Saturday, November 16, 2024

World Chess Championship 2024

 Gukesh will face reigning champion Ding Liren in a 14-game match. Gukesh is the clear favorite, but Liren's chances are better than I expected:



However, Gukesh's rating has been soaring while Liren has been bleeding points. To find a month in which Liren has gained points, I had to go all the way back to May 2023. I calculated the performance ratings for both players in the second half of 2024. Gukesh has been playing at the 2800+ level; his performance at the Olympiad was one of the greatest of all time. Meanwhile, Liren's performance is in the mid-2600s. If I plug in the performance ratings, then Gukesh is the huge favorite.


However, if the match goes to tiebreaks, Liren has better chances. Gukesh's rapid rating is 2654, and this does not seem to be a fluke. While his classical rating was skyrocketing in 2024, his rapid rating actually declined. The first tiebreaker is a 4-game rapid match:


If the match is still tied, then they play two more rapid games:

Putting it all together: the forecast based on the ratings alone is

Gukesh wins: 0.6172 + (1 - 0.6172 - 0.2121) (0.14155) + (1 - 0.6172 - 0.2121) (1 - 0.14155 - 0.6509) (0.1611) = 64.71%

Liren wins: 0.2121 + (1 - 0.6172 - 0.2121) (0.6509) + (1 - 0.6172 - 0.2121) (1 - 0.14155 - 0.6509) (0.51675) = 34.15%

Which leaves about a 1% chance that the match will be decided in blitz tiebreaks.

The forecast based on performance ratings is 

Gukesh wins: 0.9186 + (1 - 0.9186 - 0.0403) (0.14155) + (1 - 0.9186 - 0.0403) (1 - 0.14155 - 0.6509) (0.1611) = 92.58%

Liren wins: 0.0403 + (1 - 0.9186 - 0.0403) (0.6509) + (1 - 0.9186 - 0.0403) (1 - 0.14155 - 0.6509) (0.51675) = 7.15%




Wednesday, October 9, 2024

USA Memory Championship 2024

I had a disappointing performance in the qualifier. For several days before the event, I was hovering on the edge of getting sick, then managed to fight it off. But that meant going into the qualifier without much training in the last week. But I still did well enough to make it to the finals. I flew to Orlando to compete against 11 other finalists at the end of September. Hurricane Helene largely spared the Orlando area, so the championship went ahead as scheduled.

The first event was Memory Match Overdrive, a game from Lumosity. We would have to remember if the card on the right was the same as the one that came three cards before it (i.e., the one on the far left of the screen). Then it would be turned over face down, all the cards would shift one place to the left, and a new card was turned over. Scoring is based on speed and accuracy. I didn't train very much for this event. No one would be eliminated; it would just set the seeding. I was pleasantly surprised to almost set a personal record, good enough for 7th place. Samay Shah dominated, nearly doubling my score. He has made it to the finals multiple times. Jason Smith, who has also been to the finals several times, took second place. Erdem Duluung (winner of the qualifier) was third, and 5X national champion Nelson Dellis was fourth.



Long Term Recall was next. We had been given a spreadsheet with trivia about all the US national parks (e.g. size in square miles, what year was it established in). There was substantially less information than in the past. It was just a two page spreadsheet. In previous years, we would have to memorize about 10 pages. The spreadsheet was given to us a month in advance. Because it was so short, I procrastinated for two weeks, but I still had enough time to memorize it all. We would answer questions about the material, and they got increasingly difficult as the event continued. Three incorrect answers and you are eliminated. The event would continue until the field of 12 finalists was reduced to 9. 

I was the first one to make a mistake. I was asked about which park in Arizona was established in 1994. I thought of two parks in Arizona, but neither one was established in 1994, and I only get 15 seconds to answer the question. There is only one other park in Arizona (Saguaro), and that was the correct answer. The other finalists were doing very well. In the past, several of them would struggle to memorize the vast amount of material, and the event would end quickly. But with just a two page spreadsheet, the rest of the field did much better.

The questions started getting tougher. A few of my rivals finally started to crack. Samay Shah was eliminated, and surprisingly, 5X champion Nelson Dellis missed two questions. Highschoolers Annabella Kang and Ian Liu were eliminated, and I didn't make any more mistakes. Dark horses Briley Lewis and Patricia Lee made zero mistakes, even though it was their first memory championship. I advanced to the next round, "Words to Remember."

We were given 15 minutes to memorize a list of 300 words. Last year, I was eliminated in this event, so I trained harder for it this time. No one memorizes all 300 words. I just had to know more than the last three competitors, since this event would reduce the field from 9 finalists to 6. Historically, memorizing the first 100 words was enough. But I could do 120 comfortably, so I attempted about 115. I made one dumb mistake around the 60th word. I memorize in pairs, and it was "impression - spoons." But somehow I made impressions plural. As soon as I said it, something felt wrong. But I got everything else right, and made it through the rest of the event comfortably. Amanda Zheng, who has competed in the past, was eliminated. We also lost Makenna Good (winner of the 2020 online championship) and first timer Patricia Lee. Patricia had won the numbers event in the qualifier, memorizing 80 digits in under 25 seconds - a stunning result.

The Tea Party was next. We would "meet" a tea party guest by watching a video where they told us their name, birthday, job, and more. After "meeting" all 6 guests, we would watch the videos one more time and then have 6 minutes to look at their information printed on paper. My 5 rivals were all very accomplished. Jason Smith and Avi Chavda had both won 2nd place in past. Erdem had won the qualifier. Nelson Dellis had won the US Championship more times than anyone else. And Briley was making an excellent debut - it seems that we always have one newcomer who shines in the championship. 

Even though the guests weren't talking too fast, it is hard to keep up with all that information. I knew this would be a challenge when I was training, so I had planned to skip about a third of the information. As in the other events, you don't have to know everything; you just need to know more than the finalists who get eliminated. Three mistakes and you are out. But when I was training, I had the computer read the script very slowly, so I was caught off guard when the people in the videos talked at a normal tempo. I quickly decided to skip half the information instead of a third. In the first round, Avi stumbled, forgetting a birthday. I got it right, and felt that I had a chance to make it. Historically, I often get eliminated in the Tea Party (2018, 2021, 2022). Erdem also made a mistake in the first round. I missed the next two questions; I had skipped the pets and favorite foods of the Tea Party guests. Erdem got eliminated. Avi missed another question, while I remembered a guests' alma mater. Meanwhile, Briley and Nelson had both made one mistake so far. Avi was asked about one of the guest's pets. He was on the cusp of elimination, but he nailed it. I was up next, and was asked about a guest's hobbies. I had skipped most of those. I remembered some hobbies, but I didn't know which guest liked them, so I tried that. It wasn't right, and I was knocked out. Jason started to stumble, and he was eliminated soon afterwards. Nelson, Briley, and Avi advanced to the last round. 

They had 5 minutes to memorize 2 decks of cards. Maybe the nerves finally got to them; the performance was surprisingly shaky. Avi messed up when they were only about halfway through the first deck. In the very next round, Briley made a mistake. Nelson was the last man standing, so he was crowned as champion for a 6th time, breaking his own record. He then tried to recite the rest of the first deck, but made multiple errors. "Not my best deck," he admitted later. 

Next up: the champions tournament in November. I'll have to train better for the Tea Party. Congrats Nelson!






Saturday, October 5, 2024

Sunday, September 22, 2024

The Olympiad and MinStrength

 Gukesh had an amazing result at the Olympiad. Let's see how it compares to other great performances. Last year, my article "MinStrength: An Alternative to Performance Rating" was published on ChessBase. The problem with performance ratings is that a perfect score always leads to an infinite performance - even against very weak opposition. By that metric, my 2/2 at the 2024 Virginia State Championship was better than Caruana's 2014 Sinquefield Cup. Obviously Caruana is the stronger player. 


In the article, I explained, "MinStrength asks, 'Who would not be expected to score as well as you did?'  Consider a 1500 player. Sometimes they have a bad tournament and play like a 1300. But there are other times when they perform at the 1700 level. But even in their best tournament ever, they don’t perform at the 2700 level and earn a GM norm. That is outside of their range. How low would your rating have to be for a result to be outside of your range? That is your minimum strength, or MinStrength." 


Gukesh scored 9.0/10 and his average opponent was 2690. That leads to a MinStrength of 2758, which is the second highest of all time. He is only behind Caruana's 2014 Sinquefield Cup (MinStrength 2830). He is ahead of many great performances, such as Karpov in Linares 1994 and Fischer's 6-0 victory against Larsen.


Sunday, August 18, 2024

Sinquefield Cup 2024

The reigning world champion is the 9th seed in this event. That wasn't a sentence I ever thought I would write. He will face his challenger in one of his games.



Tuesday, April 2, 2024

Saturday, January 6, 2024

Tata Steel 2024

 World Champion Ding Liren is the top seed in the 14-player round robin.