Hot match alert?

Like most of you, I am bummed that I can be in St. Louis over the next 3 days.

But, since I didn’t leave town, I also have 10,000 other things I need to get done between now and Einstein.

Does anyone know of an alert system that can send a text or e-mail or something (tweet?) with details on hot matches starting in 5 minutes?

Ideally it would give whos in the match and why it’s important (e.g.Poofs, Box of Rocks & Orange Crate vs Robonauts, Mechanical Mayhem, & HOT. Poofs planning to break out their 9 ball Auto if they can convince B.O.R and O.C. to give up their bounders :wink:

Dr. Joe J.

I believe you can use myTBA to get notifications in the Android app when teams that you follow are playing but as far as I know there isn’t any other way to get match notifications.

I could send Phil a text and ask if he could add it, though :stuck_out_tongue:

You can get individual match alerts on the TBA Android app. Just click on the match, click on the star, and check the notification. But we don’t have the matches in our database yet, so you’ll have to wait until FIRST makes them official.

TBA can already send “upcoming match notifications” which go out about 5-7 minutes before the target match is about to start.

I don’t see it being too hard to add a button to the TBA Site to “subscribe me to hot matches” that adds in all the relevant Subscriptions in for you.

What (mathematical) criteria would you suggest for determining a “hot” match? Or does anybody have a hand-curated list?

EDIT: I need a “bat symbol” that calls me to threads where people have questions for me :rolleyes:

I’m guessing most people don’t want to watch a heavy on defense match. Here is my attempt at an exciting to watch match:

When Average OPR for both Alliances is in the top 1% of all matches. I probably want to watch it.

I might also want to emphasize autonomous high goal scores.

There’s 1000 matches at Champs. You’re suggesting that only 10 of them would qualify as “hot”. I’m not disputing the metric, but your calibration might be a bit off.

If predicted match score is greater than X
If predicted auton score is greater than X
If OPR for one or more teams is greater than X% of teams/teams at champs

Those will allow for more matches to get a “hot” rating.

Joe is a busy guy, with 10,000 things to do, cutting out a minimum of 22 minutes and 30 seconds out of his day might be hard!

But you bring up a good point:
Top 1% Hottt
Top 2% Hott
Top 3% Hot

Without taking the time to really think about implementation (or calibration), this is roughly what I would do for “hot matches”:


if(
BlueAllianceOPRinTopXXpercent || BlueAllianceOPRinTopXXpercent (I'm not sure if I'd do an "or" or an "and" here)
OR 
two red alliance in top 20 vs two blue alliance in top 20
OR
(>=1 red alliance in top 10 vs >=1 blue alliance in top 10) && (abs(RedAllianceOPR - BlueAllianceOPR) < 10)

I’ve never looked at the blue alliance API, but I would think all of that information should be available. I’d also think that for the first 3 or so matches I’d use the existing OPR available rather than the OPR from champs. Intent would be to be looking for matches that should either be close (and at least somewhat important), impact standing, or should be high scoring.

It’s a whole match rating system!

1% - Ghost Pepper
2% - Habanero
3% - Scotch Bonnet
5% - Cayenne
10% - Serrano
25% - Jalapeno
50% - Bell Pepper

Any match rating system based on OPR is going to miss hot matches that are relevant because of ranking implications, though.

Challenge accepted… for 2017.

Which matches?

That is a good question.

I am thinking this could be a kind of a mix match criteria for rules folks that can be customized for a particular person.

Maybe you have a rules with weights. The rule gives a match a number then you multiply by the weight and bob’s your uncle, you add them all up and the top numbers are in your queue.

So, I’m an old timer so I can put in some teams with low numbers I haven’t forgotten yet in my dotage. The more of them that are in a match the better.

Maybe I like to have games that matter. So I put in some sort of a rule about teams in the hunt for top seeds. The more teams that have skin the game in the rank hunt (e.g. if this team wins, they have a lock on #1 seed), the higher the rule triggers. Maybe the later the round the more important the match gets. …

Maybe you have a way to add other people’s rules to yours with new weights.

Maybe you can just punt and say if it is on Karthik’s queue, then I want it too.

Maybe you have ways of crowd sourcing matches where if enough people in the alert system are watching it, you should too.

I just want the tool. If you build it, Boy Howdy will I use it.

Dr. Joe J.

M

If we can devise a formula that uses a logarithmic scale we totally have a case for using the Scoville heat scale, and then we can use a lookup function to correlate the score to the appropriate pepper. Bringing in Carolina Reaper, Naga Viper and all the peppers.

Standings Impact Scores sounds like an interesting challenge. I see that corollary to the Match Hottness. Match Hottness wants to prioritize the offense a audiencability of a match, and really determine, if i’m going to watch one match, which one should it be? Standing Impact Scores should only measure the impact to the rest of the field.

A simple method to rank the Standing Impact Scores

Standard Deviation of All Teams Rank After - Rank Before)]

If you did that historically of all the results to date, I’m thinking you can probably build out a bell curve and then you’ll have a way to rank them and be able to gauge whats really important.

If you measure that in real time (replacing rank after with expected rank after), you’ll have a score and be able to determine what is decisive as the opportunity presents itself.

From what I heard from Phil, the TBA Android app will use this list for matches to watch:

http://frc.divisions.co/matches-to-watch

Not quite. Plus, that match list isn’t final yet (schedules are still preliminary).

We’re still working out a way to figure out which matches are ones to watch (if we want to calculate on TBA using prediction data or just “seed” it with an external list).

Pretty sure you just need to watch Newton…
Or you could post your cell# here and CD will take care of it :wink:

I’d be interested in some sort of digest. Best matches of the season for the impartial viewer (because any other match with 868 in it will be the best for me :smiley: )

CD IS the alert system. :stuck_out_tongue:

Fun feature announcement time! Thanks to Eugene for doing the implementation.

Check out this page: https://www.thebluealliance.com/account/mytba#my-matches

You’ll see a new button to Subscribe to hot matches! Click it, select a division, and then check the matches in the list that look interesting.

You will then receive an upcoming match notification to your Android phone when the matches are about to start.

Each match has a “hotness” factor that is designed to favor close, high scoring matches. We took some inspiration from the 2010 ranking algorithm and use a factor (winner score + 2*loser score), which comes from the match score predictions, to sort the matches. This gets scaled 0 to 100 relative to the rest of the division (100 being the highest factor and 0 being the lowest) to give the “hotness indicator”.

Here’s the code, for those who want to see.

That was fast! :smiley:

Is there no end of the awesomeness of The Blue Alliance?

Thanks for this!