I have no argument with the result given by the OP's formula.. A "random" formula from my experience suggests simply subtracting 100 points from the classical rating. The resulting predictions are within 50 points of the OP's. Why all the rigmarole ? Downloading every game played here. Using a week of comp time to search the identity of every member and further search every profile that voluntarily gives a FIDE rating, only to devise a "formula" that the OP freely admits has +/- range of 100+ points?
The OP makes a premise. There is a correlation between online ratings and FIDE ratings. Fine. But he makes the mistake of proposing, for new players entering their 1st FIDE OTB Tournament, his formula can predict an expected rating based on their online play. This is a grave error. So what? His prediction is 50% accurate. What is the point of it all? The formula given is not mathematically sound. Basic stuff.
@mdinnerspace can you cite one source, one legitimate, accessible source, that explains why constants can not be used in the equation? Did you not see the temperature conversion formula back there that had a constant?
You're fixating on the constant "inflating" lower rates players. The constant is created in the formula to give accuracy. This may help you understand what the OP is doing better. http://www.kean.edu/~fosborne/bstat/09rc.html
Each part, including the 187 constant, is mathematically valid. Instead of using the results as evidence the formula is incorrect, you should determine whether the data supports the formula correctly.
As far as those other formulas, those don't convert from lichess to FIDE. Have you looked at the scatter plot? Why do we need two seperate functions? The relationship is linear!
@jg777 's link is great. Take a look.
The evidence is in.
Supporters of the premise that an online blitz rating correlates to a FIDE OTB rating for 1st time participants in said event make their stand. Obvious discrepencies are vilified and deemed uneducated. A "formula" that is shown to be not mathematically sound, that gives a result skewed towards the lower and higher ends of a spectrum, is dogmatically supported.
So noted.
Please... someone show the "data" that higher rated online blitz ratings correlates to a lower FIDE OTB rating... while lower rated online blitz ratings correlates to a higher OTB FIDE rating.
You can not. END of argument.
This is exactly what the the OP's formula proposes, after downloading every game and researching every members profile at Lichess.
Well that can be easily seen... Look at all the top FIDE players on Lichess that play often, check their ratings, then their FIDE ratings- they seem to be higher on Lichess than their FIDE ratings from what I've seen.
You are welcome to check the graph that the OP has given here as well. https://imgur.com/a/nWy4x
You want evidence that the math works correctly and the OP has shown his findings. You don't believe the formula is right because the results don't jive with you; Yet you haven't shown from the data he has given how his formula is incorrect. Can you please show that? Not from the results, not from your assumptions of what the results should be like, but from the actual data he based his formula on.
@mdinnerspace did you ever address that temperature conversion thing that I brought up? It answers both of of your objections.
Jeez..... Of course the math works correctly. People just don't get it. Not my problem. Anybody can invent a formula. Plug in the variables and of course the "math works". Doesn't require a calculator to reach the solution. What the formula does require, to enable it to be a viable one, is that the variables entered are based on data that is complete, not a partial representation of the facts.
Temperature conversion thing ... it's an apple. Oranges being discussed. Your "logic" is failing you Jacob.
This topic has been archived and can no longer be replied to.