Zakamulin 2016
“Revisiting the Profitability of Market Timing with Moving Averages” (SSRN 2743119, March 2016 revision) by Valeriy Zakamulin of the University of Agder. The paper re-runs Glabadanidis’s (2015) widely downloaded study, which had reported “striking” performance for moving-average market-timing strategies, and demonstrates that the reported edge is entirely an artefact of look-ahead bias in the trading simulation. Re-simulating the same strategies on the same data without the bias, Zakamulin finds the timing strategy is at best marginally better than buy-and-hold on the Sharpe ratio and worse on alpha — statistically indistinguishable from passive holding — and supplies R code for full replication. It appears in this vault as the cleanest concrete demonstration that look-ahead / smoothed-label bias fabricates apparent regime-timing alpha, directly supporting Lookahead Bias from Smoothed Regime Estimates.
Connections
- Regime Classification — contradicts, source: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2743119
- Lookahead Bias from Smoothed Regime Estimates — supports, source: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2743119
- Valeriy Zakamulin — proposes_model, source: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2743119
- Buy-and-Hold Benchmark — compares_benchmark, source: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2743119