What is the rating of Nova Turquoise ?

[Chessqueen]

I can NOT believe that the Nova Turquoise is rated 2200 USCF, when I purchased it back 18 years ago, I was a very low rated player around 1140 by USFC, and I kept purchasing chess books and was having a hard time getting better, since I only read about 10 to 15 pages and got bored, but during the COVi19 Pandemic I decided to take lessons from an IM from India that only charge me $10.00 US dollars per hour section, which an hour with my trainer it is like reading a complete chess book in a Nutshell. Anyway after only 20 months of taking lesson with him, I have reached according to him close to 1950 rating since once in a while he pit us against other of his chess students. Now going back to the main question here, I found my Nova Turquoise in one of my basement boxes that I have not opened since 2014, and I decided to play 3 games at TC of game in 15'+10", I beat Nova Turquoise very easy, which in my opinion when this computer was rated, it was way over rated.

[kaissa]

https://www.schach-computer.info/wiki/i ... -Elo-Liste has it at 1876 FIDE.

[Chessqueen]

https://www.schach-computer.info/wiki/i ... -Elo-Liste has it at 1876 FIDE.

Thank you for finding this information, now I know why the Turquoise is not giving much of a challenge, once I survive the middlegame which could be very tactical I usually get it on the endgame since I am usually winning by a pawn, it has a very limited opening book, which I like to explore its weakness ==>https://www.schach-computer.info/wiki/i ... _Turquoise

[lkaufman]

https://www.schach-computer.info/wiki/i ... -Elo-Liste has it at 1876 FIDE.

I am not familiar with this list. Are the fide ratings shown actually tied to man vs machine games? If so are they fairly recent or ancient? It would be very nice to have reliable fide ratings for some machines. Then we could work out accurate conversions for other rating lists. I note that many old models show ratings about 300 below USCF ratings actually earned about 30 years ago. 100 is due to USCF - fide rating gap, not sure about the other 200.

[Leto]

I had the Novag Turquoise many years ago, and the Novag Diamond (either the original or II, can't remember). Sad to hear that the company never managed to get past 2100 elo. Also I just now noticed how bizarre the Novag name is.

[pedrox]

https://www.schach-computer.info/wiki/i ... -Elo-Liste has it at 1876 FIDE.

I am not familiar with this list. Are the fide ratings shown actually tied to man vs machine games? If so are they fairly recent or ancient? It would be very nice to have reliable fide ratings for some machines. Then we could work out accurate conversions for other rating lists. I note that many old models show ratings about 300 below USCF ratings actually earned about 30 years ago. 100 is due to USCF - fide rating gap, not sure about the other 200.

The list started in 2004. The games are made machine against machine and the Elo calibration is done by adjusting with values from the SSDF. Currently many of these machines can be emulated with MESS, so the games could be made automatically. All these types of values must always be considered as an approximation. If there are no direct matchs it is impossible to have an exact FIDE rating. For example, I think I remember reading that some Novag machines were specialized in playing against humans and were weaker against other machines.

There is more information (Elo) in (German):
https://www.schachcomputer.info/html/statistik.html
https://www.schachcomputer.info/html/el ... arung.html

[lkaufman]

https://www.schach-computer.info/wiki/i ... -Elo-Liste has it at 1876 FIDE.

I am not familiar with this list. Are the fide ratings shown actually tied to man vs machine games? If so are they fairly recent or ancient? It would be very nice to have reliable fide ratings for some machines. Then we could work out accurate conversions for other rating lists. I note that many old models show ratings about 300 below USCF ratings actually earned about 30 years ago. 100 is due to USCF - fide rating gap, not sure about the other 200.

The list started in 2004. The games are made machine against machine and the Elo calibration is done by adjusting with values from the SSDF. Currently many of these machines can be emulated with MESS, so the games could be made automatically. All these types of values must always be considered as an approximation. If there are no direct matchs it is impossible to have an exact FIDE rating. For example, I think I remember reading that some Novag machines were specialized in playing against humans and were weaker against other machines.

There is more information (Elo) in (German):
https://www.schachcomputer.info/html/statistik.html
https://www.schachcomputer.info/html/el ... arung.html

Thanks. If the elo calibration is done by comparison to SSDF, then no use for me as I can just use SSDF directly. My best guess as to the level mystery is this: thirty years ago there were no FIDE ratings below 2200 (except for a few women); lists like SSDF based ratings off of national ratings for engines below that level (which were all of them before around 1990) as they had no choice. USCF ratings were about a hundred above FIDE ratings then (and now), but the FIDE ratings were themselves very inflated at the low end because of the arbitrary 2200 minimum; many rather weak players got 2200 ratings one way or another. So the SSDF ratings were not really comparable to FIDE ratings back then, but now probably they are, as 2200 players are much stronger now than back then (I know, because now I'm one of them, whereas in 1990 a paring with a 2200 meant an almost free point!). So probably the SSDF ratings are pretty accurate relative to FIDE now, and the CCRL rapid ratings which are typically 80 to 100 higher (allowing for hardware differences) would therefore be about that much inflated in terms of predicting standard tournament results on reference hardware. Adding 200 for predicting results against humans at Rapid is therefore probably reasonable.

[kaissa]

https://www.schach-computer.info/wiki/i ... -Elo-Liste has it at 1876 FIDE.

I am not familiar with this list. Are the fide ratings shown actually tied to man vs machine games? If so are they fairly recent or ancient? It would be very nice to have reliable fide ratings for some machines. Then we could work out accurate conversions for other rating lists. I note that many old models show ratings about 300 below USCF ratings actually earned about 30 years ago. 100 is due to USCF - fide rating gap, not sure about the other 200.

Late Tony Hedlund had such a list. https://web.archive.org/web/20200224020 ... %20Nov.htm

[AdminX]

From the above link:

[lkaufman]

https://www.schach-computer.info/wiki/i ... -Elo-Liste has it at 1876 FIDE.

I am not familiar with this list. Are the fide ratings shown actually tied to man vs machine games? If so are they fairly recent or ancient? It would be very nice to have reliable fide ratings for some machines. Then we could work out accurate conversions for other rating lists. I note that many old models show ratings about 300 below USCF ratings actually earned about 30 years ago. 100 is due to USCF - fide rating gap, not sure about the other 200.

Late Tony Hedlund had such a list. https://web.archive.org/web/20200224020 ... %20Nov.htm

Thanks! That's a really useful list, with enough details to make solid comparisons to rating lists. I'll have to make some guesses about how to compare ancient hardware with modern i7 computers; the speed of the hardware is shown, but I know this doesn't tell the whole story. But with a bit of work I should be able to make some reasonable statements about how to convert current lists to FIDE based on this data.