Page from the Sandbagger's playbook

[maverick1]

I thought the USTA's dynamic NTRP algorithm was supposed to be a guarded secret so that sandbaggers couldn't game the system. But apparently, any Team captain who has been around can get hold of it, if I can. 15 months ago, I had never heard of NTRP or USTA.

------------------------------------------------------

A 4.0 player can have a dynamic rating between 3.50 and 3.99. As soon as you hit 4.0, you are automatically bumped to 4.5. Between 4.0 and 4.05, you can appeal, and they would automatically grant your appeal. After 4.05 (4.06 or higher), you better have a convincing reason (usually medical) or evidence of error in the results or something of that nature.

Now, let's say you start a season as a medium 4.0 player with a rating of say 3.70 and you play a self-rated 4.0 player with an actual dynamic rating of 3.5 (lowest 4.0 rated player possible). If you end up winning the match with an average score of 6-4 (either 6-4, 6-4, or 7-6, 6-3)., the computer takes the difference between the average games won (6) and lost (4), which in this case is 2 and multiplies it by the dynamic rating of your opponent /100 and adds the result to your rating. This now becomes your new or “current” dynamic rating. In this example, 6-4 = 2, then 2 * 3.5 /100 = 0.07, then 3.7+0.07 = 3.77 is your new rating after beating a 3.5 player 6-4, 6-4. If you beat that same player 6-0, 6-0, your rating goes up by 6*3.5/100= 0.195 and your new rating will be 3.7+0.195 = 3.895. The reverse happens when you lose a match. i.e., your dynamic rating decreases by that much. As you are playing matches throughout the season, every match resulting in your dynamic rating staying above 4.0 generates a strike. 3 strikes and you are out. That is why it is important that (if you can control it) not to win 6-0, 6-0, because it really could hurt your rating or get you DQ'ed.

So, what I am saying is that your “dynamic” rating could be already at 3.95, due to your previous wins in last season, and you get disqualified by a couple of more close wins, but someone with a lower dynamic rating, say a self-rated 3.5 player, who has more wins and more convincing wins than you, does not get disqualified because it takes time for his rating to reach the threshold where strikes are generated against him.

 

[Ripper]

Where I live, you're either a C, a B or an A. When you sign up for your first tournament, you do so as a C. Then, maybe, progress "all the way up" to A. Or stay in C for ever, either by sandbagging or sucking

 

[Geezer Guy]

From what I've seen, players that self-rate are placed at the top of the range, not the bottom. I've seen several self-rated 3.5 players (and they were indeed "true" 3.5 players in my opinion) get bumped up to 4.0 after just a couple of fairly close wins over other 3.5 players. At the 4.0 level, they got thrashed and eventually bumped back down.

 

[andrew_b]

From what I've seen, players that self-rate are placed at the top of the range, not the bottom. I've seen several self-rated 3.5 players (and they were indeed "true" 3.5 players in my opinion) get bumped up to 4.0 after just a couple of fairly close wins over other 3.5 players. At the 4.0 level, they got thrashed and eventually bumped back down.

Here's my main gripe with the "dynamic" NTRP system: It's possible to be bumped up during the season, meaning your wins no longer count for your team. Yet, it's not possible to be bumped down except for once a year, when the annual ratings come out.

At least, that's my understanding of it. It's only "dynamic" in an upward direction.

play well,
Andrew

 

[maverick1]

Here's my main gripe with the "dynamic" NTRP system: It's possible to be bumped up during the season, meaning your wins no longer count for your team. Yet, it's not possible to be bumped down except for once a year, when the annual ratings come out.

At least, that's my understanding of it. It's only "dynamic" in an upward direction.

play well,
Andrew

I agree it is ridiculous that you can be bumped in the middle of the season and results reversed. Once you start the season, you should be allowed to finish it. At the least they should make this allowance for players who have played the previous season. I don't have any experience with disqualification of anyone I know. It is quite possible that DQs never happen to players who played the previous year, and my whining is moot.

I don't particularly care whether bumping down is symmetric. It has no consequences since the demoted player is still eligible.

 

[maverick1]

From what I've seen, players that self-rate are placed at the top of the range, not the bottom. I've seen several self-rated 3.5 players (and they were indeed "true" 3.5 players in my opinion) get bumped up to 4.0 after just a couple of fairly close wins over other 3.5 players. At the 4.0 level, they got thrashed and eventually bumped back down.

If the algorithm I found is correct, I don't think it can be true that self-rated players start at the top of the range. I started as a self-rated 3.0 this year, won all my matches until the states where I lost in doubles. In my 3 singles matches, I never lost more than 3 games in a set, including against one player who won his #1 singles matches against every other team. Just beating him 6-1 6-2 should have been worth about .13 points.

I am actually quite disappointed at not being bumped up. I am going to play 3.5 anyway next year.

 

[Geezer Guy]

Well, maybe it works differently in different places, or maybe it's changed over time. I'm on a 3.5 team and a couple of years ago we added some new players who self-rated themselves at 3.5. I think most of these guys were truely 3.5 - they played very even with the rest of us. There was one guy that played better than the rest of us, but didn't totally blow us out. We played a couple of matches and our team did very well against the other 3.5 teams. Then, about a month into the league, all the self-rated 3.5 players got bumped up and all of the computer-rated players stayed the same.

Needless to say, losing all those players in the middle of the season was pretty hard on our team. The one player that was the best was able to get on a 4.0 team and did OK but not great. The other players got smoked trying to play 4.0.

 

[maverick1]

Sorry to hear that.
I wonder if the bump-ups were triggered by opposing teams' whining. The league I played in was very gentlemanly. The captains all know each other and are good friends.

 

[Geezer Guy]

Humm... That's possible, I guess. I hadn't thought about that. As a possibly related note, our Captain had mistakenly (he claims) brought in a player with the ugliest strokes you'd ever seen. He played pretty even with the rest of us, but it was painfull to see him across the net. Anyway, he's on our 3.5 team and is winning matches. One of the opposing captains googled his name and found out he played college tennis. (He hadn't told our captain he played college ball.) This was the first (or maybe second) year of that "Elite Player" rule, and our captain didn't know about it. He was the lowest guy on the college squad, and had only played 1 college match (their last match of the season - which meant nothing so the coach put him in - and he lost). BUT that was good enough to make him an Elite Player, get him bumped up to 4.5, and get our captain put on a 2 year probation. So, knowing our captain had a "history" of this sort of thing, maybe that made opposing captains a little more suspicious of any success we had.

 

[SunDog]

That formula cannot be correct. It doesnt take into acount the rating differential between oponents. That method as outlined, would punish higher rated players (even if the player rating differential remained the same) by applying a bigger correction for a given match score. If, as in your example, you substitute a 4.70 player vs a 4.50 player, then for the same match score, the winning players rating would be adjusted +0.09 (vs +0.07). This method does not take into acount what the score of a match SHOULD be for a given player rating differential. In other words - a match with a wider player rating differential should result in a more lopsided score. If it does, then little to no dymanic correction will take place. If a match with a large player rating differential does not end up in a lopsided victory for the higher rated player, then one would suspect that there would be more of a dynamic correction.

Also - in the Southern Section, only dyamic DQs of self rated players result in a reversal of all that players court victories. Dynamic DQ of competer rated players only results in a reversal of the third strike match and any subsequent matches until notification has been made (the following tuesday - i believe).

 

[maverick1]

That formula cannot be correct. It doesnt take into acount the rating differential between oponents. That method as outlined, would punish higher rated players (even if the player rating differential remained the same) by applying a bigger correction for a given match score. If, as in your example, you substitute a 4.70 player vs a 4.50 player, then for the same match score, the winning players rating would be adjusted +0.09 (vs +0.07). This method does not take into acount what the score of a match SHOULD be for a given player rating differential. In other words - a match with a wider player rating differential should result in a more lopsided score. If it does, then little to no dymanic correction will take place. If a match with a large player rating differential does not end up in a lopsided victory for the higher rated player, then one would suspect that there would be more of a dynamic correction.

Well, I can't vouch for the formula beyond saying that I got this via email from some long time 4.0 players.

What seems even more wrong to me is if a 3.5 player plays up in a 4.0 league and wins half his matches against 4.0 players, his rating could go down.

 

[oldguysrule]

Thanks for the post Maverick.

SunDog is correct that an additional element to what Maverick posted is a comparison of the actual player differential when adjusting the dynamic NTRP. The scores are used to project what the "expected" player differential would be and then an adjustment is made to the dynamic rating accordingly.

Two other elements are missing that I know of. First, self-rated players are not used in the system until they have played 4 matches with computer rated players. This will create the first dynamic rating for the self-rated player. Secondly, if the actual rating differential between two players is large enough so that a "competitive" match doesn't occur, the results are not used to adjust the dynamic rating of either player....essentially, the results are thrown out.

Yes, I am sure some people try to "work" the system. To me, that takes away from the fun of playing.