Over the past couple of weeks the USTA has sent an inordinate number of email messages to its membership heralding the imminent arrival of the World Tennis Number. For anyone who has ever bothered to register with the ITF, the World Tennis Number is already here. It appears on the ITF player profile page. In order for that to work, the ITF has to be able to receive information from all its member organizations in some standardized format. That is where the Tennis Open Data Standard (TODS) comes in.
This post was prompted by a comment left on “The FMLC Mullet” by Charles over at CourtHive. In that comment he asked if I had considered the implications of the TODS as it underpins the World Tennis Number. The high level answer is that I was aware that TODS existed, but had in fact not given it much thought. As the Fiend at Court is a daily tennis diary, today’s post is a summary of my exploration resulting from that inquiry.
The ITF set up a public Confluence page for the TODS. Confluence is a collaboration tool that is widely used in software development projects. The source code and data schemas for the TODS are published to a GitHub site. In other words, the ITF is operating the project just like a reputable open source development project. In fact, that is exactly what it is.
The TODS Confluence portal provides a summary of the objectives of the project as overarching imperatives.
Universal – TODS should be able to represent information relating to any tennis event, player, or ranking, either current or historical.
Open – Anyone can freely access and use TODS.
Neutral – TODS is not specifically tailored to any individual organisation, vendor, operational model or usage scenario.
Extendable – TODS must be able to handle additional types of tennis-related information as they become generally required and provide a mechanism for individual custom requirements (that are not included in the formal schema definition) to be used without compromising or affecting the more general interoperability requirements.
Flexible – Excluding some key fields, most data fields specified in TODS should be optional and need only be present and populated if such data is required by the requester.
Scalable – TODS should be usable in small one-off data transfers just as easily as they allow for regular automated high-volume data exchange.
Community-Driven – TODS should represent a collaboration between all interested parties and should be developed in accordance with the collective need of the whole global tennis community.
Usable – TODS are designed to facilitate automated data exchange via web service APIs and so must be compatible with this usage.
Objectives of TODS, ITF Confluence Portal
Fundamentally, TODS is a data management and exchange protocol. The data structure is public and developers and interested parties can examine the current schema via a download off the Confluence site in either JSON or XML format. I personally don’t have the bandwidth to perform a detailed technical evaluation of those work products at the moment, but it is reassuring to know that any interested party could.
In “World Tennis Number: Second Serve” I mentioned that I asked questions during an earlier USTA informational webinar on this topic. Specifically, I asked about the management and ownership of the database of match results required to perform the World Tennis Number calculations. The TODS Confluence and GitHub pages essentially answer those questions.
Based on this first look at TODS, there is a lot that the USTA can learn from the ITF standard. First and foremost is that transparency is critically important to a large scale data management project of this nature. Even a cursory look at TODS reflects a well structured schema. The project also enjoys effective configuration management and a public change history.
In the absence of access to the equivalent information that underpins the USTA Serve/Play Tennis platform, we can only speculate on the underlying data architecture. As revealed through some of the rankings calculations errors, the data structure in that area appears to be horrific.
For the World Tennis Number to work, as apparently it already does, there has to be some export/data exchange mechanism with the ITF. Unfortunately, based on the limited publicly available information and errors that keep turning up in the USTA system, my own comfort level is low. For the World Tennis Number to work correctly for USTA players, the source data coming from that federation has to be accurate.
Under the legacy TennisLink system, I regularly scraped data off the public website using python and “beautiful soup” for data analysis. For example, I used the USTA’s own data to answer the question of whether or not splitting NTRP tournament divisions by age increased participation. (Spoiler alert! It did not.)
That analysis was possible because the data was structured well enough under TennisLink to make the coding effort for the web crawler relatively straight forward. It would be significantly harder to do under the new USTA platform primarily because the data presentation is disjointed. I don’t know if it is intentionally that bad to thwart 3rd party sites that scrape data for profit (like all of those unofficial NTRP ratings estimators) but that is a possibility. The World Tennis Number, underpinned by TODS, will likely once again make that possible.
This weekend I will be writing about the various philosophies of ranking systems and how it influences player behavior. Understanding that is essential for creating a robust tournament framework that ultimately benefits the entire tennis ecosystem.
The deployment of the World Tennis Number will push tennis into an unprecedented era of public and widespread access to player performance data. It is hard to predict the future on this topic. However, we should certainly try to learn from the past.
- Tennis Open Data Standards, ITF Confluence Portal, viewed May 14, 2022.
- Tennis Open Data Standards, ITF GitHub Site, viewed May 14, 2022.
In the paragraph just under the table, in the second sentence, “via a downloaded” should be “via a download”.
On the topic at hand, I find the UTR to be more accurate than the USTA’s current system. Do you know it stacks up to the World Tennis Number?
Please contact me charles@courthive.com as I’d like to understand your perspective on the structure of the TennisLink data vs. the structure of the data coming from the new system. I also have a bit of experience working with ranking systems of a number of other governing bodies.