Memories, heteroscedasticity, and price limit in Currency futures markets

Testing for long-term memory in yen/dollar exchange rate

“…This is in contrast to the result reported by Booth, Kaen and Koveos (1982) although that study does not include yen data. Furthermore, this result is similar to that reported by Kao and Ma (1992) for currency futures price change series.…”

“…Ambrose, Ancel and Griffiths (1993) apply Lo's modified R/S technique and show that there is support for the hypothesis that stock market returns follow a random walk. Kao and Ma (1992) also apply Lo's modified technique and find significant short-term and not long-term dependence in currency future series. A brief description of Lo's modified R/S technique now follows.…”

Examining the dependency in intra‐day stock index futures

“…The results indicate that for the two contracts with different trading days toward maturity, the RR and MRR statistics are less than the critical value of 1.862, a finding indicating no long memory. The result of no long memory is consistent with two recent studies using MRR procedure in interest rate and currency futures markets [Fung and Lo (1993); Kao and Ma (1992)l. However, Helms, Kaen and Rosenman ( 1984) found long-memory in the agricultural futures prices.…”

“…Using the modified rescaled range (MRR) in analyzing stock market returns, he did not find the long-range memory found in earlier studies employing RR [e.g., see Greene and Fielitz (1977)l. Using MRR, Fund and Lo (1993) and Kao and Ma (1992) did not find long-memory in interest rate and currency futures prices changes.…”

Heterogeneous trading strategies with adaptive fuzzy Actor–Critic reinforcement learning: A behavioral approach

“…Katz (1992) indicates that an optimal number of hidden neurons can be found between one-half to three times the number of inputs, whereas Ersoy (1990) proposes doubling the number of neurons until the network's RMSE performance deteriorates. 6 Levich and Thomas (1993) and Kao and Ma (1992) found that hyperbolic sigmoid and logistic transfer functions are appropriate for financial markets data because they are nonlinear and continuously differentiable which are desirable properties for network learning. 7 The validation error starts decreasing until the network begins to overfit the data and the error will then begins to rise.…”