WIG20 Index
The WIG20 (“Warszawski Indeks Giełdowy 20”) is the principal stock index of the Warsaw Stock Exchange (GPW), a capitalisation-weighted index of the 20 largest and most liquid Polish companies, established 16 April 1994, with a rule capping any single sector at five constituents. It is tradeable as a cash index, through GPW WIG20 index futures (the exchange’s first derivative, introduced in 1998; current series multiplier PLN 20 per index point), WIG20 index options, and WIG20 ETFs. Futures trade in GPW’s continuous system 08:45-17:05 CET. Critically, the WIG20 is a smaller, shallower market than the developed-equity indices: although the WSE is the largest exchange in Central and Eastern Europe, FTSE classifies Poland as an Advanced Emerging market, and foreign investors accounted for roughly 67% of Main Market equity turnover in the first half of 2024 — so liquidity is real but foreign-flow-dependent, and effective trading costs are higher than on the S&P 500 or DAX.
In this vault the WIG20 is one of the two test instruments in Wilinski 2019, a peer-reviewed Expert Systems with Applications paper that applied a heterogeneous (non-homogeneous) Markov Chain Trading Model — re-estimating its transition matrix over rolling windows — to 4,000 daily WIG20 candles ending 28 April 2017. Wiliński reports “good results of profit according to the Calmar criterion” for both first- and second-order chains. Of the four discrete-Markov price-prediction papers tracked here it is the only one in a peer-reviewed journal and the only one reporting actual trading profit, which is why it warrants the closest scrutiny.
But the profit claim does not survive grading, and the WIG20’s market structure sharpens the problem. Three hyper-parameters — window length, window count, interval count — were tuned by machine learning to maximise predictive efficiency on the simulation data, so the reported Calmar ratio is an in-sample / data-snooped figure exposed to overfitting; the paper discloses no transaction costs, no slippage, no buy-and-hold or random-walk benchmark, no Sharpe and no replication package. The omission of costs is decisive here in particular: on a smaller emerging-market index like the WIG20, spreads are wider and depth is lower than on developed-market benchmarks, so a frequently-trading Markov-chain strategy faces a higher cost hurdle than the same strategy on the S&P 500 — exactly the cost that Wiliński’s “good Calmar” figure never pays. The WIG20 thus appears in this vault as an emerging-market equity testbed where a peer-reviewed Markov-chain profit claim looks favourable on paper but rests on a parameter-optimised, cost-free, benchmark-free in-sample simulation in a market where real costs would bite hardest.
Wilinski 2019 [tests_strategy] Markov Chain Trading Model WIG20 Index [relates] Transaction Costs and Slippage Transaction Costs and Slippage [contradicts] Wilinski 2019
Connections
- Markov Chain Trading Model — trades_market, source: https://www.sciencedirect.com/science/article/abs/pii/S0957417419303033
- Wilinski 2019 — tests_strategy (heterogeneous rolling-window chain, 4,000 daily candles), source: https://www.sciencedirect.com/science/article/abs/pii/S0957417419303033
- Transaction Costs and Slippage — relates (higher emerging-market costs undercut the profit claim), source: https://www.gpw.pl/contract-specifications-trading-rules