Data Project for Classic Racquets

In the short period of time I've been on TT, the wealth of knowledge here has truly blown me away. To take advantage of the racquet historians here, I'm starting a personal side project to compile a comprehensive spreadsheet for the specs and playing characteristics of classic frames. It will be a fun longterm project, and the more people contribute, the cooler the results will be. I plan to leverage unsupervised machine learning algorithms to find natural groupings of similar racquets in terms of specs (similar to TW's racquet finder but specifically for classic racquets). The end product will be a dashboard where you can visually see not only clusters of similar frames based on their objective specs, but also based on similar playing characteristics (i.e. solid, not hollow, very flexible, etc.). A possible feature could also be plots that show the changes in the a model's specs over the years it was produced (i.e. the different Head Prestige models). I realize contradictions in specs may arise from different contributors, and that's not a bad thing; it'll serve as a platform to really pin down the correct information over time and promote continued discussion. Here is the link to the Google spreadsheet that you can edit - https://docs.google.com/spreadsheets/d/1gsDhqoKimtio8LnCrd2_pQ8HL0W4i0qQd4EE722ecf4/edit?usp=sharing
 

vsbabolat

G.O.A.T.
In the short period of time I've been on TT, the wealth of knowledge here has truly blown me away. To take advantage of the racquet historians here, I'm starting a personal side project to compile a comprehensive spreadsheet for the specs and playing characteristics of classic frames. It will be a fun longterm project, and the more people contribute, the cooler the results will be. I plan to leverage unsupervised machine learning algorithms to find natural groupings of similar racquets in terms of specs (similar to TW's racquet finder but specifically for classic racquets). The end product will be a dashboard where you can visually see not only clusters of similar frames based on their objective specs, but also based on similar playing characteristics (i.e. solid, not hollow, very flexible, etc.). A possible feature could also be plots that show the changes in the a model's specs over the years it was produced (i.e. the different Head Prestige models). I realize contradictions in specs may arise from different contributors, and that's not a bad thing; it'll serve as a platform to really pin down the correct information over time and promote continued discussion. Here is the link to the Google spreadsheet that you can edit - https://docs.google.com/spreadsheets/d/1gsDhqoKimtio8LnCrd2_pQ8HL0W4i0qQd4EE722ecf4/edit?usp=sharing
The Wilson Pro Staff originally came out in 1984 and was made in Chicago. Then Production switched to St. Vincent, WI, then Taiwan, and then finally to China. Back in 1984 Wilson did not have their swing index marketing BS. When Wilson released the Pro Staff in 1984 it came in 3 head sizes: Midsize, Largehead, and 125.
 
Also a spell of production at Donnay in Belgium, I believe contemporaneously with early St V.

I’d suggest you borrow some of the fine work already done by @Dino Lagaffe on this subject:

Cool thanks, for sure! There are plenty of online resources but I’m rather slow on my own. Data crowd-sourcing is the way to go. I’ll work on this some more this weekend. For now I still need to fulfill responsibilities for my real job haha.
 

Sanglier

Semi-Pro
This is an ambitious project and I wish you success, but I suspect the amount of input you are likely to receive from others will be disappointingly limited in the end. If you haven't already done so, have a look at tenniswiki, an even more ambitious project that someone started more than 10 years ago, with the aim of capturing everything there is to know about every piece of tennis equipment ever made, including images. However, other than the initial upload (mostly just model names from the best-known brands) by the owner/admin, and some spambot posts in a virtually empty discussion forum, no one appears to have touched this platform over the course of an entire decade. https://thetenniswiki.com

Crowdsourcing works in some situations but not in others. I think part of the reason tenniswiki hasn't attracted participants the way wikipedia has is that the latter is of general interest to an infinitely wider audience than the former, and covers many topics where conflicting views are common. There is often an emotional element that compels the editors to ensure that their version of the truth is told and preserved against those of their challengers (consequently the discussion pages are frequently as interesting and illuminating to read as the articles themselves). I'm not sure that the average racquet-geek (especially the more "mature" and sedate type populating the "Classic" sub-forum) is emotionally invested enough in the playability and RA value of his/her favorite toys to want to defend such data on an open platform, where some equally anonymous stranger can come along at any time to change or erase his/her input without any justification. And if they couldn't be bothered to defend their contributions, what would motivate them to submit them in the first place? Should there be a disagreement over a particular entry, how would you know that the version left standing in the end is the "correct" one? How meaningful or comparable are the "feel" data on different frames if they came from different contributors?

For something as esoteric as vintage racquet data, and as difficult to quantify as "playability" and "feel", I think it's probably more straightforward to keep the spreadsheet private and to fill it out using publicly accessible materials in the first instance, and your own experience in the second. Other than Dino's generous contributions mentioned above (which he clearly spent a lot of time and effort compiling), a subscription to USRSA (https://www.racquettech.com) will also give you access to a treasure trove of good stuff on older racquets that you could transcribe to your spreadsheet. There are probably quite a few of us who have compiled something similar over the years for our own reference and amusement, from which we would select bits and pieces to share with the community when a relevant topic comes up. It does take time and work to create a reasonably comprehensive database, but the result is more internally consistent than a crowdsourced version, I think, because the person creating it knows exactly where every data point came from and what standard was used for comparisons.

This isn't meant to pour cold water on your project, merely to suggest that the alternative approach, which you seem to believe will take too long, might actually be the better one in practice, irrespective of whether your spreadsheet is meant for personal or public consumption.

Have fun, and good luck! :)
 
This is an ambitious project and I wish you success, but I suspect the amount of input you are likely to receive from others will be disappointingly limited in the end. If you haven't already done so, have a look at tenniswiki, an even more ambitious project that someone started more than 10 years ago, with the aim of capturing everything there is to know about every piece of tennis equipment ever made, including images. However, other than the initial upload (mostly just model names from the best-known brands) by the owner/admin, and some spambot posts in a virtually empty discussion forum, no one appears to have touched this platform over the course of an entire decade. https://thetenniswiki.com

Crowdsourcing works in some situations but not in others. I think part of the reason tenniswiki hasn't attracted participants the way wikipedia has is that the latter is of general interest to an infinitely wider audience than the former, and covers many topics where conflicting views are common. There is often an emotional element that compels the editors to ensure that their version of the truth is told and preserved against those of their challengers (consequently the discussion pages are frequently as interesting and illuminating to read as the articles themselves). I'm not sure that the average racquet-geek (especially the more "mature" and sedate type populating the "Classic" sub-forum) is emotionally invested enough in the playability and RA value of his/her favorite toys to want to defend such data on an open platform, where some equally anonymous stranger can come along at any time to change or erase his/her input without any justification. And if they couldn't be bothered to defend their contributions, what would motivate them to make them in the first place? Should there be a disagreement over a particular entry, how would you know that the version left standing in the end is the "correct" one? How meaningful or comparable are the "feel" data on different frames if they came from different contributors?

For something as esoteric as vintage racquet data, and as difficult to quantify as "playability" and "feel", I think it's probably more straightforward to keep the spreadsheet private and to fill it out using publicly accessible materials in the first instance, and your own experience in the second. Other than Dino's generous contributions mentioned above (which he clearly spent a lot of time and effort compiling), a subscription to USRSA (https://www.racquettech.com) will also give you access to a treasure trove of good stuff on older racquets that you could transcribe to your spreadsheet. There are probably quite a few of us who have compiled something similar over the years for our own reference and amusement, from which we would select bits and pieces to share with the community when a relevant topic comes up. It does take time and work to create a reasonably comprehensive database, but the result is more internally consistent than a crowdsourced version, I think, because the person creating it knows exactly where every data point came from and what standard was used for comparisons.

This isn't meant to pour cold water on your project, merely to suggest that the alternative approach, which you seem to believe will take too long, might actually be the better one in practice, irrespective of whether your spreadsheet is meant for personal or public consumption.

Have fun, and good luck! :)
Appreciate the feedback and thanks for the suggestions. Yeah, think I'll change the spreadsheet permissions to just viewable. Was hoping crowdsourcing might work, but understand people wouldn't want to do work for others for free, haha. Of course will still share the results when the project reaches a good place. I think what will set this one apart will be the UI and visualizations that people can access. The subjective descriptions (hollow, solid, firm, flexy, crisp, etc.) can actually be extracted via web-crawling this forum pretty easily and then using some basic word-gram analysis, but we'll see how difficult that actually is. Stay tuned. I've done a side data project with basketball before too, but this will be on another scale altogether.
 
Last edited:
For the subjective descriptors of playability, I think word cloud representations might be best. If "solid" is mentioned very frequently in association with a racquet, that'll be represented in the word cloud in larger size, etc.
 

onehandbh

Legend
Strung weights without overgrip.
From LEFT to RIGHT:

14.0 oz / 397g : Wilson Advantage (Medium L5) : 396g / 14.0 oz
13.1 oz / 371g : Wilson Jack Kramer Midsize
13.3 oz / 378g : Wilson Jack Kramer Autograph (Medium L5)
12.7 oz / 360g : Wilson Jack Kramer Prostaff (Light)
13.6 oz / 385g : Wilson Jack Kramer Prostaff (Light L5)

I have not hit with all of these yet. I will add playing characteristics after I get a chance to hit with them all.
 

onehandbh

Legend
Some more:

Prince Spectrum 90
12.8 oz / 365g (strung weight)

Wilson Hyper Pro Staff 85
11.6 oz / 329g (unstrung weight)

Wilson Pro Staff Limited 95 (17mm beam)
12.5 oz / 355g (strung with overgrip)

Head Pro Tour 630
12.5 oz / 356g strung weight with 1 overgrip
 
Top