Plug in the numbers jacob. You will find exactly that.
I like the formula, good job @dudeski_robinson ;)
@mdinnerspace
Ah, I thought that you were claiming that as one of the ratings increased, the other decreased. What you are saying is that the online rating points increase at a different speed than OTB rating points, to the point where some players are predicted to have OTB ratings higher than their online ratings, and some are predicted to have OTB ratings below their online ratings.
Right?
If that is the case, it just means that one FIDE rating point does not have the same value as one lichess blitz/classical point.
It appears that you argue that this fact immediately invalidates the equation by violating a basic mathematical principal. It does not. As an example, in converting between Celsius and Fahrenheit, 0 degrees Celsius results in 32 degrees Farenheit, making Farenheit "higher"
At -50 degrees Celsius, the Farenheit temperature is -58 degrees, making Celsius "higher."
Now, I don't think that the conversion equation is unsound. And don't you tell me that it's oranges and apples here. The fact is, you just claimed that a property in the FIDE estimator equation made it mathematically invalid. Inherently mathematically laughably unsound. And yet that same property appears in another equation that is universally accepted.
Ah, I thought that you were claiming that as one of the ratings increased, the other decreased. What you are saying is that the online rating points increase at a different speed than OTB rating points, to the point where some players are predicted to have OTB ratings higher than their online ratings, and some are predicted to have OTB ratings below their online ratings.
Right?
If that is the case, it just means that one FIDE rating point does not have the same value as one lichess blitz/classical point.
It appears that you argue that this fact immediately invalidates the equation by violating a basic mathematical principal. It does not. As an example, in converting between Celsius and Fahrenheit, 0 degrees Celsius results in 32 degrees Farenheit, making Farenheit "higher"
At -50 degrees Celsius, the Farenheit temperature is -58 degrees, making Celsius "higher."
Now, I don't think that the conversion equation is unsound. And don't you tell me that it's oranges and apples here. The fact is, you just claimed that a property in the FIDE estimator equation made it mathematically invalid. Inherently mathematically laughably unsound. And yet that same property appears in another equation that is universally accepted.
@Jacob531 you are right, of course, but the Celsius/Farenheit example was brought up about 100 posts ago, and @mdinnerspace did not appear to grasp why it invalidates his argument.
Also, he never replies to any of my posts when they are more than 2 paragraphs long. Fighting him is risky business.
Also, he never replies to any of my posts when they are more than 2 paragraphs long. Fighting him is risky business.
The C/F example was addressed. Old stuff. The equation gives consistent results. The "logic" I just read and the comparison made is from another planet.
No. The point I make is not understood at all. The interpretations of which entirely miss the issue.
I'll try a last time:
As online ratings decrease the predicted FIDE rating progressively becomes higher (relative to the online rating). So much so that the predicted FIDE rating, using the formula is HIGHER than the online blitz rating. As the lower the rating is, the predicted FIDE rating becomes that much greater than the online rating.
The "trend" reverses direction at approx 1900. The predicted FIDE rating is LOWER than the online blitz rating. As online ratings increase, the formulas prediction of a FIDE rating progressively becomes lower relative to the online rating.
The OP looked at the numbers and Agreed. Said there is nothing out of order. In fact his chart shows this to be in fact representative of his data findings.
I dispute the formula as being mathematically sound.
The OP makes the claim, in general, online blitz ratings are 78 points higher than FIDE ratings with classical ratings being 169 points higher. This can be applied to all rating groups. He "derived" his formula based on this data. YET, when the numbers are inserted into the formula, an entirely different result is seen. Predicted FIDE ratings are higher than their online blitz rating for all ratings below 1900. This raises problems in the formula.
No. The point I make is not understood at all. The interpretations of which entirely miss the issue.
I'll try a last time:
As online ratings decrease the predicted FIDE rating progressively becomes higher (relative to the online rating). So much so that the predicted FIDE rating, using the formula is HIGHER than the online blitz rating. As the lower the rating is, the predicted FIDE rating becomes that much greater than the online rating.
The "trend" reverses direction at approx 1900. The predicted FIDE rating is LOWER than the online blitz rating. As online ratings increase, the formulas prediction of a FIDE rating progressively becomes lower relative to the online rating.
The OP looked at the numbers and Agreed. Said there is nothing out of order. In fact his chart shows this to be in fact representative of his data findings.
I dispute the formula as being mathematically sound.
The OP makes the claim, in general, online blitz ratings are 78 points higher than FIDE ratings with classical ratings being 169 points higher. This can be applied to all rating groups. He "derived" his formula based on this data. YET, when the numbers are inserted into the formula, an entirely different result is seen. Predicted FIDE ratings are higher than their online blitz rating for all ratings below 1900. This raises problems in the formula.
@mdinnerspace writes: "The OP makes the claim, in general, online blitz ratings are 78 points higher than FIDE ratings with classical ratings being 169 points higher. THIS CAN BE APPLIED TO ALL RATINGS GROUP."
The text in CAPS is the point I disagree with.
You have provided no explanation for why the gap between online skills and OTB skills should be exactly constant across all ratings groups.
In fact, the observed data shows that this is NOT the case. In Lichess profiles, we see clearly that the gap between people's online ratings and their self-reported Fide ratings changes across ratings groups. For certain groups of players, that gap is greater than for others. This is an EMPIRICAL finding, from the dataset. I did not make this up. And this is exactly what the formula takes into account.
This is not a "bug" of the formula. The fact that the formula accounts for changing gaps is actually a positive feature, because it reflects an observed tendency in the data.
The text in CAPS is the point I disagree with.
You have provided no explanation for why the gap between online skills and OTB skills should be exactly constant across all ratings groups.
In fact, the observed data shows that this is NOT the case. In Lichess profiles, we see clearly that the gap between people's online ratings and their self-reported Fide ratings changes across ratings groups. For certain groups of players, that gap is greater than for others. This is an EMPIRICAL finding, from the dataset. I did not make this up. And this is exactly what the formula takes into account.
This is not a "bug" of the formula. The fact that the formula accounts for changing gaps is actually a positive feature, because it reflects an observed tendency in the data.
For what it's worth, the fact that the observable gap between Fide and Lichess ratings is NON-CONSTANT across ratings groups is clearly shown in the graph I presented in the very first post in the this thread:
imgur.com/a/nWy4x
imgur.com/a/nWy4x
The problem is not that you didn't express your argument clearly. The problem is that you don't understand why it's a terrible argument.
The OP has stated that both his formula and the data (which shows Blitz ratings to be +78 points higher than FIDE ratings and +169 points higher for classical ratings) that a strong correlation exists for either method in predicting a FIDE rating.
Let's see if this is true by inserting the numbers.
A very common Blitz rating of 1078
and a classical rating of 1169 shall be used.
One would expect the formula to predict a 1000 FIDE rating. (78 points lower than blitz and 169 lower than classical.)
We have 1078 x .48 = 517.44
and 1169 x .38 = 444.22
Add +187
The sum = 1149 !!
Wow! What happened ?
The predicted FIDE rating is higher by 149 points than the expected 1000. In fact it is Higher than the blitz rating by 71 points. I thought the data shows online blitz ratings to be HIGHER than FIDE ratings, but this "great" formula gives just the opposite result and by a big margin, predicting a OTB FIDE rating to be greater than the online rating.
This is suggesting 1st time OTB players should expect a higher FIDE rating than their online blitz rating. Evidence simply proves this is not the case. The trend in the formula does not start predicting a lower FIDE rating than the online blitz rating until the rating 1900, whereby the trend reverses itself and progressively predicts lower FIDE ratings for players in comparison to their online blitz rating.
Let's see if this is true by inserting the numbers.
A very common Blitz rating of 1078
and a classical rating of 1169 shall be used.
One would expect the formula to predict a 1000 FIDE rating. (78 points lower than blitz and 169 lower than classical.)
We have 1078 x .48 = 517.44
and 1169 x .38 = 444.22
Add +187
The sum = 1149 !!
Wow! What happened ?
The predicted FIDE rating is higher by 149 points than the expected 1000. In fact it is Higher than the blitz rating by 71 points. I thought the data shows online blitz ratings to be HIGHER than FIDE ratings, but this "great" formula gives just the opposite result and by a big margin, predicting a OTB FIDE rating to be greater than the online rating.
This is suggesting 1st time OTB players should expect a higher FIDE rating than their online blitz rating. Evidence simply proves this is not the case. The trend in the formula does not start predicting a lower FIDE rating than the online blitz rating until the rating 1900, whereby the trend reverses itself and progressively predicts lower FIDE ratings for players in comparison to their online blitz rating.
"(which shows Blitz ratings to be +78 points higher than FIDE ratings and +169 points higher for classical ratings)"
Really? Where are you getting this from?
"Evidence simply proves this is not the case."
What evidence?
"One would expect the formula to predict a 1000 FIDE rating."
No. Look at the graph. That's not what we expect.
"The trend in the formula does not start predicting a lower FIDE rating than the online blitz rating until the rating 1900, whereby the trend reverses itself and progressively predicts lower FIDE ratings for players in comparison to their online blitz rating."
No trend is reversed.
Really? Where are you getting this from?
"Evidence simply proves this is not the case."
What evidence?
"One would expect the formula to predict a 1000 FIDE rating."
No. Look at the graph. That's not what we expect.
"The trend in the formula does not start predicting a lower FIDE rating than the online blitz rating until the rating 1900, whereby the trend reverses itself and progressively predicts lower FIDE ratings for players in comparison to their online blitz rating."
No trend is reversed.
This topic has been archived and can no longer be replied to.