Questions about TennisRecruiting.net

[TennisFan2Day]

HI Dallas,

Per your request I am starting a new thread and I have attached the last part of our conversation to this one.

Is there any difference betweeen winning a match against the #1 player in the ITf and #1200 since they are both international players?

With all due respect, I am surprised that you asked this question. All head-to-head systems handle each player individually - there are no "groups of players" - just like there are no "classes of tournaments".

TRN uses a head-to-head system. All players and results are loaded into the system, and then the system assigns a rank value to all players. The system does not have any idea which players are domestic or which players are international.

International players end up with rank values just like domestic players.

Every single player with results ends up with a rank value. The longer the record, the more confident we are about the rank value. That is why we insist on a minimum number of tournaments and a minimum number of wins for a player to receive an actual ranking.

Back to our main disagreement... we feel that our rank values for international players are interesting. While we do not always get the fine details correct (e.g., you could probably argue that the #20 international player is better than #25 player), the general rankings are quite good and very predictive. Those international kids from South America that you referred to earlier do not have very high rank values in our system. But, for the purposes of ranking the U.S. kids, we think that using these international results enhances our rankings.

How does winning against an international player compare with beating a blue chip, 5 star, 4 star, etc?

Answered above, but just to reiterate... players are handled individually rather than in groups. We do not lump 5 Stars together any more than we lump international players together.

Personally I don't even think it is arguable that the #20 international player is better than #25 player.

I pretty much know that if I have a player who is a 5 star and he beats a player who is ranked way above him that he is going to move up in the rankings as long as he hasn't lost to somebody who is ranked lower than him. Provided that he doesn't have anything significant dropping off.

The problem lies in that nobody, outside of your group, knows where an international player sits in the rankings. So, how does a player know which wins really count over an international player? I know that beating a bunch of 1 stars will not help my player's ranking.

I'm just going to use an easy example. At the Orange Bowl, Shane Vinsant(2012) lost to Oliver Golding who is an international player. How would anybody know if he lost to a player who is not ranked very high or an incredible player? I would assume that this loss didn't hurt Vinsant since Oliver is the #1 junior British player.

International players end up with rank values just like domestic players.

If International players end up with rank values just like domestic players why not just show them?

Quoted from your FAQ, "One factor in the College Reqruiting list rankings is the average of the opponent's quality of a player's best 8 wins." Isn't there a way that you could just highlight the wins used in each players ranking? This would simplify things a great deal. This would eliminate your need to rank the international players on your site and it would provide some sort of clarity on which international wins count and which do not.

I am kind of shocked that there is no weight to tournaments. Ask any top college coach and they will tell you that they are looking for players who can play under pressure. There is a lot more pressure playing in the quarters of a Supernational than playing in the finals of a local, even if you play the same player.

 

[ClarkC]

Quoted from your FAQ, "One factor in the College Recruiting list rankings is the average of the opponent's quality of a player's best 8 wins." Isn't there a way that you could just highlight the wins used in each players ranking? This would simplify things a great deal. This would eliminate your need to rank the international players on your site and it would provide some sort of clarity on which international wins count and which do not.

Sounds like a good idea. I understand that it might not be a higher priority than what the TRN staff is already working on, but it would be informative and would remove some of the uncertainty about results vs. international players as compared to domestic opponents.

 

[TennisFan2Day]

Thanks ClarkC. I guess Dallas is just ignoring this.

 

[ClarkC]

Thanks ClarkC. I guess Dallas is just ignoring this.

Well, he could be busy, you know.

They just released a major feature article (the women's winter team recruiting rankings) today.

 

[dallasoliver]

Apologies. I have been meaning to respond to this thread for a while... and I will (lamely) plead being busy. Let me fire off a few quick thoughts in response to what I think are very constructive points from TennisFan2Day...

If International players end up with rank values just like domestic players why not just show them?

This is a fair question. Basic answer is that we only use a subset of ITF tournaments and so will only have rankings for a subset of international players. We thought it would be confusing to rank only the international players who play regularly against U.S. competition. I don't have any data to back that up... We might revisit that question at some point, but I have a feeling that this suspicion will bear out.

Quoted from your FAQ, "One factor in the College Reqruiting list rankings is the average of the opponent's quality of a player's best 8 wins." Isn't there a way that you could just highlight the wins used in each players ranking?

This does indeed seem fair. Our ranking system is iterative (i.e., it runs through the data multiple times revising the rankings until the rankings converge), and I know that the system tracks the Top wins on each iteration. Right now we are not storing that data (it is only in memory while the ranking system runs). We could store that data and highlight the Top 8 wins on the Activity tabs of players.

I will talk with the other folks at TennisRecruiting.net about this feature, and I can what can be done. I agree that it would be a very good change that addresses the international issue without the problems I mentioned in the answer to your first question.

However, note that such a feature will not be any time soon - we have a number of irons in the fire right now, and I know that this feature will require changes to the ranking system, to our database, and to our webpage builders. All of those changes typically take a lot of time - so those costs will be factored in.

I am kind of shocked that there is no weight to tournaments. Ask any top college coach and they will tell you that they are looking for players who can play under pressure. There is a lot more pressure playing in the quarters of a Supernational than playing in the finals of a local, even if you play the same player.

All fair points. I suppose you can think of USTA PPR and the TennisRecruiting.net CRL as opposites on a spectrum. The USTA PPR only counts the quality of the tournament (with a small nod to the very top players with the Bonus Points list), while TR.net CRL only counts the quality of the opponent. There are certainly possibilities in between, but neither system really does anything there (with all due respect to the PPR bonus points list). Here are a couple of thoughts:

- We have gotten a lot of thanks from many parents whose kids only compete regionally for financial reasons. These kids only travel to one supernational a year when they can get financial help from their section. Those kids can still become Blue Chips or 5 Stars if they compete in the top tournaments in their area.

- Given a choice between the two, I like the lean of the TR.net CRL for predictability.


Again, great questions. Hopefully my ramblings help some.

Best,
Dallas

 

[ClarkC]

I am kind of shocked that there is no weight to tournaments. Ask any top college coach and they will tell you that they are looking for players who can play under pressure. There is a lot more pressure playing in the quarters of a Supernational than playing in the finals of a local, even if you play the same player.

This is often asserted, but it only affects the rankings if the results come out differently. How many players play each other enough times, in different level tournaments, so that you can definitely say that Player A has a much better record head to head against Player B when they face each other at nationals than when they face each other at sectionals? Examples, please?

 

[dallasoliver]

This is often asserted, but it only affects the rankings if the results come out differently. How many players play each other enough times, in different level tournaments, so that you can definitely say that Player A has a much better record head to head against Player B when they face each other at nationals than when they face each other at sectionals? Examples, please?

Great point. What he said.

- Dallas

 

[tenniscrazed]

Great point. What he said.

- Dallas

I have looked at TRN, Star, and PPR. And, although this is an informal analysis, it clearly appears to me that Star, and TRN are the closest to an actual head to head analysis and therefore are more reliable in their predictive properties. PPR is simply a collection of pennies with exponentially more pennies issued based on rounds reached. This eliminates any predictive value. I'm not saying that one type is better than another, my contention is that TRN and Star have greater predictive value over PPR.

 

[TennisFan2Day]

I've been away for a while. Thank you for at least considering some of the issues at hand.

I couldn't help but notice today that Taro Daniel from the Class of 2011 is a "new' Blue Chip. This is very striking considering that he hasn't played on American soil in the past year. There really needs to be something done about this international ranking thing. He has only played one Blue Chip and one 5 Star. While that is a great accomplishment I would think you would need more wins than that over ranked players to be a Blue Chip. I'm not saying that he isn't a good player, there is just no way to verify it.

Why don't you just use tournaments on U.S. soil since those are the only colleges that are recruiting?

Thanks,

Mike

 

[ClarkC]

I
Why don't you just use tournaments on U.S. soil since those are the only colleges that are recruiting?

Thanks,

Mike

Only American colleges are recruiting, but they are not just recruiting American kids, and even the American kids are not just playing on American soil. I don't understand the point of your question.

If an ITF is held in Mexico and a lot of the same kids show up who play ITFs and other big events in America, then the tournament provides a good basis for comparison among players already in the head to head database. Why would you argue for excluding such useful information?

 

[ClarkC]

Just a quick question instead of starting my own thread.
A certain player I beat has a 13-16 record. He's 0-1 against 5 stars and 3 stars, 0-2 against 2 stars, 4-12 against 1 stars, and 9-0 against unranked players. Yet he's a 2 star.

Now this other player is 41-28, is 0-1 against Blue Chips, 1-3 against 5 stars, 0-5 against 4 stars, 3-5 against 3 stars, 2-2 against 2 stars, 25-9 against 1 stars, and 10-3 against unranked players. And he's only a 1 star.

In a later post in the other thread, you stated that the 2-star is ranked #700 with an RPI of 1234. I have checked all ranking lists for boys, 6th grade through 12th grade, and in none of the lists is the player who is ranked #700 a 2-star recruit. All seven of them are 1-star recruits. None of them has an RPI even close to 1234.

I have a premium account on tennisrecruiting.net. I will be glad to look at the detailed results that are not available for free and analyze why two players are ranked where they are, but you have to provide accurate information first.

 

[ClarkC]

Just a quick question instead of starting my own thread.
A certain player I beat has a 13-16 record. He's 0-1 against 5 stars and 3 stars, 0-2 against 2 stars, 4-12 against 1 stars, and 9-0 against unranked players. Yet he's a 2 star.

Now this other player is 41-28, is 0-1 against Blue Chips, 1-3 against 5 stars, 0-5 against 4 stars, 3-5 against 3 stars, 2-2 against 2 stars, 25-9 against 1 stars, and 10-3 against unranked players. And he's only a 1 star.

Ok, 2 star RPI is 1234, 700 Recruiting.
1 star RPI is 884, 754 Recruiting.

In no age group does the #754 player have a record anywhere close to 41-28, nor an RPI of 884.

 

[TennisNinja]

[x]

 

[ClarkC]

2 star: http://tennisrecruiting.net/player.asp?id=680615 Actually 751 with 1324 RPI
1 star: http://tennisrecruiting.net/player.asp?id=623221 Actually 763 with 901 RPI

I kind of realized my mistake, I was looking at 2010 high's, not weekly.

The 2-star is actually 775, not 751.

As I suspected, your question is one that has been answered 1000 times here. The numerical rankings are updated every week. The star rankings are updated once a year. Before you bother to ask why, read the FAQs at tennisrecruiting.net and the threads on this board that discuss that site.

The 2-star has seen his ranking drop to 775, which is way out of 2-star range (401-600). If he does not improve his results, he will not be a 2-star next fall when the star rankings are updated.

Meanwhile, the 1-star has a higher ranking (763 vs. 775). It would be better for you to beat the player ranked 763 than to beat the player ranked 775, because the current numerical ranking is the key to the strength of your opponent.

 

[ClarkC]

The 2-star recruit has only a 4-12 record against 1 star players. However, he has apparently been playing most of his tournaments up one age group, because 12 of his 16 matches against 1-star players were against older players. In fact, 5 of those matches were against players who are three graduating classes older than he is. He won 1 of those and lost the other 4. All five of those players are well ahead of him in the overall strength rating across age groups, so he actually got some boost to his ranking from those 5 matches, for example.

One of the key components of your TRN ranking is your 8 best wins. While this player has been playing up, and his losses don't count much against him, his schedule has prevented him from having 8 good wins. He has a lot of impressively close losses to older players, but closeness of score is not a criterion. So, he has dropped from 2-star to 1-star territory because his wins from a year ago have been dropping off his 12-month window and are not getting replaced by wins.

 

[ClarkC]

Meanwhile, the 1-star recruit ranked #763 is three years older. His record is against players his own age. He is an 11th grader, so he cannot be playing guys three years older than he is in USTA tournaments, unlike the 2-star ranked #775. He has a record of 41-28, so he has 8 good wins, but they are against his own age group.

In the internal database at TRN, an 11th grader ranked 763 is WAY ahead of an 8th grader ranked 775. They are not even close. Don't get hung up on star ratings, which is what 90% of the critics of TRN do on this board. Age matters. A LOT.

 

[tenniscrazed]

To Clark C or Dallas;

The "playing up" issue is an interesting issue from a rankings standpoint. What has been your review of kids doing this where they are highly ranked but have lower RPI's? Where they have wins against various "star levels" in birth years as much as 2 or 3 years older. In reverse would be kids that are highly ranked that are having 5 star and blue chip wins but against equal or younger birth years?

 

[TennisNinja]

Thanks for all your help Clark C.

 

[TennisTaxi]

Dallas...

The re-design of the website looks great!!