Arbitrage in Taiwan Stock Index Futures and Application of Neural Networks
碩士 === 國立中山大學 === 資訊管理學系 === 87 === This study investigates the signals and the profitability of index arbitrage for Taiwan Stock Index Futures that trade on SIMEX and TAIMEX. The pricing model used in this study incorporates different transaction costs, different borrowing and lending rates, and se...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
1999
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/56098132707792085707 |
| Summary: | 碩士 === 國立中山大學 === 資訊管理學系 === 87 === This study investigates the signals and the profitability of index arbitrage for Taiwan Stock Index Futures that trade on SIMEX and TAIMEX. The pricing model used in this study incorporates different transaction costs, different borrowing and lending rates, and seasonal dividend payments. This study also constructs a neural network model to forecast the direction of the next five-minute''s basis movement and uses a quadratic programming model to select baskets of stocks which mirror the market index. The result is as follows:
1. The futures contracts are undervalued for the period of July to October 1998 and are overvalued for the period of November to December 1998.
2. Compared to futures contracts that trade on SIMEX, there is a larger number of arbitrage signals in futures contracts that trade on TAIMEX.
3. The market in SIMEX has gradually matured with time but the market in TAIMEX is not yet matured.
4. After considering the true value of futures contracts and the exchange rate, the arbitrage profit of arbitrageurs in SIMEX is larger than that in TAIMEX.
5. The neural network predicts the correct sign of the next five-minute''s basis change of a futures contract 80% of the time.
6. The mean square error and average absolute error of using the quadratic programming model to construct the surrogate stock baskets is about 0.0162% and 0.0112% when the average number of stocks is 42.33333.
|
|---|