Sabse gehra reason: ek backtest look-ahead bias aur survivorship bias ka shikaar ho sakta hai, jabki paper trading, apni construction ke hisaab se, aage nahin dekh sakti kyunki future literally abhi hua hi nahin hai.
Hume backtest performance ko paper performance se usi period par aage jaate hue compare karna hoga. Aao tracking metric ko scratch se derive karte hain.
Maano rtbt day t par backtest return hai aur rtpt usi din paper-traded return hai. Daily difference define karo:
dt=rtpt−rtbt
Subtract KYUN karte hain? Kyunki ek systematically negativedt matlab paper trading woh value bleed kar rahi hai jo backtest ne tumse kabhi charge nahin ki — usually slippage aur fees.
Har din average leakage:
dˉ=N1∑t=1Ndt
Tracking error us difference ka standard deviation hai — gap kitna unpredictable hai:
Maano tum Q shares khareedna chahte ho lekin demand ki har unit price ko k factor se upar push karti hai (impact coefficient). Quantity q par marginal share ki paid price P0(1+kq) hai. Total cost:
Socho tumne ek video-game move invent ki jo hamesha jeetti hai... jab tum purani recorded games replay karte ho. Cool — lekin replays tumhe surprise nahin kar sakti, tum pehle se jaante ho enemy kya karta hai. Paper trading matlab live game ke against abhi khelna hai, lekin ek save-file ke saath taaki asli coins na jaayein. Agar tumhari move live bhi jeetti hai, toh real hai. Agar sirf replays mein jeetti thi, toh tum apne real coins khone se bach gaye.
Paper trading kaun sa data aur kaun sa paisa use karti hai?
Live/streaming market data, simulated (nakli) capital ke saath.
Ek backtest jhooth KYUN bol sakta hai lekin paper trading structurally timing ke baare mein jhooth KYUN nahin bol sakti?
Ek backtest accidentally future data use kar sakta hai (look-ahead bias); paper trading tick-by-tick decide karti hai jab future genuinely unknown hota hai.
Backtest aur paper returns ke beech tracking error define karo.
Daily return differences dt=rtpt−rtbt ka standard deviation, yaani N−11∑(dt−dˉ)2.
Consistently negative mean difference dˉ kya indicate karta hai?
Systematic leakage — usually slippage aur fees jo backtest ne ignore kiye.
Linear price impact k ke saath derived slippage-per-share formula?
21kP0Q — order size Q mein linear.
Ek strategy jo 1 lot par kaam karti hai woh large size par KYUN fail ho jaati hai?
Slippage quantity ke saath linearly badhta hai (∝Q), toh scale par market impact edge kha jaata hai.
Validation funnel ka order?
Backtest → Paper trade → Live (chhoti size).
Backtest se perfect paper match ek warning sign KYUN hai?
Iska matlab hai paper engine bhi costless instant fills assume karta hai — usne backtest ki fantasy inherit kar li, realism add karne ki bajaye.
Paper results statistically meaningful hone se pehle roughly kitne trades chahiye?
Kam se kam ~30, kyunki mean ka standard error 1/n se shrink hota hai.
Do cheezein jo paper trading test karti hai jo ek backtest kabhi nahin kar sakta?
Real data-feed/latency behaviour aur trader ki apni psychological discipline.
