein kab enter ya exit karna hai, quantifiable market conditions ke basis par. Ek trading signal ek trigger hota hai jo ek algorithm ko "BUY," "SELL," ya "HOLD" batata hai — predefined rules ke basis par jo market data par apply hote hain.

Ye kyun important hai: Signals ke bina, tum sirf guess kar rahe ho. Signals intuition ko testable, repeatable logic mein convert karte hain. Ye kisi bhi algorithmic trading system ka decision engine hain.

Kya kaam karte hain signals: Continuous market data ko discrete decisions mein convert karte hain. Kaise kaam karte hain: Mathematical/logical rules ko data streams par apply karte hain, actionable triggers output karte hain. Kyun chahiye rules: Discretion khatam karne ke liye, backtesting enable karne ke liye, aur consistency ensure karne ke liye.

Signal Types & Generation Logic

1. Trend-Following Signals

Purpose: Sustained price movements capture karna.

Example Rule: Moving Average Crossover

• Long Signal: Jab fast MA, slow MA ke upar cross kare
• Short Signal: Jab fast MA, slow MA ke neeche cross kare

Derivation from First Principles:

Moving average price noise ko smooth karta hai: $M{A}_n(t)=\frac{1}{n}{\sum }_{i=0}^{n-1}P(t-i)$

Crossover momentum shift detect karta hai. Jab $M{A}_{50}(t)>M{A}_{200}(t)$ ho lekin $M{A}_{50}(t-1)\le M{A}_{200}(t-1)$, toh buyers ne recently sellers ko overwhelm kiya hai (fast MA tezi se utha).

Signal Generation Rule (Long, +1):

1 & \text{if } MA_{50}(t) > MA_{200}(t) \land MA_{50}(t-1) \leq MA_{200}(t-1) \\ 0 & \text{otherwise} \end{cases}$$ **Signal Generation Rule (Short, −1):** $$S_{\text{short}}(t) = \begin{cases} -1 & \text{if } MA_{50}(t) < MA_{200}(t) \land MA_{50}(t-1) \geq MA_{200}(t-1) \\ 0 & \text{otherwise} \end{cases}$$ Full signal hai $S(t) = S_{\text{long}}(t) + S_{\text{short}}(t) \in \{-1, 0, +1\}$ (dono conditions mutually exclusive hain, isliye unka sum well-defined hai). **Ye step kyun?** Humein DONO conditions chahiye: current crossover (pehli inequality) AUR ye naya hai (doosri inequality check karti hai ki pehle cross nahi hua tha). Doosri condition ke bina, crossover ke baad har bar signal generate hota. Short rule, long rule ka mirror hai reversed inequalities ke saath — ek downward crossover. > [!example] Worked Example: MA Crossover > **Diya hua:** Stock XYZ, 50-day MA = ₹520, 200-day MA = ₹518 aaj. Kal, 50-day MA = ₹517, 200-day MA = ₹518. **Step 1:** Current condition check karo $520 > 518$ ✓ (fast MA, slow MA ke upar) **Step 2:** Previous condition check karo $517 \not> 518$ ✓ (fast MA kal slow MA ke upar nahi tha) **Step 3:** Signal evaluate karo Dono conditions meet hui → **Long Signal = 1** (BUY) **Ye step kyun?** Kal ki comparison confirm karti hai ki ye ek *fresh* crossover hai, continuation nahi. ### 2. **Mean-Reversion Signals** **Purpose:** Overextended price moves exploit karna jo wapas snap back karte hain. **Example Rule:** Bollinger Band Reversal - **Long Signal:** Price lower band touch kare (oversold) - **Short Signal:** Price upper band touch kare (overbought) **Derivation:** Bollinger Bands mean se standard deviation measure karte hain: $$\text{Upper Band} = MA_n(t) + k \cdot \sigma_n(t)$$ $$\text{Lower Band} = MA_n(t) - k \cdot \sigma_n(t)$$ jahan $\sigma_n(t) = \sqrt{\frac{1}{n}\sum_{i=0}^{n-1}(P(t-i) - MA_n(t))^2}$ **$k \cdot \sigma$ kyun?** Chebyshev's inequality ke mutabik, kisi bhi distribution ke liye, $\geq 75\%$ data 2 standard deviations ke andar hota hai. Jab price is range se bahar nikle, toh statistically rare hai → reversion likely hai. **Signal Generation Rule (Long, +1):** $$S_{\text{long}}(t) = \begin{cases} 1 & \text{if } P(t) \leq \text{Lower Band}(t) \\ 0 & \text{otherwise} \end{cases}$$ **Signal Generation Rule (Short, −1):** $$S_{\text{short}}(t) = \begin{cases} -1 & \text{if } P(t) \geq \text{Upper Band}(t) \\ 0 & \text{otherwise} \end{cases}$$ Full signal hai $S(t) = S_{\text{long}}(t) + S_{\text{short}}(t) \in \{-1, 0, +1\}$ (price ek saath lower band ke neeche aur upper band ke upar nahi ho sakta, isliye conditions mutually exclusive hain). **Short rule kyun?** Jab price **upper** band pierce kare toh statistically overbought hota hai (+2σ se upar), isliye hum reversion *neeche* ki taraf bet karte hain — exactly oversold long entry ka mirror. > [!example] Worked Example: Bollinger Reversion > **Diya hua:** 20-day MA = ₹1000, σ = ₹20, k = 2. Current price = ₹955. **Step 1:** Bands calculate karo - Lower Band = $1000 - 2(20) = 960$ - Upper Band = $1000 + 2(20) = 1040$ **Step 2:** Price compare karo $955 < 960$ ✓ (price lower band ke neeche) **Step 3:** Signal generate karo **Long Signal = 1** (BUY — upar reversion expect karo) **Ye step kyun?** ₹955 par price statistically oversold hai (2σ ke neeche). Reversion bet ye hai ki ye ₹1000 ki taraf wapas jaata hai. (Agar instead price ₹1045 hoti, toh $1045 \geq 1040$ → **Short Signal = −1**, reversion neeche ki taraf bet karo.) ### 3. **Momentum Signals** **Purpose:** Price direction mein acceleration par ride karna. **Example Rule:** RSI Thresholds - **Long Signal:** RSI 30 ke upar cross kare (oversold se bahar nikle) - **Short Signal:** RSI 70 ke neeche cross kare (overbought se bahar nikle) **Derivation:** RSI, $n$ periods mein average gains ko average losses se compare karta hai: $$RS = \frac{\text{Avg Gain}_n}{\text{Avg Loss}_n}$$ $$RSI = 100 - \frac{100}{1 + RS}$$ **Ye formula kyun?** Gain/loss ratio ko 0-100 scale par normalize karta hai. Jab $RS \to \infty$ (saare gains), $RSI \to 100$. Jab $RS \to 0$ (saare losses), $RSI \to 0$. **Signal Generation Rule (Long, +1):** $$S_{\text{long}}(t) = \begin{cases} 1 & \text{if } RSI(t) > 30 \land RSI(t-1) \leq 30 \\ 0 & \text{otherwise} \end{cases}$$ **Signal Generation Rule (Short, −1):** $$S_{\text{short}}(t) = \begin{cases} -1 & \text{if } RSI(t) < 70 \land RSI(t-1) \geq 70 \\ 0 & \text{otherwise} \end{cases}$$ Full signal hai $S(t) = S_{\text{long}}(t) + S_{\text{short}}(t) \in \{-1, 0, +1\}$. > [!example] Worked Example: RSI Crossover > **Diya hua:** 14-day RSI = 32 aaj, RSI = 28 kal. **Step 1:** Threshold breach check karo $32 > 30$ ✓ (oversold level ke upar) **Step 2:** Check karo ki ye fresh breach hai $28 \leq 30$ ✓ (kal oversold tha) **Step 3:** Signal generate karo **Long Signal = 1** (momentum upar turn ho raha hai) **Ye step kyun?** Hum *crossing* event chahte hain, sirf 30 ke upar hona nahi. Ye ensure karta hai ki hum tab enter karein jab momentum shift hota hai, na ki jab already motion mein ho. ## Rule Design Framework > [!formula] Signal Rule Components > Har signal rule ke teen parts hote hain: 1. **Indicator Calculation:** $I(t) = f(\text{Price}, \text{Volume}, \ldots)$ 2. **Threshold/Condition:** $C(t) = \text{boolean expression on } I(t)$ 3. **State Transition:** $S(t) = g(C(t), C(t-1), \ldots, S(t-1))$ **Example:** MA Crossover 1. $I_1(t) = MA_{50}(t)$, $I_2(t) = MA_{200}(t)$ 2. $C(t) = [I_1(t) > I_2(t)]$ 3. $S(t) = C(t) \land \neg C(t-1)$ (aaj true, kal false) ### **Ye structure kyun?** - **Indicator:** Noisy price data ko interpretable metric mein compress karta hai - **Condition:** Indicator ko binary decision boundary mein translate karta hai - **State Transition:** Duplicate signals prevent karta hai (backtesting ke liye key) > [!example] Worked Example: Volume Breakout Signal > **Rule:** Buy karo jab price 20-day high break kare AUR volume > 1.5× average volume ho. **Step 1:** Indicators calculate karo - $I_1(t) = P(t)$ - $I_2(t) = \max(P(t-1), \ldots, P(t-20))$ (20-day high) - $I_3(t) = V(t)$ (aaj ka volume) - $I_4(t) = \frac{1}{20}\sum_{i=1}^{20}V(t-i)$ (average volume) **Step 2:** Conditions define karo - $C_1(t) = [P(t) > I_2(t)]$ (breakout) - $C_2(t) = [V(t) > 1.5 \cdot I_4(t)]$ (volume surge) **Step 3:** State transition - $S(t) = C_1(t) \land C_2(t) \land \neg S(t-1)$ (dono true, pehle signal nahi tha) **Diya hua Data:** - Aaj: Price = ₹550, Volume = 2M shares - 20-day high = ₹548 - 20-day avg volume = 1.2M shares **Evaluation:** - $C_1 = [550 > 548] = \text{True}$ - $C_2 = [2M > 1.5 \times 1.2M = 1.8M] = \text{True}$ - Assume karo $S(t-1) = 0$ (pehle koi signal nahi) - $S(t) = \text{True} \land \text{True} \land \text{True} = 1$ → **Long Signal** **Ye step kyun?** Volume confirmation false breakouts filter karta hai. Price akele low volume par spike kar sakta hai (weak move). High volume = institutional participation = zyada reliable. ## Common Mistakes & Fixes > [!mistake] Mistake 1: Signal Redundancy > **Galat Idea:** Har bar "BUY" signal generate karo jab fast MA > slow MA ho. **Ye sahi kyun lagta hai:** Tum trade mein rehna chahte ho, isliye signal repeat karna safe lagta hai. **Fix:** Signals *state changes* trigger karne chahiye, states confirm karne nahi. Ek alag position tracker use karo. Signal sirf crossover par generate karo, phir opposite signal aane tak position hold karo. **Steel-man:** Confusion isliye hoti hai kyunki "signal" aur "position" ko ek saath mila diya jaata hai. Signal ek *event* hai, position ek *state* hai. Events ek baar hote hain; states persist karte hain. ```python # WRONG: Signal every bar if fast_ma > slow_ma: signal = 1 # Fires repeatedly! # RIGHT: Signal on transition if fast_ma > slow_ma and fast_ma_prev <= slow_ma_prev: signal = 1 # Fires once at crossover elif fast_ma < slow_ma and fast_ma_prev >= slow_ma_prev: signal = -1 else: signal = 0 ``` > [!mistake] Mistake 2: Look-Ahead Bias > **Galat Idea:** Aaj ke close se aaj ka MA calculate karo, phir "aaj" signal generate karo. **Ye sahi kyun lagta hai:** Tumhare paas aaj ka saara data hai, toh use kyun na karo? **Fix:** Time $t$ par signals sirf woh data use kar sakte hain jo $t$ se *pehle available* tha. Agar tum close par trade karte ho, toh tumhara signal $t-1$ tak ke data use karna chahiye. **Steel-man:** Backtesting mein, saara data ek saath exist karta hai, isliye accidentally future mein "jhankna" aasaan hai. Live trading mein ye impossible hai — jab tak market band na ho, tum aaj ka close nahi jaante. **Example:** ```python # WRONG: Calculate MA including today, signal today ma = prices[:today+1].mean() # Uses today's close! if price[today] > ma: signal = 1 # RIGHT: Calculate MA up to yesterday, signal today ma = prices[:today].mean() # Excludes today if price[today] > ma: signal = 1 # Now valid ``` > [!mistake] Mistake 3: Over-Optimization (Curve Fitting) > **Galat Idea:** Parameters (MA periods, RSI thresholds) tune karo jab tak backtest returns maximize na ho jaayein. **Ye sahi kyun lagta hai:** Higher returns = better strategy, hai na? **Fix:** Optimized parameters jo historical data ko perfectly fit karte hain, naye data par fail honge (overfitting). Train/test splits, cross-validation, ya out-of-sample testing use karo. Wide parameter tolerance wale robust rules prefer karo. **Steel-man:** Optimization logical lagti hai — hum best parameters chahte hain. Lekin "past data par best" ≠ "future data par best." Tum signal nahi, noise fit kar rahe ho. **Example:** - Strategy A: 50/200 MA crossover → train par 12% return, test par 11% - Strategy B: 73/183 MA crossover → train par 18% return, test par 3% - A choose karo (robust), B nahi (overfit) ## Advanced: Multi-Condition Signals Real strategies signals combine karti hain: $$S_{\text{final}}(t) = \begin{cases} 1 & \text{if } \sum_{i=1}^n w_i S_i(t) > \theta \\ -1 & \text{if } \sum_{i=1}^n w_i S_i(t) < -\theta \\ 0 & \text{otherwise} \end{cases}$$ jahan har $S_i(t)$, $i$-ve individual signal ka output hai, $w_i$ weights hain (jaise, trend ke liye 0.4, volume ke liye 0.3, momentum ke liye 0.3), aur $\theta$ ek threshold hai. **Weighted kyun?** Saare signals equally reliable nahi hote. Trend signals, trending markets mein reversion se stronger ho sakte hain. Weights is hierarchy ko encode karte hain. > [!example] Worked Example: Combined Signal > **Teen Signals Diye Gaye Hain:** > 1. MA crossover: $S_1(t) = 1$ (bullish) > 2. RSI: $S_2(t) = 0$ (neutral) > 3. Volume breakout: $S_3(t) = 1$ (bullish) **Weights:** $w_1 = 0.5$, $w_2 = 0.3$, $w_3 = 0.2$ **Threshold:** $\theta = 0.6$ **Calculation:** $$\sum_{i=1}^3 w_i S_i(t) = 0.5(1) + 0.3(0) + 0.2(1) = 0.5 + 0 + 0.2 = 0.7$$ **Evaluation:** $0.7 > 0.6$ → **Final Signal = 1** (BUY) **Ye step kyun?** Weighted sum evidence aggregate karta hai. Bhale hi RSI neutral hai, strong trend + volume confirmation use outweigh kar deta hai. > [!recall]- Ek 12-Saal ke Bacche ko Samjhao > Socho tum ek video game khel rahe ho jahan tum coins collect karte ho. Tum chahte ho ki ek helper bot tumhe bataye kab jump karna hai. Tum use rules dete ho: **Rule 1:** "Jump karo jab coin tumse neeche ho" (neeche se kharido). **Rule 2:** "Jump karo jab tum 3 coins ek line mein upar jaate dekho" (trend). Bot game dekhta hai, aur jab rules true hote hain, toh chillata hai "JUMP!" Woh chillaana ek **signal** hai. Stock trading mein, bot prices dekhta hai. Jab rules meet hote hain (price ek line cross kare, volume spike kare), toh chillata hai "BUY" ya "SELL." Rules **signal generation rules** hain — ye boring numbers ko action commands mein badal dete hain. Tricky part: bure rules bot ko galat samay par jump karwate hain (lava mein!). Acche rules testing se aate hain: "Kya is rule ne mujhe pichle levels mein zyada coins collect karne mein madad ki?" Agar haan, rakh lo. Agar nahi, fix karo. > [!mnemonic] TRIPS for Signal Rules > **T**hreshold: Kaunsa level signal trigger karta hai? > **R**ule logic: AND/OR conditions > **I**ndicator: Tum kaunsa data use kar rahe ho? > **P**rior state: Duplicates avoid karne ke liye kal check karo > **S**tateful: Track karo ki position mein pehle se ho ya nahi ## Practical Implementation Checklist 1. **Objective Define karo:** Trend-following vs. mean-reversion? Time horizon kya hai? 2. **Indicators Select karo:** Market regime ke basis par choose karo (trending: MA, ranging: BB) 3. **Thresholds Set karo:** Historical volatility use karo, arbitrary numbers nahi 4. **Filters Add karo:** Volume, time-of-day, market cap constraints 5. **State Management:** Positions track karo, re-entry signals avoid karo 6. **Backtest karo:** Train/test split, walk-forward analysis 7. **Risk Rules:** Max drawdown, position sizing, stop-loss integration ## Connections - [[Moving averages and crossovers]] — Trend signals ki foundation - [[RSI and momentum indicators]] — Core momentum signal inputs - [[Bollinger Bands and volatility]] — Mean-reversion signal framework - [[Backtesting fundamentals]] — Signal rules ko historically validate kaise karein - [[Risk management in algo trading]] — Signals ko position sizing ke saath integrate karna - [[Market regimes and adaptivity]] — Signal types kab switch karein - [[Order execution and slippage]] — Signals ko actual trades mein translate karna --- #flashcards/stock-market Trading signal kya hota hai? :: Ek discrete output (BUY/SELL/HOLD) jo tab generate hota hai jab market data predefined rules satisfy kare, continuous price streams ko actionable decisions mein convert karta hai. Kisi bhi signal rule ke teen components kya hote hain? ::: (1) Indicator calculation $I(t)$, (2) Threshold/condition $C(t)$, (3) State transition logic $S(t)$ jo prior state check karta hai. Signal rules mein "state transition" check kyun zaroori hai (jaise $\neg C(t-1)$)? ::: Ek condition meet hone ke baad har bar duplicate signals generate hone se bachne ke liye; hum sirf *transitions* (crossovers) trigger karna chahte hain, continuous states nahi. Signal generation mein look-ahead bias kya hai? ::: Time $t$ ya baad ka data use karke time $t$ "par" signal generate karna, jo live trading mein impossible hai (aaj ka close market band hone se pehle nahi pata). MA crossover long signal formula :: $S_{\text{long}}(t) = 1$ agar $MA_{\text{fast}}(t) > MA_{\text{slow}}(t)$ AUR $MA_{\text{fast}}(t-1) \leq MA_{\text{slow}}(t-1)$, warna 0. MA crossover short signal formula ::: $S_{\text{short}}(t) = -1$ agar $MA_{\text{fast}}(t) < MA_{\text{slow}}(t)$ AUR $MA_{\text{fast}}(t-1) \geq MA_{\text{slow}}(t-1)$, warna 0. Bollinger Band lower band formula ::: $\text{Lower Band} = MA_n(t) - k \cdot \sigma_n(t)$ jahan $\sigma_n$, $n$ periods par standard deviation hai. Bollinger short (overbought) signal ::: $S_{\text{short}}(t) = -1$ agar $P(t) \geq \text{Upper Band}(t)$, downward reversion par bet karte hue. RSI formula ::: $RSI = 100 - \frac{100}{1 + RS}$ jahan $RS = \frac{\text{Avg Gain}_n}{\text{Avg Loss}_n}$. RSI > 70 overbought kyun suggest karta hai? ::: High RSI matlab recent gains, losses se bahut zyada hain ($RS$ bada hai), jo indicate karta hai ki price tezi se badhi hai aur pullback ho sakta hai. Breakout signals mein volume confirmation ka kya purpose hai? ::: Volume false breakouts filter karta hai; high volume institutional participation aur price move ke peeche strong conviction indicate karta hai. Signal tuning mein curve fitting (over-optimization) ::: Parameters ko backtest returns maximize karne ke liye tune karna noise fit karta hai, signal nahi; strategy naye data par fail hogi kyunki ye historical quirks par overfit hai. Multiple signals kaise combine karte hain? ::: Weighted sum $\sum_i w_i S_i(t)$ use karo aur threshold $\theta$ se compare karo; weights ke zariye zyada reliable signals ko prioritize karna possible hai. Signal aur position mein kya fark hai? ::: Signal ek *event* (trigger) hai, position ek *state* (holding) hai. Signals transitions par ek baar fire hote hain; positions exit signal aane tak persist karti hain. Bollinger Bands mein standard deviation ($k\sigma$) kyun use karte hain? ::: Chebyshev's inequality ke mutabik, zyaatar data (~75%+) mean ke 2σ ke andar hota hai; isse exceed karna statistically rare hai, jo reversion suggest karta hai. ## 🖼️ Concept Map ```mermaid flowchart TD MD[Market Data] -->|input to| SG[Signal Generation] RULES[Predefined Rules] -->|applied by| SG SG -->|outputs| TS[Trading Signal] TS -->|encoded as| DISC["Discrete values -1,0,+1"] DISC -->|means| ACT[Buy Sell or Hold] SG -->|eliminates| DISCR[Emotional Discretion] DISCR -->|enables| BT[Backtesting and Consistency] RULES -->|example type| TF[Trend-Following Signals] TF -->|uses| MAC[MA Crossover] MAC -->|built from| MA["Moving Average smooths noise"] MAC -->|needs| DUAL[Current cross AND newly crossed] DUAL -->|prevents| REPEAT[Repeated signals each bar] ```