Why a Computer Algorithm made me feel like Fischer

It is like a dream: Winning against a player who's chess.com ranking is 250 points better than mine? This does not happen too often to me. I guess your experience is not much different. I did some days ago and was kind of shocked: The chess.com analysis told me that my accuracy rate for this game was — 99 %!

Well, this was hard to believe. It was not a game against a beginner who just found out that a pawn cannot move three fields at once. Neither was it against an opponent who timed out after a few moves, no: It was 31 moves in a serious game against a fine player. Being proud was one thing — but then a terrible thought appeared in my head: Would chess.com ban me as a cheater?

99 % — no, I am not as good in chess to play on this level against someone with a chess.com Daily ranking of 1875 (March 20, 2021). My Daily level then was 1628 — and seriously: I would not even had bet on a draw.

The Stats

Looking at the all time statistics of my Daily games I have an average accuracy rate of 79.4 % while writing this. 99 % seems to be very unlikely, right? On the other hand my chess level varies a lot. I had an accuracy rate of below 30 in a few games that caught me on the wrong foot. On brighter days with better footwear I reached more than 90 % — and I even played two games with an accuracy of 98.7 % in April 2021. Looking at these two the reason is obvious: They ended after 7 and 12 moves.

Getting better? Being lucky?

Yes, this really would be a wonderful explanation: My chess could have got better. I somehow feel that it actually had over the last weeks (when looking at Daily games — my Rapid play still is a mess and I still avoid it). But this always is a very temporary feeling until I mess up another game and think that my cat could do better.

Maybe "pure luck" is a more reasonable explanation for the 99 % game? It could be possible: Me getting up with the right foot and having my inspiring cat sitting on my knees, purring,  while my Indian opponent had really bad days with a dog in his office barking for hours when he made the four mistakes the chess.com analysis found — yes, that's more likely than me beating this fine player.

Having another look: Ooops!

The chess.com analysis — they offer a new feature called "Game Review". I was curious and had another look at the game today. What a surprise to see that my play was rated only at an accuracy level of 94 % by the "Game Review", not 99 %! 94 is still a good number — but this reads way more realistic. Just like my opponent being lifted on a level of 82.5 % from the former 71.1%.

Computers!

I am glad to see that the reason for all this doubt and confusion could be a confused computer algorithm that got up with the wrong foot. Maybe a mad cat walked over the developer's keyboard? Who knows?

Now that my numbers are lower and I am relaxing again I got the idea to play against this confused computer. Maybe I can reach 100 %?

Cheers, Frank

Here's the link to the game:

https://www.chess.com/analysis/game/daily/321008406?tab=analysis