Backtest · PSX Alpha

AICLRs 31.0+9.1%BERGRs 36.5-14.4%PRLRs 147.7+44.2%AGLRs 16.9+69.1%KAPCORs 51.5+57.5%SITCRs 52.5-4.6%NMLRs 374.1+210.3%NETSOLRs 5,771.3+29.7%BOPRs 138.7-11.7%PSORs 443.1+105.3%NPLRs 18.7+89.3%FMLRs 288.6+94.9%MARIRs 23.5-11.0%THALLRs 140.6+110.3%NRLRs 61.3-22.3%HBLRs 451.9-6.7%SYSRs 1,566.1+36.2%GATMRs 79.9+33.4%GHGLRs 1,075.6+1.5%FCEPLRs 85.0+61.6%PNSCRs 566.1-42.5%AVNRs 463.8+57.3%FFLRs 135.5+33.3%AGTLRs 159.5-34.9%FEROZRs 274.0+45.3%SRVIRs 38.0+63.0%ILPRs 26.7+20.5%KELRs 71.8+64.8%MLCFRs 1.8-5.9%HTLRs 157.8+46.7%PIOCRs 54.6+58.7%IGIHLRs 98.5-4.8%EFERTRs 76.5+20.3%FFCRs 14.2-30.0%GHNIRs 476.1+34.4%PTCRs 14,713.1+1.5%TPLPRs 2,078.1+10.1%UBLRs 355.3-25.8%ISLRs 171.0+2.7%APLRs 38.3+9.0%ABOTRs 4.4-26.4%PKGSRs 61.2+35.3%FABLRs 263.0-1.9%ATLHRs 10.6-5.5%OGDCRs 71.0+43.4%MTLRs 106.9-27.7%NCLRs 5.7+56.9%UNITYRs 758.2+41.8%LUCKRs 67.7+22.9%KTMLRs 158.5+51.2%SNBLRs 172.8+16.1%BWCLRs 196.5+40.1%CHCCRs 25.0-4.0%ASTLRs 9.1+14.5%ITTEFAQRs 12.3-9.9%AGPRs 30.8-42.3%HINOONRs 61.3-35.9%ENGRORs 68.6+46.4%ATRLRs 26.4+81.9%SHELRs 187.1-5.4%JDWSRs 2.1+20.6%EFUGRs 50.9+44.1%KOHCRs 79.1+42.0%BNWMRs 54.3+62.8%GLAXORs 56.9+14.0%CNERGYRs 49.2-1.8%JSBLRs 377.8-24.1%SEARLRs 21.7-21.2%MCBRs 81.8-4.4%BAFLRs 110.1+14.1%TGLRs 236.2-8.1%NCPLRs 27.7+19.9%AIRLINKRs 81.3+23.7%PAELRs 27.5+77.4%PAKTRs 1.7+3.6%INDURs 74.3-34.6%EPCLRs 17.8-50.1%TRGRs 100.9+18.5%SNGPRs 3.6-7.2%MEBLRs 95.1+11.0%PPLRs 12.0+27.2%POLRs 149.8+69.1%BAHLRs 123.3-17.6%GWLCRs 18.0+9.8%ACPLRs 41.8+19.4%FCCLRs 146.4+6.4%HCARRs 752.9-10.6%NBPRs 173.6-16.9%FATIMARs 237.9+68.5%SSGCRs 12.9+40.7%AKBLRs 268.1-10.1%LOTCHEMRs 65.9-7.3%NATFRs 5.0-57.8%MUGHALRs 76.8-47.2%PSMCRs 514.2-32.5%ICIRs 268.1+5.8%HUBCRs 48.3+16.8%NESTLERs 4.6-50.6%ARPLRs 89.9-12.9%DGKCRs 681.9+39.9%PSELRs 113.5-0.4%AICLRs 31.0+9.1%BERGRs 36.5-14.4%PRLRs 147.7+44.2%AGLRs 16.9+69.1%KAPCORs 51.5+57.5%SITCRs 52.5-4.6%NMLRs 374.1+210.3%NETSOLRs 5,771.3+29.7%BOPRs 138.7-11.7%PSORs 443.1+105.3%NPLRs 18.7+89.3%FMLRs 288.6+94.9%MARIRs 23.5-11.0%THALLRs 140.6+110.3%NRLRs 61.3-22.3%HBLRs 451.9-6.7%SYSRs 1,566.1+36.2%GATMRs 79.9+33.4%GHGLRs 1,075.6+1.5%FCEPLRs 85.0+61.6%PNSCRs 566.1-42.5%AVNRs 463.8+57.3%FFLRs 135.5+33.3%AGTLRs 159.5-34.9%FEROZRs 274.0+45.3%SRVIRs 38.0+63.0%ILPRs 26.7+20.5%KELRs 71.8+64.8%MLCFRs 1.8-5.9%HTLRs 157.8+46.7%PIOCRs 54.6+58.7%IGIHLRs 98.5-4.8%EFERTRs 76.5+20.3%FFCRs 14.2-30.0%GHNIRs 476.1+34.4%PTCRs 14,713.1+1.5%TPLPRs 2,078.1+10.1%UBLRs 355.3-25.8%ISLRs 171.0+2.7%APLRs 38.3+9.0%ABOTRs 4.4-26.4%PKGSRs 61.2+35.3%FABLRs 263.0-1.9%ATLHRs 10.6-5.5%OGDCRs 71.0+43.4%MTLRs 106.9-27.7%NCLRs 5.7+56.9%UNITYRs 758.2+41.8%LUCKRs 67.7+22.9%KTMLRs 158.5+51.2%SNBLRs 172.8+16.1%BWCLRs 196.5+40.1%CHCCRs 25.0-4.0%ASTLRs 9.1+14.5%ITTEFAQRs 12.3-9.9%AGPRs 30.8-42.3%HINOONRs 61.3-35.9%ENGRORs 68.6+46.4%ATRLRs 26.4+81.9%SHELRs 187.1-5.4%JDWSRs 2.1+20.6%EFUGRs 50.9+44.1%KOHCRs 79.1+42.0%BNWMRs 54.3+62.8%GLAXORs 56.9+14.0%CNERGYRs 49.2-1.8%JSBLRs 377.8-24.1%SEARLRs 21.7-21.2%MCBRs 81.8-4.4%BAFLRs 110.1+14.1%TGLRs 236.2-8.1%NCPLRs 27.7+19.9%AIRLINKRs 81.3+23.7%PAELRs 27.5+77.4%PAKTRs 1.7+3.6%INDURs 74.3-34.6%EPCLRs 17.8-50.1%TRGRs 100.9+18.5%SNGPRs 3.6-7.2%MEBLRs 95.1+11.0%PPLRs 12.0+27.2%POLRs 149.8+69.1%BAHLRs 123.3-17.6%GWLCRs 18.0+9.8%ACPLRs 41.8+19.4%FCCLRs 146.4+6.4%HCARRs 752.9-10.6%NBPRs 173.6-16.9%FATIMARs 237.9+68.5%SSGCRs 12.9+40.7%AKBLRs 268.1-10.1%LOTCHEMRs 65.9-7.3%NATFRs 5.0-57.8%MUGHALRs 76.8-47.2%PSMCRs 514.2-32.5%ICIRs 268.1+5.8%HUBCRs 48.3+16.8%NESTLERs 4.6-50.6%ARPLRs 89.9-12.9%DGKCRs 681.9+39.9%PSELRs 113.5-0.4%

Walk-forward backtest

Positions at each rebalance were chosen only from filings already published and prices already printed. Costs of 50 basis points per round trip are charged on the traded fraction of the book, no sector may exceed 30% of the portfolio, and illiquid names are excluded however well they score.

Engine compound return

21.21%

vs 14.34% for the KSE-100

Years tested

9.5

38 quarterly rebalances from 1 July 2016

Calendar years ahead

7/10

Full calendar years where the engine beat the index

Cost drag paid

21.19%

Total commission and slippage charged across the test

Growth of Rs 100

Each strategy holds the top-ranked names under its own score. The benchmark is the cap-weighted KSE-100.

Growth of Rs 100 · click a legend entry to isolate

Every strategy, side by side

The rule-based score has no fitted parameters, so it cannot overfit. The machine-learning ranker is retrained at every rebalance on data that was available then. The blend is what the dashboard reports as the engine.

Measure	Blended engine	Rule-based score	ML ranker (XGBoost)	KSE-100 benchmark
Compound annual return	21.21%	19.48%	16.85%	14.34%
Total return	521.6%	442.2%	339.1%	257.2%
Annualised volatility	17.6%	17.3%	17.8%	16.1%
Sharpe ratio	0.56	0.46	0.31	0.18
Sortino ratio	0.94	0.77	0.51	0.29
Worst drawdown	-53.3%	-50.1%	-54.7%	-55.5%
Calmar ratio	0.40	0.39	0.31	0.26
Beta vs KSE-100	0.96	0.95	0.97	1.00
Annual alpha	6.98%	5.29%	2.60%	0.00%
Tracking error	8.3%	8.3%	8.5%	0.0%
Information ratio	0.83	0.62	0.29	0.00
Quarters ahead of the index	63%	63%	55%	—
Average turnover per rebalance	56%	31%	62%	—

Drawdown

How far the book sat below its own high-water mark, against the same measure for the index.

The engine is a long-only equity strategy, so it falls with the market. Its worst drawdown of -53.3% against the index's -55.5% is a difference of degree, not of kind.

Calendar year returns

Where the excess return actually came from, year by year.

The engine finished ahead of the KSE-100 in 7 of 10 calendar years. Losing years matter: a strategy that never lags the index over a decade is usually a sign of look-ahead leaking into the test.

Rebalance log

Every quarterly trade the blended engine made, with the book it held and how that book performed against the index over the following quarter.

BERGChemicals74STRONG BUY

TPLPTechnology & Communication73STRONG BUY

GATMTextile Composite72STRONG BUY

PRLRefinery71STRONG BUY

AVNTechnology & Communication70STRONG BUY

SITCChemicals69STRONG BUY

PTCTechnology & Communication69STRONG BUY

FFCFertilizer69STRONG BUY

FMLTextile Composite68STRONG BUY

NMLTextile Composite67STRONG BUY

MARIOil & Gas Exploration67STRONG BUY

GHNIAutomobile Assembler66BUY

FFLFood & Personal Care66BUY

AIRLINKTechnology & Communication65BUY

KAPCOPower Generation65BUY

101 names were scored; 3 were excluded for thin traded volume before selection. Equal weight, 6.7% each.

A backtest is evidence about a rule, not a forecast. It assumes every order filled at the closing price on the rebalance day, ignores the market impact of scaling the book, and is measured on a single simulated history of the KSE-100. Research and educational project. Not investment advice. Figures are produced by a point-in-time simulation of the KSE-100 built for reproducible backtesting.