Originally posted by no1marauderDoes Fritz have the ability to generate this data, or do you go through each game and count?
I analyzed all my games with Fritz8 some months ago and updated it to 697 games (since then I resigned 9 tournament games without making a move). By my criteria, exact match or a move within .05 of a pawn equivalent to Fritz's first choice, I matched up 10.896/16,933 or 64.3%. In my two games with Xanthosnz, I matched uo 32/49 or 65.3%; he matched up 32 ...[text shortened]... her matchups are a little higher than I remembered; in our 12 games, they are 156/281 or 55.5%.
2 edits
Originally posted by WulebgrIf it does, I'm tooooooooo computer illiterate to figure it out. I do it the old-fashioned way, usually doubling and tripling-checking (I was an accountant before I went to law school, so tiresome counting up of numbers isn't new to me).
Does Fritz have the ability to generate this data, or do you go through each game and count?
EDIT: I don't have to "rediscover the wheel"; I have a notebook with a summary of all my RHP games and various info concerning them(openings, rating of opponents at the time, etc. etc.) of which my and my opponents matches with Fritz are included.
Originally posted by no1marauderThat's a lot of work.
If it does, I'm tooooooooo computer illiterate to figure it out. I do it the old-fashioned way, usually doubling and tripling-checking (I was an accountant before I went to law school, so tiresome counting up of numbers isn't new to me).
EDIT: I don't have to "rediscover the wheel"; I have a notebook with a summary of all my RHP games and various ...[text shortened]... opponents at the time, etc. etc.) of which my and my opponents matches with Fritz are included.
I do not know about the most recent version of Chessmaster, but the analysis function of CM 7000--the latest version I've used provides this sort of data up front. In other respects, it is far inferior to Fritz, however.
Originally posted by WulebgrWell, the whole point of doing an analysis is to actually look at it and determine where you could have played better moves. That requires a certain amount of time to do right. I often don't agree with Fritz's choices anyway; it might be tactically perfect within its horizons but at least at my settings it's long-term positional judgment is inferior to a strong, experienced player. My main use for Fritz is to see where I could have made tactically stronger moves; I've found quite a few times that I make an automatic positional move like putting a Rook on an open file or a Knight on a weak square but miss a tactical opportunity that could have led to material gain. Studying Fritz's analysis has made me look deeper in all my games for tactical chances.
That's a lot of work.
I do not know about the most recent version of Chessmaster, but the analysis function of CM 7000--the latest version I've used provides this sort of data up front. In other respects, it is far inferior to Fritz, however.
Originally posted by wormwoodQuite right. But they would normally be weak players.
still, somebody [b]has to be in the tail, because there is a tail.[/b]
If we started with a group of strong players only, someone would still be in the tail, that's true. However, you could measure whether the extent of the mismatch is "normal" or whether it is a statistical aberation. I'm certain there are a large number of statistical aberations at the upper end of the curve, but have no idea about the lower end. That's why this is such an interesting thread.
Originally posted by no1marauderI agree, and use Fritz the same way. But, with 40+ games going at once, I don't go through every game as I might.
Well, the whole point of doing an analysis is to actually look at it and determine where you could have played better moves. That requires a certain amount of time to do right. I often don't agree with Fritz's choices anyway; it might be tactically perfect within its horizons but at least at my settings it's long-term positional judgment is inferior t ...[text shortened]... l gain. Studying Fritz's analysis has made me look deeper in all my games for tactical chances.
I also use a lot of engines in addition to Fritz for analysis. Suppose a cheater wanted to conceal the activity from your Fritz analysis, there are hundreds of other engines he or she might use. Dozens of these are within 50 ELO of Fritz.
1 edit
Originally posted by WulebgrI'm not a Game Mod; I don't do my analysis for the purpose of detecting cheaters. I report when my analyses shows very high matches to Fritz as I believe that is strong evidence of engine use. If I were a Game Mod, I have most of the good freeware engines as well (Crafty, Ruffian and others) and could check for those as well. There are many techniques someone using an engine could try to hide his engine use, at least in part: I have never claimed that I have any infallible method for detecting all cheats. But I do say that the ones I have ID'ed has cheats ARE based on Fritz analyses and other factors I consider in my "totality of evidence" approach.
I agree, and use Fritz the same way. But, with 40+ games going at once, I don't go through every game as I might.
I also use a lot of engines in addition to Fritz for analysis. Suppose a cheater wanted to conceal the activity from your Fritz analysis, there are hundreds of other engines he or she might use. Dozens of these are within 50 ELO of Fritz.
1 edit
Originally posted by WulebgrThe amount of time you need to allow for analysis depends on the computer. I have an A64 3500+ and it averages around 1500-2000kN/s (thousands of positions per second). This means it gets to around 16 ply in 60 seconds. Of course when analysing whole games this figure is actually higher because Fritz goes backwards through the game and using hash tables has access to analysis done on later positions when analysing earlier ones.
Sixty seconds seems a little quick. If I decided to cheat in a correspondence game, I'd let my engine run for a lot longer than that.
When I am checking my games using Fritz in some positions this isn't enough however. I have one OTB game where Fritz was nigh on useless for analysis (That was a freakish game though, 32 pieces on the board at move 19 and still 28 left on move 40).
4 edits
Originally posted by XanthosNZi think this gets to the heart of the matter ...
.... I have one OTB game where Fritz was nigh on useless for analysis (That was a freakish game though, 32 pieces on the board at move 19 and still 28 left on move 40).
there are two ways to generate a low match up ...
either:
1/ play openings to regularly enter middlegame positions where there are many"correct" moves.
or
2/ somehow magically choose different moves to the engines in all positions.
there are two ways to generate a high match up ...
either
1/ play openings to regularly enter middlegame positions where there is only one"correct" move
or
2/ somehow magically choose identical moves to the engines in all positions.
Originally posted by XanthosNZI think it depends more on what you're trying to do. My point was that sixty seconds seemed short relative to the amount of CPU time that a cheater could devote in a correspondence game. If catching cheaters is the intent (No1 said this is not his primary objective), you may need more engines and longer time per move.
The amount of time you need to allow for analysis depends on the computer.
On the other hand, my old, now deceased, 80386 running Chessmaster 2100 with 1 MB RAM could find plenty of tactical oversights, and would have been sufficient to beat most of the players on this site.
Are you running the analysis while you have your browser open? My head-to-head tests of an AMD64 3200+ and a P4 540 HT revealed that the AMD is faster when Fritz is the only application open, but when multiple programs are running, the P4 runs circles around the AMD.
What is the Fritzmark on your machine? That number is far more useful than kN/s and ply depth, as these can vary dramatically at different points in the game. It is also useful to remember that Hiarcs 9, which runs within +/- 30 ELO of Fritz 8 generally examines far fewer than half the number of kN/s.
Originally posted by WulebgrFritzmark of 1650 when I've got the computer setup to run analysis or play in the engine room of playchess.
I think it depends more on what you're trying to do. My point was that sixty seconds seemed short relative to the amount of CPU time that a cheater could devote in a correspondence game. If catching cheaters is the intent (No1 said this is not his primary objective), you may need more engines and longer time per move.
On the other hand, my old, now decea ...[text shortened]... ich runs within +/- 30 ELO of Fritz 8 generally examines far fewer than half the number of kN/s.
Originally posted by XanthosNZYou must be running multiple applications. That's about what my girlfriend's AMD 3200+ gets when multi-tasking. It has been as high as 3100 when running Fritz alone. My P4 runs above 2200 when multi-tasking, but tops out just above 2700 when Fritz is the only application.
Fritzmark of 1650 when I've got the computer setup to run analysis or play in the engine room of playchess.
Originally posted by flexmorePlayers over 2100 in their games against me including only those games which were really played (not timeouts or game management resignations):
how about all players above 2000 ... or if that is still too many then all above 2100 or even 2200 ... you choose.
thanks for all the effort 🙂
Player #Games W-L-D Match/Non-Opening Book Moves Pct
Weyerstrass 6 0-6-0 228/260 87.7%
Meman 12 0-10-2 315/393 80.2
Quirine 4 1-3-0 94/120 78.3
Hww 2 0-2-0 36/53 67.9
David Tebb 1 0-0-1 20/41 48.8
Cludi 11 1-4-6 219/308 71.4
Nicohuyboom 4 1-3-0 134/172 77.9
Jimster 10 1-6-3 232/329 70.5
BlackBuck 10 0-5-5 243/320 75.9
LittleBear 2 0-2-0 35/44 79.5
Gatecrasher 4 1-2-1 77/112 68.8
BBarr 2 0-1-1 48/67 71.6
Jerzy 2 0-2- 0 38/46 82.6
Dave's game wasn't one of his best; I had a won endgame position but neglected to check Fine's book and offered a draw that he was happy to accept.
Some of the games with the same players were played quite far apart in time, so the results should be interpreted with care. But you asked for the data, so here it is.
Originally posted by WulebgrIt's odd, I have nothing running in the background (if I do the mark drops to 1200 or so) and this rig constantly outperforms anything with similar or even much better fritzmark ratings. I've tried various things to improve the benchmark (hoping for an increase in performance) but nothing seems to work.
You must be running multiple applications. That's about what my girlfriend's AMD 3200+ gets when multi-tasking. It has been as high as 3100 when running Fritz alone. My P4 runs above 2200 when multi-tasking, but tops out just above 2700 when Fritz is the only application.
The only information I can find online is a single article by Steve Lopez back when Fritz 5 was cutting edge.
Perhaps it's a odd thing about my particular system.
Any tips?