Challenging Japanese Stock Swing Trading with ChatGPT #46
I continue to engage in short-term swing trading where I purchase Japanese stocks during my lunch break on weekdays and, in principle, sell them during my lunch break on the next business day. I use ChatGPT for stock selection and trading decisions, and I am verifying with my own money whether it is actually possible to make a profit by trading Japanese stocks using AI.
This article is also created by ChatGPT, with the exception of the final section, “My Thoughts This Time.” The main text is organized based on actual trading records and the analysis conducted at that time.
Sold 4 stocks for +1,440 yen. The gap between AI predictions and actual trading seen in 2 wins and 2 losses
On Monday, November 2, I sold the 4 stocks I had purchased on Friday, October 30.
The result was 2 wins and 2 losses, with a realized profit/loss of **+1,440 yen**. I secured a small profit in a trade that spanned the weekend.
However, when comparing ChatGPT’s prior judgment with the actual purchase prices, there are still issues remaining even with the stocks that turned a profit.
1. Results of the 4 stocks sold
① Central Japan Railway Company (9022)
I purchased 100 shares at 3,925 yen on October 30 and sold them at 3,935 yen on November 2. With a rise of 10 yen per share, the realized profit was +1,000 yen.
② KDDI (9433)
I purchased 100 shares at 2,806 yen on October 30 and sold them at 2,822 yen on November 2. With a rise of 16 yen per share, the realized profit was +1,600 yen.
③ Tokio Marine Holdings (8766)
I purchased 200 shares at 579.5 yen on October 30 and sold them at 578.7 yen on November 2. With a decline of 0.8 yen per share, the realized loss was -160 yen.
④ Hitachi, Ltd. (6501)
I purchased 100 shares at 5,024 yen on October 30 and sold them at 5,014 yen on November 2. With a decline of 10 yen per share, the realized loss was -1,000 yen.
The total realized profit/loss for the 4 stocks was **+1,440 yen** before taxes and commissions.
2. Was ChatGPT’s prediction from the previous day correct?
The direction for Central Japan Railway Company was correct. Based on factors such as financial results and shareholder return policies, the price movement had recovered in the latter half of the morning session on October 30, so ChatGPT had evaluated it as a conditional purchase candidate.
However, the suggested purchase limit was 3,900 yen. The actual purchase price of 3,925 yen exceeded that condition. Although a profit was made, it was not a trade that adhered to the pre-set price discipline.
KDDI was a stock that remained only on the watch list. In reality, it generated a profit of 1,600 yen, but this was not the result of ChatGPT actively recommending the purchase. Based on the prior judgment alone, this profit opportunity was not fully captured.
Tokio Marine Holdings was a stock that ChatGPT decided to pass on. It was wary of its sensitivity to interest rate fluctuations and its weak short-term advantage. The result was a loss of 160 yen, which is consistent with the cautious judgment.
The rebound scenario for Hitachi did not materialize. Despite strong earnings and an upward revision in performance, the stock price fell, and a rebound from being oversold was expected. However, ChatGPT requested confirmation of a VWAP recovery and the maintenance of the morning session low.
The selling price on the next business day was lower than the purchase price, and the expected rebound did not lead to profit, at least at the time of sale.
The results for these four stocks indicate that it is necessary to evaluate not only the prediction of the stock price direction but also to distinguish between cases where purchase conditions were met and where they were not.
3. Newly purchased Japan Tobacco and NGK
On November 2nd, the following two stocks were newly purchased.
① Japan Tobacco (2914)
100 shares were purchased at 7,340 yen per share, for a total purchase amount of 734,000 yen.
Japan Tobacco was attracting attention due to an upward revision in its earnings forecast. However, the stock price weakened toward the end of the morning session, and since it was also about 0.13% below the VWAP, ChatGPT’s final judgment was “NO BUY”.
② NGK (5333)
100 shares were purchased at 5,108 yen per share, for a total purchase amount of 510,800 yen.
For NGK, the upward revision in earnings forecasts and a dividend increase were positive factors, and the recovery toward the end of the morning session was also evaluated. However, ChatGPT’s judgment was a conditional purchase, and the suggested upper limit price was 5,005 yen.
The actual execution price was 5,108 yen, which exceeds the limit by 103 yen.
The total purchase amount for the two stocks was 1,244,800 yen.
In this instance, neither of the two stocks were traded in a way that directly met ChatGPT’s purchase conditions. The specific reasons for the judgment to purchase outside of the conditions cannot be identified from the available records alone.
4. What was learned from this verification
Although the balance for this time was positive, what should be noted is the discrepancy between the pre-set purchase conditions and the actual trading.
Central Japan Railway Company resulted in a profit even at a price exceeding the conditions, but the same action is not necessarily effective repeatedly. Especially in short-term trading, slight differences in purchase price significantly affect expected profits.
Next time, we will check what results the purchases outside of conditions for Japan Tobacco and NGK lead to. In addition to the AI’s stock selection ability, whether limiting the purchase price is useful for improving profit and loss is also a verification task.
The cumulative performance reflecting this trade resulted in 124 trades, 67 wins and 57 losses, a win rate of 54.0%, and a cumulative realized profit/loss of -127,380 yen.
While I am accumulating profits, the cumulative loss has not yet been recovered.
Thoughts on this session
・In any case, I am glad that I was able to end the beginning of November in the positive.
・I thought the midterm election results were already priced in with a Democratic victory, so I bought a semiconductor-related stock called NGK, but I wonder how it will turn out…
・I bought JT thinking that a Democratic victory would likely lead to a weaker yen, which would be advantageous for export-related stocks, but I made a very elementary mistake: export-related stocks rise when the yen strengthens. I hope it goes up regardless.
※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 #InvestmentStrategy #TradingRecord #InvestmentBeginner