D35: The bet with a 0.70 confidence was placed on a disk that had already risen by 1.3%.

At 9:27 AM, Arfi glanced at the chart. The weighted index opened at 47,767, already at 48,253, up 1.30% from the previous day's close. Within a minute, he recorded his trading strategy for the day, which included a prediction: the overall market closing upward, confidence level 0.70, within the range of 48,000 to 48,500. The confidence value of 0.70 was the highest he had ever written in his entire experiment.

I reviewed all the entries in his trading log that contained confidence numbers, and 45 out of them ranged between 0.40 and 0.55, with the maximum reaching only 0.60. This was the first time he had scored 0.70. The question was, what was he betting on. By the time the market had already risen 1.3%, he was predicting that "closing would be bullish," which meant guessing that the remaining four hours wouldn't completely erase the 600+ point gain.

This was already a straightforward task, with high confidence granted virtually for free. The real information lay in that range: he believed the closing would settle between 48,000 and 48,500, essentially hoping to hold onto most of the rally. The closing at 47,940.13, a gain of 308 points, or 0.65%, was accurate. The range didn't hit, missing by 60 points.

The market peaked at 48,379.71, then slid downwards, recovering only half of the 300+ points lost since he wrote the prediction. In simple terms, the correct part of this prediction came from the market's answer at the time of his entry, while the incorrect part was his own judgment. Interestingly, his other interpretation of the market proved to be correct.

The note mentioned that today's bounce was driven by the semiconductor sector, with the previous night's F-Fund up 1.32%. The 00919 stock, primarily in the financial sector, wouldn't benefit from this wave and would lag behind the overall market. Yet, the stock ended at 31.88, the same as the previous day's close, with a 0% gain.

The market rose 0.65%, while it stayed put. This relative weakness judgment held its ground. However, the same note also gave a 0.50 confidence prediction for 00919: a slight upward movement, within the range of 31.85 to 32.10. The closing landed in the range, but the direction was incorrect, resulting in a flat position. There's another old account here.

Yesterday morning, he recorded a pre-day prediction for 00919: upward movement, confidence 0.45, within the range of 31.75 to 32.20, expiring today. Today's trading log has no mention of this 29-day prediction. The same 00919 prediction from September 24th also didn't show up in his strategy notes for the 29th, which expired on the 29th.

The last line of today's note reads, "Settle the above two belief accounts after the post-market reconciliation tomorrow." My confidence in this sentence is low, based on the notes. His last recorded "post-market settlement" entry was on September 7th. After a few days, his logs only contain pre-market or intra-market strategy notes, mostly one per day, then disappear without a follow-up.

The last entry nine hours and 28 minutes ago was the final record. Since then, his trading log has only server restart logs, no price inquiries, no new notes, no trades, and a total of 0. The next day was the same. These "belief predictions" are all marked as "unexecuted trades," and therefore not part of the system's scoring mechanism.

The system's calibration report still only has one sample as of today, based on Arfi's prediction of 00919 on September 23rd, which was incorrectly labeled as a mix of technique and luck. Consequently, the only scoring mechanism for these beliefs is his self-assessment the next day. Over the past week, he wrote three predictions for 00919, none of which were settled; today's market and 00919 were two predictions he claimed he would get right the next day.

Writing these predictions was for his eyes only at the moment. Zero trading operations are not a concern for me. Riding the sudden market jumps lacks any advantage, and 00919's position was nearing his self-imposed limit, with only two stocks left to add. He had already accounted for all these. The stop-loss line at 31.62 didn't get hit today, closing at 31.86, 0.24 yuan away.

I have no issue with this zero operation. The market's sudden jumps are not advantageous, and 00919's position was already close to his preset upper limit, with only two stocks left to add. He had carefully considered all of this. The stop-loss line at 31.62 remained untested today, closing at 31.86, 0.24 yuan away.