UMIE2005
This year’s convention was a joint bid with U-Mart2005, usually organized in autumn.
From this convention, all tools were unified to U-Mart system version 2, with redevelopments of the batch processing system for acceleration experiment, and the tool for Pareto ranking. In total six teams, three teams each from the science and engineering faculties and from the economics science faculties, and 30 agents participated in this convention. The winner was the agent which used the classical technical assay, and the agent who achieved good results in UMIE2005 also got better grades in U-Mart2005.
No agent was bankrupted from Ex1 through Ex3. But it seems as if the advance tuning was not sufficient, since agents who were top in all items of Ex1 were at 3, and agents who were in first place according to the Pareto rank were at 11. Additionally, one of the convention characteristics is to have few participants of AI-related agents; those were active until last year.
After the convention, there was a report on a new plan for the next fiscal year. A continuous session version of U-Mart now under development, and a new league plan, with the theme of “Market maker roles in thin board market”, were announced.
|
･Date |
12, September, 2005 | |
|
･Place |
Kyoto University (Event of Social Infomatic Fair) |
|
|
･Participants |
6 teams with 30 agents | |
|
･Agent type |
Machine agent |
【Machine】
| ID | Team | University | Agent name | Description |
| T01 | OCU-Nakajima | Osaka City University | MarketMakerStrategy_MM1 |
・There are five agents in total, i.e., three market maker agents, and two random agents, as for comparative experiments. ・Liquidity is provided to the market by sending an order for one selling and one buying in each session. The spread is adjusted by judging the phase direction upward or downward through observation of its own position. |
| T02 | Team titeCHuo | Tokyo Institute of Technology・ Chuo University |
Trend_fuzzyStrategy Trend_swiftStrategy |
・Two other agents, developed by three researchers, were entered. One agent enhanced the management ability position from the Fuzzy agent and is active for a couple of years. The other agent has a bigger position through catching the bigger trend, it erases the position when the trend is over. |
| T03 | Kyoto University, Pocket Seminar Kita 2005 |
Kyoto University | A0027279aStrategy A0027544kStrategy A0027757aStrategy |
・Other agents were developed within the Kyoto University training course and were implemented by three independent students. The three agents were: firstly, an agent that does contrary trade having its own indicator by means of the past price divergence or repurchase divergence; secondly, an agent which conducts pseudo arbitrage transactions, and finally an agent which, applying its own calculation method, combines both contrary trade methods by divergence and market following factors with short-term trends. |
| T04 | Kinki University | Kinki University | TestStrategy TestStrategy2 TestStrategy3 |
・An agent implemented by Kinki University. ・It does buying and selling taking advantage of moving average conversion premiums for future prices of short duration when its band exhibits more volatility. Usually it judges buying and selling by using bigger price conversion premiums, comparing both the moving average conversion premium for short duration spot prices, and the moving average conversion premium for future prices. Three agents were entered which had different thresholds for judging moving average conversion premiums. |
| T05 | Hobomegane | Shizuoka University | FireModoki3Strategy FireModoki4Strategy FireModoki5Strategy FirstLogisticStrategy FMACDStrategy FMACDStrategy0 FMACDStrategy2 IMACDStrategy MACDStrategy MACDStrategy02 Psycho4Strategy SecondLogisticStrategy |
・New entry team from Shizuoka University. ・Three agents entered: an agent that encodes the agent’s action, future price and spot price, then does matching of the action and input by means of the bucket-brigade algorithm, and an improved version agent; and two agents that judge buying and selling approximating the future price movement by logistic curve, and an improved version agent. There were six agents using MACD, i.e., the classical technical assay. And, this is an agent that employed the psychological line. These twelve agents, having various measures, all entered. |
| T06 | Chuo University | Chuo University | Murakami_spreadStrategy Nakamura_spreadStrategy Osako_pivotStrategy Shinohara_WmaRsiStrategy Tabira_bolingerStrategy |
・Five students from Chuo University implemented five independent agents. ・These five agents implemented and entered were as follows: an agent that sends an order of quantity that follows a stepwise increase according to the conversion premium of cash goods and future goods; a similar type of agent that sends an arbitrage order considering repurchase spread, together with an order preparing for bulge and collapse; an agent that does buying and selling using the pivot; an agent that sends an order by combining both WMA and RSI; and lastly an agent that trades using a classical technical agent “Bollinger band” but also implements an independent strategy. |
●Pareto-ranking
| Rank | Agent name | ID | Team | University |
| No.1(No.1 in Ex1,Ex2 and EX3) | Nakamura_spreadStrategy | T06 | Chuo University | Chuo University |
| No.1(No.1 in Ex1,Ex2 and EX3) | Trend_swiftStrategy | T02 | Team titeCHuo | Tokyo Institute of Technology・Chuo University |
| No.1(No.1 in Ex1,Ex2 and EX3) | Osako_pivotStrategy | T06 | Chuo University | Chuo University |
| No.4(No.1 in Ex1 and EX2, No.4 in Ex3) | Shinohara_WmaRsiStrategy | T06 | Chuo University | Chuo University |
| No.5(No.1 in Ex1, No.5 in Ex2, No4 in EX3) | Trend_fuzzyStrategy | T02 | Team titeCHuo | Tokyo Institute of Techonology・Chuo University |
| No.5(No.1 in Ex1, No.5 in Ex2, No.4 in EX3) | A0027544kStrategy | T03 | Kyoto University, Pocket Seminar Kita 2005 | Kyoto University |
| No.5(No.1 in Ex1, No.5 in Ex2, No.4 in EX3) | Murakami_spreadStrategy | T06 | Chuo University | Chuo University |
○All Rank(pdf)
