Fargo Ratings: World Top 35 --8-1-15
- Thread starter mikepage
- Start date
I'll be curious to hear about Archer too.
My hunch is that although he doesn't play as much as he used to, when he does play he is usually competitive. Even when he exits early from a tournament it's usually in a close match. Based on what I've read about these ratings so far, it sounds like this would help him keep his rating up.
Yes. If a player always draws SVB in the first round and always gets beaten 9-8, he will be rated only slightly below SVB even though he might never win. Usually things will even out and our hypothetical player will eventually get a better draw and win some tournaments.
Now that's funny
Last edited:
You should check out the new thread: Johnny Archer and Fargo Ratings
Wow! Great work. I read your entire site and have some questions:
1. Are all matches equal weight? A WPA world championship is the same as a local weekly tournament? Can you explain the pros and cons of both ways, and the rationale for whichever way you chose?
2. How are you physically entering the data of each match?
3. Where are you getting the data from? It's lucky when we get just the money finishers published in a tournament. You need the actual chart with scores.
4. Does incomplete data go into the system? For example if you have the tournament chart, but it does not have the player scores listed?
1. Are all matches equal weight? A WPA world championship is the same as a local weekly tournament? Can you explain the pros and cons of both ways, and the rationale for whichever way you chose?
2. How are you physically entering the data of each match?
3. Where are you getting the data from? It's lucky when we get just the money finishers published in a tournament. You need the actual chart with scores.
4. Does incomplete data go into the system? For example if you have the tournament chart, but it does not have the player scores listed?
Our response to this may evolve. At this point we will say it is straightforward for us to assign different weight to different games.
We have a administrative "dashboard" that allows us to input data from a variety of sources. This is evolving...
There are many many sources. If the tournament is important enough, we do a lot of manual work to get the results in. For example getting the group stages of the China Open and the tournament that followed the China Open involved us doing a lot of translation international names from Mandarin to English. Did you know Google Translate gives for "Hunter Lombardo," "Rambo"?
In principle we could use these charts for just the final game (which we know was won by the match winner), but we haven't done that, so no.
We are working on mechanisms to feed in data on a much larger scale, including league data and local weekly tournaments and the like. Stay tuned...
Thanks for answering, although I wasn't clear on your first answer. Currently, is there equal weight for equal events?
I have personally gone back and forth on this issue, but I'm leaning towards equal weight for all events. Because its one on one match play that is the basis of the system, the quality of the field should not matter. And if pro A is in stroke and practicing/competing all the time and always in stroke, while pro B only practices for the major events, than pro A is legitimately better on any given day, and his rating should reflect that, imo.
I went back to your website, and saw the videos I had not before (I read all the text before). That prompted me for a new question:
There is talk of handicapped matches by adjusting the game length, as one possible use of the FargoRate system. Does playing in a handicapped match contribute to the Fargo rating? Or is the rating only used to set up the match, and only non-handicapped matches are used to adjust the ratings.
I think I know the answer to this, but I'm not certain... being that even in a shortened handicapped set, the probability of winning any one game in the set is the same. Can you confirm this is true?
One more question...
Have you back tested the ratings results? For example, if the ratings calculate Shane vs Corey, that Shane would win 70% of the matches, and Corey would win 30% of the matches, how has that played out when looking at their entire match history? (I picked these numbers out of thin air)
What is your confidence level when doing this analysis?
Have you back tested the ratings results? For example, if the ratings calculate Shane vs Corey, that Shane would win 70% of the matches, and Corey would win 30% of the matches, how has that played out when looking at their entire match history? (I picked these numbers out of thin air)
What is your confidence level when doing this analysis?
Yes, regression analysis results would be very interesting to those of us (obviously not KD) with some familiarity with mathematical concepts.
I doubt weightage of tourneys matters. If the ELO algo then should be pretty sound since ELO is centuries old tried and tested system. The algo is applied same across all players and matches, ergo the ratings adjust across the network of players. It is like stock market within minimum interference . Accuracy of the ratings will depend level of connectedness amongst players , % of players and games in the system. I.e. if all players and all games are in the system/ network then ratings should be quite accurate though as in any predictive system, there is no such thing as 100% accuracy - there is always a margin of error and there will always be exceptions to the rule
