PDA

View Full Version : Shanghai: Slam Bracket Prediction Contest Pilot



James7
10-06-2015, 05:45 AM
Entry Form:
http://www.talkabouttennis.com/contests/SlamTest/brackettest.php?tour=ATP&tournament=Shanghai

Picks Tracking (Flat 1 point per correct pick):
http://www.talkabouttennis.com/contests/SlamTest/picks_res.php?tour=ATP&tournament=Shanghai

Picks Tracking (Progressive Scoring):
http://www.talkabouttennis.com/contests/SlamTest/picks_res_alt.php?tour=ATP&tournament=Shanghai

What is it?
This is an additional contest to be run at Slam events for both WTA and ATP. It does not replace anything and is intended to be for fun as well as potentially awarding medals. Since this is newly built, we'd like to run a pilot of this event during a Master's tournament to test it out and also to elicit feedback.

Basic Concept
To provide a fun standalone full bracket prediction contest to complement other ongoing contests at Slam events. The only requirement for participation being a TAT account. The majority of achievement should be the ability to get as close as possible in successfully predicting the results of the entire tournament. Bonus points are added for guessing game totals of later rounds to provide some late-tournament action when most predictions action is locked-in.

How does it work?
After the bracket is finalized and prior to the first round of play, participants select match winners for the full bracket. Additionally, participants are asked to guess how many games will be played in each of the QF, SF, and finals matches. Brackets for WTA and ATP events are separate.

Points are awarded as follows:
Each correct bracket pick (having the correct player in that part of the draw) awards one point for each and every slot correctly filled. This forms the majority of the prediction score.
Bonus points are awarded for how close you come to game totals in each of those predicted matches. 5 bonus points for getting the exact game count in that match, 4 for being within one game, 3 for being within 3, 2 for being within 5, and 1 point for being within 7. If a retirement takes place, the number of games played until the retirement is the 'correct' answer for that match, and a walkover is treated as playing 0 games.
Prediction and bonus points are added together to form a final score for the event.

There will be a pick tracker so you can follow scoring as it occurs, complete with some plots of the data and breakdowns of choices in real time.

What do I need to do to participate?
Nothing beforehand. Entries will need to be made before the start of main draw play and a thread will be up in order to do that at least one day prior to the start of first round play.
This is intended as a fun pilot of the prediction contest, so please share any and all feedback you may have. Including ideas for re-tooling, aesthetic changes, "I don't like this thing about it", etc. I'd like your help in building something we can all enjoy.


Thanks!

James7
10-07-2015, 08:09 AM
One possible complication with the pilot on Shanghai: I notice there is the possibility of having overlap with qualies and first round play. If that is the case, we will either be predicting with 'empty' qualifier slots as possible choices, or I may just extend the entry to the next day and give virtual freebies for some first round matches. As this is just for fun at this point (as a small stress test of my scripting skills) I don't think it matters. Any strong preferences?

I'll try to tackle and additional questions here as they arise. Please also post feedback comments here as well.

Thanks!

JTContinental
10-07-2015, 08:53 AM
Looking forward to the pilot launch!

James7
10-09-2015, 12:33 PM
Minor update regarding entry.

A schedule for Shanghai still isn't available, but guessing by the Schedule from last year, the deadline will either be Start of play Monday (Oct 12) OR Start of play Tuesday (Oct 13), depending on the qualifying match schedule.

A bracket entry form link will be posted after qualifying matches complete and entries will need to be submitted before the start of play the next day.

Sorry for the uncertainty. I will quote hard deadlines once the tournament schedule becomes more apparent.

Plan on a Monday deadline that will likely translate to Sunday evening for many.

James7
10-10-2015, 06:00 AM
Looks like qualifying will be completed after Sunday play, which will also include 2 main draw matches. So two "free" match picks there.

This bracket will periodically refresh, but will NOT auto-fill picks past the first round.

Links will be in the first post now as well but here they are:

Entry Form:
http://www.talkabouttennis.com/contests/SlamTest/brackettest.php?tour=ATP&tournament=Shanghai

Picks Tracking:
http://www.talkabouttennis.com/contests/SlamTest/picks_res.php?tour=ATP&tournament=Shanghai

Note, you may pick anytime now, but Qualifier picks will be problematic until they are actually placed. The deadline again will be Start of Play Monday.

fastbackss
10-10-2015, 07:28 AM
Hello, thanks for putting this together. I presume if we submit another bracket it will overwrite the previous?

This is so much fun - so I had some fun with a couple of my picks. (well at least early on :-))

PS - I think it would be interesting after the fact to check the scoring via other means, like the NCAA tournament for example (points double with each round, or go up 50%, or only are worth more for the quarters, etc)...

James7
10-10-2015, 07:48 AM
Entries are tied to your TAT login. So if you re-enter it will indeed overwrite your previous one. So no problem if you want to redo things as many times as you want.

Submissions will automatically close off after the deadline, however. So no redoing after that point :) (Though, that was something I was considering... the ability to redo entries as the tournament goes on at a growing penalty, but I didn't see how that would really work).

Our existing prediction contest does progressive scoring (points go up each round). I was thinking about it, but thought that with the single point scoring it adds extra weight to your ability to pick overall. So even the first round match results are very important. Part of this is to explore how well this works and retool. If people like progressive scoring that can be done.

Thanks for the feedback!

James7
10-11-2015, 04:31 AM
The bracket is actually finalized now, though there is still one first round match of the two scheduled being played providing a freebie pick.

Note that the deadline for entries is 1:30 PM Monday Shanghai local time (1:30 AM Monday EDT USA, 10:30 PM Sunday PDT USA)

Remember to get your predictions in! Thanks!

ptmcmahon
10-11-2015, 05:46 AM
Woo hoo! Tied for first...we're all one point ahead of JTContinental ;) (assuming that was a Verdasco over Tomic pick?)

Kirkus
10-11-2015, 10:27 AM
I resubmitted a second bracket to include # of games (I forgot the first time). I hope that was OK to resubmit.

Verdasco and Dolgopolov. LOL. I'm not submitting a third one. :)

James7
10-11-2015, 03:47 PM
Less than 8 hours left to get your entries in. :)

fastbackss and JTC may want to update their entries as well, even though there are not any real stakes to this.

Kirkus
10-11-2015, 04:00 PM
OK. 3rd submission. I'm done now. :D

JTContinental
10-11-2015, 04:18 PM
Less than 8 hours left to get your entries in. :)

fastbackss and JTC may want to update their entries as well, even though there are not any real stakes to this.

I'm too lazy... :D

Kirkus
10-11-2015, 04:48 PM
I'm too lazy... :D

It's predicting that first round that takes it outta ya. LOL.

James7
10-12-2015, 04:30 AM
Heh. And you had Byes and everything here. You certainly won't like the Slams then where there's double that in the first round and no freebies.

----

For clarity:
Pick scoring is happening somewhat automatically. Refreshes are not constant, so it may take a little while for a completed match to be scored (I scheduled updates every hour or two during approximate play times). A Correctly picked match will be Bolded in the table. A Mis-pick will get a strikethrough. An uncompleted match is just regular text.

Even though it won't change the cursor, column headers in the picks table ARE clickable, and will pop-up an new window with a graph of selections for that slot.

As mentioned before, any and all feedback and recommendations are welcome and we can discuss how to alter the contest to be more interesting for the slams, or what does or doesn't work well.

Thanks again for participating!

fastbackss
10-12-2015, 05:44 AM
I'm too lazy... :D
Haha - I actually thought the same thing. Then I felt guilty, so I re-did bracket. After that I found out about the 2 freebies...and then I was really too lazy!

Drop-shot
10-12-2015, 05:51 AM
Heh. And you had Byes and everything here. You certainly won't like the Slams then where there's double that in the first round and no freebies.

----

For clarity:
Pick scoring is happening somewhat automatically. Refreshes are not constant, so it may take a little while for a completed match to be scored (I scheduled updates every hour or two during approximate play times). A Correctly picked match will be Bolded in the table. A Mis-pick will get a strikethrough. An uncompleted match is just regular text.

Even though it won't change the cursor, column headers in the picks table ARE clickable, and will pop-up an new window with a graph of selections for that slot.

As mentioned before, any and all feedback and recommendations are welcome and we can discuss how to alter the contest to be more interesting for the slams, or what does or doesn't work well.

Thanks again for participating!

THIS is awesome.

JTContinental
10-12-2015, 09:49 AM
Heh. And you had Byes and everything here. You certainly won't like the Slams then where there's double that in the first round and no freebies

I'll have a couple of months to train and build up endurance. :D

The real reason I was too lazy is that we tiled our kitchen this weekend, so spending any additional effort on anything else seemed exhausting. :D

ptmcmahon
10-12-2015, 10:31 AM
Still tied for first! Graph selections are fun.

James7
10-12-2015, 12:03 PM
Made it more visually obvious that the column headers are clickable (cursor and color indicators).

Everything here seem good as far as the bracket entry and picks summary go? Any suggested changes to any of that?

James7
10-13-2015, 12:29 PM
Looks like some people lost a projected winner. Many a projected finalist.

I wonder how tricky it would be to program a "maximum points" projection... as in how many points from here it's possible to achieve. Or maybe the Bonus point section makes that useless.

Should I try experimenting with progressive point scoring here? Maybe as a separate section for contrast to help decide?

Any other thoughts? Anything people would like to see?

ptmcmahon
10-13-2015, 04:34 PM
Was just thinking projected maximum would be interesting...like you say may have to omit the # games predictions.

ptmcmahon
10-13-2015, 05:17 PM
Should I try experimenting with progressive point scoring here? Maybe as a separate section for contrast to help decide?

Sure...then well see how different it is. 1 point per match no matter what would separate it from predictions...but could mean late tournament round action won't affect standings...so hard to say what's best.

James7
10-15-2015, 06:55 AM
Ok, let's take a look and see what we think:

Single point scoring:
http://www.talkabouttennis.com/contests/SlamTest/picks_res.php?tour=ATP&tournament=Shanghai

Progressive Scoring test (first round 1 point, each subsequent round increases the points per correct slot by one):
http://www.talkabouttennis.com/contests/SlamTest/picks_res_alt.php?tour=ATP&tournament=Shanghai

Also added an "In Play" column that says how many more prediction points are possible for the participant based on who has lost already.

Let me know what you think. Or if you think the points should be tweaked. Or whatever your thoughts are.

James7
10-15-2015, 07:51 AM
I should also add that there are two "extra" picks brackets in the mix in case it wasn't already noted.

"TennisAbstract" comes from the tennisabstract site projections for the tournament (At the same point our deadline was). Highest likelihoods for each slot.
"RandomBracket" was just using the randomize button.

I've had the suggestion to add something from ProPelicanTennis.

I've also thought about adding in a meta 'most popular' bracket after the selection deadline. I hope that there would be a bunch of entries for Slams making that at least somewhat interesting.

Any other thoughts for 'flavor' are welcome.

ptmcmahon
10-15-2015, 11:37 AM
Ok, let's take a look and see what we think:

Single point scoring:
http://www.talkabouttennis.com/contests/SlamTest/picks_res.php?tour=ATP&tournament=Shanghai

Progressive Scoring test (first round 1 point, each subsequent round increases the points per correct slot by one):
http://www.talkabouttennis.com/contests/SlamTest/picks_res_alt.php?tour=ATP&tournament=Shanghai

...either way James is first.... FIX FIX! Fix I tell ya :D

Drop-shot
10-15-2015, 11:40 AM
...either way James is first.... FIX FIX! Fix I tell ya :D

As long as Omegadoom isn't winning, it works perfectly :cheesy:

ptmcmahon
10-15-2015, 11:58 AM
*checks medal counts* I stand corrected!

James7
10-15-2015, 12:01 PM
Lol. :cool:

I honestly didn't expect to, I swear. I'm seeing a trend where I only do well as long as there is absolutely no reward at stake. As soon as there would be a medal up for grabs I'm certain to fail.

I'm thinking the progressive scoring may be better overall. More separation, and the Random Bracket is more pleasingly behind where it should be. Wold have been further behind if I didn't give it the byes and freebies.

This was just with 1 additional point per round, so the first round picks still carry a good amount of weight which is what I was hoping for (less throwaway compared to more aggressive scaling). And more late round action is a good thing.

But I do want to know what others think. I'm willing to try other things. I'm completely serious that I want people's input on how to design it.

*Tweaked the max score projection a bit to include possible Bonus and overall totals.
*Should I also pre-strike earlier round losers in the summary table since I'm now figuring that out for scoring projections?

Drop-shot
10-15-2015, 07:00 PM
I like progressive scoring more.

fastbackss
10-15-2015, 07:29 PM
I will add more feedback after the event has concluded, but I am confident the sentence will start with "I like progressive scoring better."

That being said I think getting the scores correct is worth too many points...

James7
10-16-2015, 03:04 AM
I'm not sure either, but let's see in a real-world situation what happens.

It should be higher bonus scoring here in a best-of-3 tournament than in an actual slam where you are guessing game totals in a best-of-five. Though that also means that bonus scoring in WTA is going to be easier than in ATP at the Slam events. This is more of a proxy for how it would work in a WTA Slam event, so we'll have an indication of that.

With 2 of the quarters finished there are some +3s, and surprisingly the Random number generator got close too (despite generating numbers from 0 up to a best of 5 match) and got +4, currently beating others in the bonus category. I may just set Random brackets in general to 0s in bonus predictions as I plan to do with other 'flavor' entries.

ptmcmahon
10-16-2015, 03:49 AM
I just realized when picking total # of games...I was thinking of 5 set matches :) Pretty sure I'm not gonna get any bonus points there.

James7
10-16-2015, 05:46 AM
I think this is interesting.

There's plenty of action available yet at this stage. Even though I'm lucky enough to have a lead (due to my nearly sole belief in Rafa having turned things around), it's possible to be overtaken depending on how remaining matches go. Even though there's a clear advantage, it's not unassailable. Which I think is a positive thing.

A bunch of people got a full 5 bonus points on Djokovic's quarterfinal.

EDIT:
*Noticed and fixed a typo in the tournament winner pick points projection. That was incorrectly evaluating the finalist_2 slot again to see if it was 'alive' rather than the winner slot (too much copy/paste!). Now correctly projecting winner points available so max possible score has gone up for many.

James7
10-18-2015, 04:11 AM
And it looks like it's done!

Flat Scoring: http://www.talkabouttennis.com/contests/SlamTest/picks_res.php?tour=ATP&tournament=Shanghai
Progressive Scoring: http://www.talkabouttennis.com/contests/SlamTest/picks_res_alt.php?tour=ATP&tournament=Shanghai

A little awkward to accidentally win your own contest pilot, isn't it? :lipsrsealed:

But Dryrunguy and Kirkus also followed up with strong overall performances in second and third by the progressive scoring algorithm and uniform scoring both.

In terms of raw bracket prediction, via the progressive scoring James7, Kirkus, and liam_valid had the strongest performances, though the order varies by scoring method.

Leading the bonus scoring with exceptional scoreline predictions are fastbackss in first, Dryrunguy again in second, and Scotty in third.

Some minor differences overall in ordering depending on which scoring method you go with.

So how do you think this went? Now that it's complete I'd love to hear your feedback and suggestions!

Thanks again for participating!

JTContinental
10-18-2015, 09:53 AM
I liked everything except the way that my picks didn't seem to score any points. I hope this can be fixed in future events.

liam_valid
10-18-2015, 10:05 AM
Thanks for doing this, it made another bland Djokovic title a little more exciting!! Technically, I liked how all the picks and percentages I wanted to look at during the comp were all there in front of you when you clicked on the link instead of having to open different pages, and i loved how you can see the amount of picks everyone still had in play

lukeorion
10-18-2015, 11:31 AM
I didn't get the chance to be too involved with the scoring or anything due to work, but I loved it. Made things interesting.

Scotty
10-18-2015, 03:23 PM
Clearly, the shoreline bonus worked best. :cheesy:

James7
10-19-2015, 06:17 AM
So I think the general rumbling I had heard was that the progressive scoring was preferable. And it seemed to me that the point increase of one point per round worked well.

To summarize points at stake at a slam:
64 points for round of 64 predictions (1 each)
64 points for round of 32 predictions (2 each)
48 points for round of 16 predictions (3 each)
32 points for round of 8 predictions (4 each QF slot)
20 points for round of 4 predictions (5 each SF slot)
12 points for finalist slot predictions (6 each)
7 points for the tournament winner (7 points)

Bonus would be up to 5 for each predicted match (35 total), but as we saw, getting the full 5 is rare enough and this was an easiest case scenario as best of 3 (WTA only is slams). The best bonus prediction here was less than half of that, with the average for actual people who did them somewhere between 8 and 9 points in total.

Did the bonus work well enough? Too many points?

Is it okay to have the scoreline guesses be done so abstractly up-front? Even though it would be a logistical challenge, score guesses could potentially be done later into the tournament. I kind of like that it's tied some to who you *think* will be playing at that point by doing it early with submission, personally, but I'm willing to be swayed.

Other random prompts/musings I'll throw out there:
Non-player brackets: Should they be separated from actual players? Or left in the same results table?
Any other stats worth compiling or viewing? What I have is pretty basic, but I'm not sure what other people might like to see, like the possible points abstraction added during this trial run.

Thanks again for the feedback!

Drop-shot
10-19-2015, 06:31 AM
I'd keep the Bonus (guessing games) as is. Makes it more "interesting".

Besides, I think you'll be opening up to a lot of people who forget to return and post game numbers.

mysterX
10-20-2015, 04:39 PM
By non-player brackets, I assume you mean the random one and the Tennis Abstract. I say leave them in the results table. If you want to distinguish them, you could use a different font for those names.