|
|
| |||||||
| From | Message | ||||||
|---|---|---|---|---|---|---|---|
|
 GK communication resultsFrankly, this might be my last year as I’m getting tired of poor management and I think these clowns are going out of business. |
||||||
|
you could use the advanced player search (APS) gameknot.com Then put your limits on current 90 day rating and so much more. |
||||||
|
"Advanced Player Search" It is a great tool to match the highest rated as well as hard to match mates (high 90 day rating, etc). |
||||||
|
Rationale for Using 90-Day-Max Rating in Setting Matches? In creating challenges, I have to skip many otherwise good opponents because the 90-day max ratings are not within the tolerance (typically 50 points) set by the opposing team. I don't find that number useful, and am curious why other teams do. I have heard that it shows "potential". However, it seems to me that it may just as well be showing a recent and short-lived winning streak. I'm eager to hear your thoughts. - Dave |
||||||
|
challenge, so he declined stating the 90 day high rating is too high. Since then other teams have used that excuse to decline challenges. There can be no other reason. Dave, I agree that 90 day high ratings are somewhat useless. In fact, the average 90 day rating would reflect the high 90 day rating. The 90 day high rating unfairly "punishes" fast moving players with many active games. There was a period of time where I told players to control the 90 day hag rating by resigning the lost games before winning the won games. |
||||||
|
|
||||||
|
@ mikemate challenge, so he declined stating the 90 day high rating is too high." Is that a true story that you know about, or an amusing imagined story? Anyhow, I think at least some of the blame belongs to whichever of GameKnot's previous owners placed the "Top rating (90 days)" directly below the current rating among the stats that appear when scrolling over a player's screen name. Also, GameKnot uses the "Top rating (90 days)" for eligibility for mini-tournaments. GameKnot promotes the 90-day high much more than it promotes the average. |
||||||
|
Advanced Player Search |
||||||
|
|
||||||
|
I could with a clear conscience nominate him for games way over the usual +-50 difference above him, and I could explain to opposing captains what had happened and that they must not underestimate him. Now, my player timed out: I can explain that- I even contacted him by telephone to find out what happened. But some opposing players are sandbaggers. They time-out and lose deliberately if the 90-day average becomes too high for them. Opposing captains cannot always explain why their players' 90-day average drops, but when I look at the statistics and analyze it, I see the trend and I am wary. That is the time that I look deeper into longer average performance. One-year and all time averages change too slowly to indicate these problems. I also see the same is true for over-the-board games at the Pretoria Chess Club, where I am secretary, although there are no sandbaggers there. When players suddenly drop in club ratings, there is usually a problem of some sort. Mostly it is just real life interfering with chess- I am a police officer and our general election is coming up on 29 May. I have regularly been working 16-18 hours a day since April, including weekends. I have had to forfeit quite a number of games in the current club competition, and in other games I have been so tired that I just couldn't concentrate and lost, when I should have won. I have dropped to a lower division already (from B1 to B2), and I believe by June I'll be dropping well into B3. However, as I explained, this is a consequence of my current predicament at work. I see the same with other players- school children who drop just before and during exams, financial year end for others, and so on. The short average at the club reacts rather quickly because we are a small group (about 60 players) with much less games than people usually have here. There is also the difference between the 90-day, 1 year and all time averages that can be a significant indicator of change. Have a look at my own profile. I used to be a sub-1300 player for over a decade. In 2019 I decided I want to improve, and I joined the Pretoria Chess Club to find a tutor. At the moment my all time average is 1208, my one year average is 1412 and my 90-day average is 1382. I peaked at over 1500 a while ago, and now I am performing exactly at the level where I am performing at the Pretoria Chess Club. My one-year average is more or less in line with my 90-day average (within +-50) but the all time average is around -200! Lucky for me, I contacted several opposing captains before I joined the Pretoria Chess Club and informed them of my intention, because if I see such a vast difference and such a sudden improvement in any unknown player I would be skeptical. A good explanation will be very helpful, though. For this reason I advocate that team captains should be able to contact a team player quickly and easily, and be involved in the lives of their players.... not being intrusive, of course! I do not disregard the 90-day average as inconsequential. It has helped team SWAT management a lot. However, I do not rely solely on the 90-day average and consider other statistics as well. Relying on a single statistic would probably be a huge error in judgement. I do not think any single captain can consider all the statistics available to make matches, but one can select a few pertinent statistics to consider, and if these set of statistics prove to be undecided, more statistics can be considered in a list prioritized by each team. The general parameters set by most captains are useful and acceptable to me. Some captains consider another set of statistics, and when I comply to those I have found that team SWAT usually draws the short end of the stick. On average one should have an outcome of roughly 50/50 either way, but I have accepted challenges from a specific team where the outcome was 85/15 in their favor. For a while I tried to send challenges which seem more fair to me, which were declined, and I have since stopped setting up challenges with that team. The result was a recent reign at number 1 for several days for team SWAT. The moment one of the co-captains accepted challenges from that team, we have dropped outside of the top 10 again. We can discuss this analyses of statistics a bit more, but right now I have to prepare a brief for my commander who will be at the office within the hour. I can pick up on this discussion later. |
||||||
|
@Riaan |
||||||
|
90 Day Max So players with a lot of games will enjoy a correspondingly higher max 90. The average 90 is a far better measure. I think the max is also interesting—like why does this 1400 level player have a 1900 max? That’s a bit of an alarm bell. I can’t answer for Riaann, but this makes good sense to me. Of course, the team score as Mike notes has to be the most critical element. I do understand players might be a bit more careless in non team games. Or in tournament games where they’ve already lost a few. And some of the few tournament victors I’ve examined seemed suspiciously underrated. |
||||||
|
|
||||||
|
Sine Wave Boost |
||||||
|
However, if a 1400 player with 10 000 games suddenly beat a 1900, it is another kettle of fish altogether. I'll download that particular game and have a look at it in my laboratory- that takes a few days, so a challenge is out of the question. When I have an answer I'll make a note on the player's profile. I'll also try to download a number of other extraordinary games and analyze them as well. If we have a cheater on our hands, I'll never nominate him for a game against my team. I'll share the detail in confidence with a few captains, who can also have a look and come to their own conclusions and opinions. The 90-day max must be considered with another set of statistics. Usually, it is a secondary set of statistics that I would consider if the primary set of statistics are inconclusive. Team SWAT also has a tertiary fall-back set of facts, but most of the time, by the time I have to rely on a third set of statistics, it is just not worth it to go to such lengths and I err on the side of caution- I decline. |
||||||
|
Primary statistics: 1. Current average (within +-50 either way) 2. 90-day average (within +-50 either way) 3. Number of games completed (at least 100) 4. Time-out statistics (less than 2%) 5. Streaks: excessive winning or losing streaks. I have a very long winning streak myself, achieved around the time I first went over 1300 many, many years ago. I didn't cheat as such, but I was strategic in how I managed my games- I played faster in the winning games and slower in the losing games. Right after the winning streak I had a massive losing streak again. Both still there. 6. Highest rating won against (no more than +150) Secondary statistics: 1. Lowest rating lost against- 90-day, one year and all time (I have a look at the particular game in question) 2. One year average 3. Latest losing streak (specifically); it may not be the same as the longest losing streak 4. Team rating 5. Rating floor 6. One year highest 7. 90-day highest Tertiary statistics: 1. All time average 2. All time high; 3. All time low; 4. Lowest opponents lost against; 5. Highest opponent won; 6. Mini tournament points; 7. Which team is he/she with? I have noted that there are 2-4 teams which always get the better of SWAT, no matter how well I research the opponents. They suddenly become strong enough to beat just about any player I have in their bracket. Our team outcomes are not around 50/50 as one would expect, but 13/2, 11/2, 9/1... One such team was disbanded a number of years ago for cheating. They were from the former Balkan states...... I can't remember the name. These days there are at least two more teams which I am less inclined to accommodate. As you can see, by the time I get to the third set of statistics it becomes rather petty and I rather decline at the end of the second set of statistics. The exceptions will be if I am specifically asked to accommodate the team, or if the opposing team captain and I are on good footing and accommodate each other as a rule. |
||||||
|
So far, it seens none of the respondents here REQUIRE a minimum fixed difference between the max's. I hope we hear from someone who does require it, and learn his/her rationale for it. |
||||||
|
Some thoughts... I think the current rating and 90-day high, together, are a better measure than the current rating or the 90-day high, alone. I used to be much more lightly involved in matchmaking than I have become. For years, I let my co-captains do most of the matchmaking. My instinct is that rating averages are much more significant than maximum ratings. For years, when captains referred to "the 90-day," I thought they were talking about the 90-day average since, to me, that statistic seems to be the more meaningful one. I like it best when *everything* matches up closely, but of course that is not always possible. If I were creating pairs all on my own and could only look at one statistic, that statistic would be the full-length rating graphs for the players, which I think contains the most information in one place. My second choice would be the 90-day averages. My third choice would be the one-year averages. [For low-volume players, I would rather see the one-year averages than the 90-day averages. But in general I have become less fond of the 1-year average than I once was because (1) I have a lot of fast players on my roster who like fast opponents, and the 90-day average is usually good enough in my opinion for fast players, and (2) the 1-year average is more likely than the 90-day average to contain timeouts that distort the average.] The Top rating (90 days) is much less likely than most other statistics to be distorted by timeouts. However, I try to check for timeouts directly. I think that any statistic by itself has potential to be misleading. For instance, the averages are very misleading in the case of new players whose ratings have been steadily climbing so far... |
||||||
|
Red Checkerboards Wouldn’t that be a sweet and easy modification? 🤷🏽♀️🏁 |
||||||
|
|
||||||
| |||||||
