Jump to content


Photo

Classifier HHFs about to get real


  • Please log in to reply
287 replies to this topic

#41 Mike Foley

Mike Foley

    Advanced Member

  • Members
  • PipPipPip
  • 907 posts

Posted 06 July 2017 - 07:48 PM

I laugh every time somebody says "this classifier is all shot out" because they believe the bit about resetting the HHFs that HQ has been selling since the inception of the classification system, yet nearly never doing.

Mike-- I suggest graphing the scores per-classifier and see what the distribution looks like before arbitrarily taking some number as the 100%. There's enough data the distribution will be clear rapidly and I've seen enough shenanigans over the years where the timer misses a shot, or the classifier was setup wrong or the scorekeeper -2-ed a friend and so on that unless you know the people and clubs involved, look like legit results. Find an unemployed Statistics major (there must be tons), hand them Excel and the data file and tell them to have at it.

I totally understand what you are saying, BUT, I don't think distribution matters since every score isn't a measure of true ability, mostly at the bottom 2/3.

#42 Motosapiens

Motosapiens

    Advanced Member

  • Members
  • PipPipPip
  • 2,521 posts

Posted 06 July 2017 - 07:50 PM

I'm going to give two examples. When I first started this job a year and a half ago, we reset a particular classifier that had never had anyone ever shoot an open score above 84%, and the same classifier in production had pages and pages of 100s all the way down to
C class. 

that's because production is full of ballers, and open is full of old fat guys who can't see.

 

but i did notice that the production HHF's are way easier than the SS HF's. I personally shoot almost exactly the same with both guns, but that is what we call an anecdote, and not actual data.


2013 National Champion C class  Limited 10


#43 Motosapiens

Motosapiens

    Advanced Member

  • Members
  • PipPipPip
  • 2,521 posts

Posted 06 July 2017 - 07:54 PM

I think that becomes far less common at the upper tiers of shooting. Are 100% GMs really trying to push their abilities and throw stages like that?

 

This is a very valid point. If you are not trying your best to get the next class so you can brag about it (like I just did), you may be more likely to shoot for match score and not worry about it.

 

I think the 5% below rule also really encourages the hero-zero approach. In the moto utopia, i would maybe take best 6 of 10-12 and forget about the 5% rule. If you want to hero/zero, fine, but if you can't hero more than 50% of your runs, your classification will continue to reflect your actual ability.


  • racetaco likes this

2013 National Champion C class  Limited 10


#44 Motosapiens

Motosapiens

    Advanced Member

  • Members
  • PipPipPip
  • 2,521 posts

Posted 06 July 2017 - 07:56 PM

Good question. I'm going to review every classifier in every division. I'll know more when I get into the numbers.

 

As a numbers guy (degree in applied math and statistics), I suspect you may want to look deeper than the first 10 for the average. OTOH, I don't really GAF. I just want to get better and beat guys that used to beat me (in matches, you homos, not in gay s/m).


  • Stubb, GuanoLoco and Will like this

2013 National Champion C class  Limited 10


#45 GuanoLoco

GuanoLoco

    Minister of Culture

  • Members
  • PipPipPip
  • 4,245 posts
  • LocationMy Safe Space

Posted 06 July 2017 - 10:23 PM

Really, get a mathematician/stats guy to look at this stuff.

You will never know what you don't know unless someone 'skilled in the art' looks at it.
  • Doc, chachacha, busdriver and 1 other like this
Are you now, or have you ever been a member of the Doodie Project?

You'll never find a more wretched hive of scum and villainy.

In Doodie, Veritas.

"You might be a little thin-skinned for this forum". -Scott.

#46 Timmy

Timmy

    Advanced Member

  • Members
  • PipPipPip
  • 38 posts

Posted 06 July 2017 - 10:33 PM

All changes to the system have been and will continue to be going forward.

 

Lame.  If someone never would have made their current class based off the new HHFs then bump them down B)



#47 aandabooks

aandabooks

    Advanced Member

  • Members
  • PipPipPip
  • 47 posts

Posted 06 July 2017 - 10:50 PM

Wonder where PCC is going to end up on this? Some of the classifiers are significantly easier in PCC than Open.

#48 busdriver

busdriver

    Advanced Member

  • Members
  • PipPipPip
  • 945 posts

Posted 06 July 2017 - 11:08 PM

I think moto just volunteered to lead the classifier HHF stats team.
  • GuanoLoco, Motosapiens and shmella like this

#49 foofighter

foofighter

    Advanced Member

  • Members
  • PipPipPip
  • 1,388 posts

Posted 06 July 2017 - 11:32 PM

I totally understand what you are saying, BUT, I don't think distribution matters since every score isn't a measure of true ability, mostly at the bottom 2/3.

 

1. How many classifier scores are really a measure of true ability though?

2. Yes, putting a data scientist to crunch numbers and coming up with a good statistical model that can be applied automatically as often as you like seems like a waste of time when you can do it manually over and over again every other year when shit becomes stale.


  • GuanoLoco and peterthefish like this

#50 foofighter

foofighter

    Advanced Member

  • Members
  • PipPipPip
  • 1,388 posts

Posted 06 July 2017 - 11:34 PM

Wonder where PCC is going to end up on this? Some of the classifiers are significantly easier in PCC than Open.

 

Gee, thats a tough one!

 

Do you think

a) they'll all get easier

b) they'll all get harder

c) some will be adjusted up and some down, like what Mike said would happen?

or

d) I'm from Area 6.


  • Just Some Random Hoser likes this

#51 Alfred Salveti

Alfred Salveti

    Advanced Member

  • Members
  • PipPipPip
  • 511 posts
  • LocationSugar City, USA

Posted 07 July 2017 - 12:45 AM

Can I retroactively get my GM card please.


I like to rewatch election night 2016 coverage and I want to marry a conservative black girl who likes to shoot.

#52 Will

Will

    Advanced Member

  • Members
  • PipPipPip
  • 1,430 posts

Posted 07 July 2017 - 02:54 AM

Honestly I think it's funny as shit when unclassified folks beat A or M shooters in the overall. Your placing in matches is a better barometer of your ability than a classification.
  • StrongHandOnly, Just Some Random Hoser and racetaco like this

#53 Mike Foley

Mike Foley

    Advanced Member

  • Members
  • PipPipPip
  • 907 posts

Posted 07 July 2017 - 05:38 AM

Really, get a mathematician/stats guy to look at this stuff.

You will never know what you don't know unless someone 'skilled in the art' looks at it.

.
The current task is to do what has been done, not reinvent the system. If we were reinventing the system, that'd be the course. What I included here was that we are updating again, like has been done in years past.

#54 Mike Foley

Mike Foley

    Advanced Member

  • Members
  • PipPipPip
  • 907 posts

Posted 07 July 2017 - 05:39 AM

Wonder where PCC is going to end up on this? Some of the classifiers are significantly easier in PCC than Open.


And some are harder. I can't seem to make GM, but then I don't pick classifiers.

#55 Mike Foley

Mike Foley

    Advanced Member

  • Members
  • PipPipPip
  • 907 posts

Posted 07 July 2017 - 05:42 AM

1. How many classifier scores are really a measure of true ability though?
2. Yes, putting a data scientist to crunch numbers and coming up with a good statistical model that can be applied automatically as often as you like seems like a waste of time when you can do it manually over and over again every other year when shit becomes stale.


I'm playing the hand dealt on this one. This time, I want to see it. The ability for auto updates will be there at some point. Part of this manual review is already automated.

#56 Mike Foley

Mike Foley

    Advanced Member

  • Members
  • PipPipPip
  • 907 posts

Posted 07 July 2017 - 05:44 AM

Honestly I think it's funny as shit when unclassified folks beat A or M shooters in the overall. Your placing in matches is a better barometer of your ability than a classification.


Yes, but we have both, and not all matches are significant. There is more talent at some locals than some section matches and even Area matches in some divisions.
  • Will likes this

#57 racetaco

racetaco

    Advanced Member

  • Members
  • PipPipPip
  • 809 posts
  • LocationMontana

Posted 07 July 2017 - 05:47 AM

I like the steel challenge classifications, hero or zero doesn't work. Being able to figure out how you did on individual stages and for the match overall is easy and accurate. Say you get %80 of the top guy at a USPSA match, makes me wonder did he have a bad day or am I truly able to perform at that level.

#58 Mike Foley

Mike Foley

    Advanced Member

  • Members
  • PipPipPip
  • 907 posts

Posted 07 July 2017 - 05:49 AM

Thanks for all the input. The announcement was supposed to help explain matches as classifiers and match bumps, what happens initially, and I threw in that we are updating HHF's soon with the traditional process.
  • Nimrod and StrongHandOnly like this

#59 Trigger Warning

Trigger Warning

    That's What She Said

  • Members
  • PipPipPip
  • 1,503 posts
  • LocationRaleighwood

Posted 07 July 2017 - 06:06 AM

I'm OUTRAGED that Foley is executing long standing procedures and updating members while personally reviewing the process.

 

The comments are interesting, and gaming up classes is certainly fucking dumb.  The GM's around here are real, it seems, as they tend to do excellent on a regular basis on classifiers (and matches).  Also, more or less, I usually do at a match about how my class rank would suggest.  Certainly we don't have GM's halfway down the list and a C (not even a PCC C) at the top.  Having watched Chris Tilley and Todd Jarrett shoot classifiers in the last week, it is clear they are not hero or zero classifier shooters, and neither are the other GM's floating around here from what I see.  

 

I've been convinced to not care about chasing classifiers for purposes of upping classes and to concentrate on overall match skills (which oddly does help in classifiers), but it's an interesting system that seems to motivate a lot of people to work on core skills in a measurable way.  Updating it is good, and finding ways to make it more accurately reflect a shooter's skill comparatively is good too.  I guess I'm not outraged.  Fuck.


  • Stubb, Doc, ToddKS and 2 others like this

On injured reserve.


#60 Doc

Doc

    Google ( o Y o ) SME

  • Moderators
  • 4,264 posts
  • Location: safety wired in the pissed off position

Posted 07 July 2017 - 07:02 AM

Yo, Beard.

If you are going to publish the new HHFs, why not also publish the old HHFs so we can see the changes?

That, and what Trigge Warning said.
  • MilkMyDuds likes this
Unload and Show Clear...
tsarbombaexplos2.jpg
"Licorne" French Polynesia 1970

It pins my irony meter when people post things like "your to stupid..."

You brought a rifle to a handgun competition?




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users