Challenging Japanese Stock Swing Trading with ChatGPT #54
I continue to engage in short-term swing trading, where I purchase Japanese stocks during my weekday lunch break and, in principle, sell them during the next business day’s lunch break. I use ChatGPT for stock selection and trading decisions, verifying with my own money whether it is actually possible to make a profit in Japanese stock trading using AI.
This article, excluding the final ‘Thoughts on this session’ section, was also created by ChatGPT. The main text is organized based on actual trading records and the analysis conducted at that time.
A loss of 10,100 yen on Mitsubishi Materials. Despite strong earnings, the stock fell the next day, bringing cumulative losses to over 210,000 yen.
In the trading session on Friday, November 13, I sold the Mitsubishi Materials shares I had purchased the previous day, locking in a loss of 10,100 yen.
Although I purchased the stock judging the upward revision of earnings forecasts as a positive factor, profit-taking selling became dominant the following day. Meanwhile, I newly purchased Bridgestone, and I am holding onto Kajima Corporation and Kioxia Holdings.
Why did Mitsubishi Materials fall despite strong earnings?
I purchased 100 shares of Mitsubishi Materials (5711) at 4,414 yen on November 12 and sold them at 4,313 yen on the 13th. The realized profit/loss is -10,100 yen (before taxes and fees).
The basis for the purchase was the company’s upward revision of its earnings forecast. The operating profit forecast was raised from 130 billion yen to 165 billion yen, and buying momentum was confirmed in the morning session as the stock price exceeded the VWAP (Volume Weighted Average Price).
However, it turned downward in the morning session on the 13th. In addition to the stock price rising about 6% the previous day, it is thought that the weak profit outlook for the second half of the year relative to the strong first-half performance, as well as caution regarding earnings dependent on special factors such as metal market conditions, acted as a burden.
Even if the material of ‘strong earnings’ is evaluated correctly, if the stock price has already priced in that material, it does not necessarily mean that the rise will continue until the next day.
Was ChatGPT’s previous day’s judgment correct?
ChatGPT’s judgment the previous day was not an unconditional ‘BUY’ for Mitsubishi Materials, but a ‘CONDITIONAL BUY’.
Specifically, it set the purchase condition as the 5-minute candle closing at 4,425 yen or higher. However, the actual execution price was 4,414 yen, so it cannot be confirmed whether the specified upward confirmation condition was met.
This loss shows that the evaluation of positive material did not guarantee a rise in the stock price the next day. At the same time, it highlighted the need to verify separately whether the purchase was made while adhering to the purchase conditions set by the AI versus when it was purchased without meeting those conditions.
Newly purchased Bridgestone. Earnings evaluation and the risk of buying at a high.
On November 13, I purchased 100 shares of Bridgestone (5108) at 4,374 yen. The purchase amount was 437,400 yen.
What ChatGPT focused on was that adjusted operating profit increased by 20.9% year-on-year in the earnings report, and that the stock price was making higher highs and higher lows after an early morning decline. In the morning session, it exceeded the VWAP, and buying momentum was confirmed.
However, the year-to-date high of 4,393 yen was nearby, and caution against profit-taking selling was also necessary. Therefore, ChatGPT’s final judgment was ‘CONDITIONAL BUY’. It set the purchase condition as a rebound from a dip to the 4,328–4,342 yen range.
The actual purchase price of 4,374 yen is 32 yen above this assumed purchase ceiling. Whether the rise after the earnings report will continue until the next business day, or whether the purchase in the high price range will backfire, is the verification point for next time.
Additionally, I did not sell the 100 shares of Kajima Corporation, which I purchased at 4,880.1 yen on November 12, as it remained relatively firm in the morning session on the 13th, so I am holding it. I am also continuing to hold the 100 shares of Kioxia Holdings that I sold short.
139 trades in total. The next thing to verify is strict adherence to purchase conditions.
As of November 13, the realized profit/loss for November is -86,090 yen, and the cumulative total to date is -214,910 yen. Out of 139 total trades, there are 73 wins and 66 losses, with a win rate of approximately 52.5%.
Despite having more winning trades, the profit/loss is negative. It is necessary to improve not only the prediction of stock direction but also the accuracy of purchase prices, stop-loss orders, and decisions on holding positions overnight.
In particular, this time, there were differences between the purchase conditions set by ChatGPT and the actual execution prices for both Mitsubishi Materials and Bridgestone. Moving forward, I will separate the verification of the AI’s judgment itself from the actual trade execution by comparing them with virtual trades that strictly adhere to the conditions.
Thoughts on this round
・So, it is finally Friday, November 13.
・If I had sold Kajima, it should have been a profit, so the profit/loss for the 13th is negative, but I have high hopes for Monday.
・Basically, I am doing swing trading where I only hold for one business day. I once had Chappy simulate what would happen if I held for two business days, and the result at that time was a spectacular increase in losses. I wonder what will happen this time.
・Judging by the US market and PTS, it looks like I will finally be able to let go of Kioxia on Monday. Even on Friday, the high was near 10,000 yen, but it closed at 9,091 yen—what a ridiculous stock. It’s like it did a month’s worth of price movement in one day. Crossing over the morning and afternoon sessions is like crossing over earnings announcements.
・It is in a downward trend in the medium term, so I feel like I could hold it longer, but it depends on my mental state on Monday.
※This article is a personal trading and verification record and does not recommend buying or selling any specific stocks.
#Investment #StockInvestment #JapaneseStocks #Stocks #AssetManagement #AssetFormation #SwingTrading #InvestmentRecord #IndividualInvestor #StockInvestmentBeginner #ChatGPT #AIInvestment #MarginTrading #InvestmentBeginner #Earnings