PractiScore Math Question

[AlexMT]

Hi all, 

 

Can someone explain me how practiscore make math per image below (C: x0.18, D : x0.53, M: x2.64, NS: x1.76 MAX Time 26)

 

Stage setup 

2 steel

14 Paper, some noshoots

Min round 30

Max points 150

 

Feeling a little bit confused 

 

[broadside72]

I usually see that when there is an incomplete upload or only partial match completion

[euxx]

There is a menu button at the top right of the screen, tap it, then tap About option, then read a first paragraph. [emoji16]

[theWacoKid]

MaxTime is the time needed to match the high hit factor on the stage while shooting all A's. 

 

For the values associated with hits/penalties that is the amount of time improvement from MaxTime for each of those hits/penalties that would be needed in order to maintain a score equal to the current high hit factor.

[lstange]

26.38 seconds divided by 150 points is 0.1759 seconds per point.

 

You lose 1 point, or 0.1759 seconds, on a Charlie.

You lose 3 points, or 0.5276 seconds, on a Delta.

You lose 15 points, or 2.638 seconds, on a Mike.

You lose 10 points, or 1.759 seconds, on a NS.

[AlexMT]

On 11/8/2019 at 8:11 PM, lstange said:

26.38 seconds divided by 150 points is 0.1759 seconds per point.

 

You lose 1 point, or 0.1759 seconds, on a Charlie.

You lose 3 points, or 0.5276 seconds, on a Delta.

You lose 15 points, or 2.638 seconds, on a Mike.

You lose 10 points, or 1.759 seconds, on a NS.

Great one @Istange .... That means  for a good stage planning we need just to estimate time need  and after that we can have clear math of how much mistakes will cost us

 

Now question of all questions is how to make fast guestimation of time needed to finish the stage with high hit factor?

 

Option 1: Estimation of HF that will lead into time and time will lead into cost of penalties

Option 2: Estimate time that will lead to math .....

 

[euxx]

You can simply walk stage at your own pace with a stop watch. With a little practice you get a good estimate.

Also, towards the end of a multi-day match some top shooters might have shot that stage, so PractiScore Competitor simply show you their best time.

PS: classifier stages are different story. The HHFs are known, so no guessing part there.

[lstange]

On 11/11/2019 at 8:48 AM, AlexMT said:

we can have clear math of how much mistakes will cost us

Then the only remaining part is figuring out how much time not making the mistakes we did not make cost us. Not sure if PractiScore is working on it already.

 

It's tempting to say "that delta cost me half a second, surely I could have spent another quarter of a second to get an alfa instead". That conveniently ignores the targets on which you did not spend extra time and still got alphas.

[euxx]

6 hours ago, lstange said:

Then the only remaining part is figuring out how much time not making the mistakes we did not make cost us.

 

It took me some time to process this "how much time not making the mistake we did not make cost us".

 

The numbers above are showing time-representation for the shooter's mistakes. If you are able to avoid mistakes by taking more time to shoot more accurate - you will be improving your results.

 

It could be hard to shoot A instead of a C with cost of C at 0.1759 seconds

But with 0.5276 seconds penalty for a D, it should be possible to shoot 0.2..0.3 seconds slower to get an A.

And with 2.638 seconds for M you could have slowed down or do a make up shot in less than 2.6 seconds.

 

However those numbers are overall for the whole stage. That D or M could be on a far partial target or could be on a full target that is at point blank - both on the same stage. We simply don't have that information in the match results.

 

6 hours ago, lstange said:

 

Not sure if PractiScore is working on it already.

 

I wonder if you have checked the About section in the app...

 

6 hours ago, lstange said:

It's tempting to say "that delta cost me half a second, surely I could have spent another quarter of a second to get an alfa instead". That conveniently ignores the targets on which you did not spend extra time and still got alphas.

 

That information is actually available in PractiScore Competitor if you use AMG Lab timer.

 

[lstange]

On 11/11/2019 at 12:51 PM, euxx said:

classifier stages are different story. The HHFs are known, so no guessing part there.

True for the high hit factor, close up classifiers. But with distant targets it might make sense to accept more Charlies, and it's not immediately obvious how many.

[euxx]

6 minutes ago, lstange said:

True for the high hit factor, close up classifiers. But with distant targets it might make sense to accept more Charlies, and it's not immediately obvious how many.

 

I was referring to the fact that you know GMs time for classifiers. And with C's and D's you will have to shoot stage faster than that to stay in GM bracket.

 

But I disagree on taking Cs for a distant targets. The HHF likely will be lower for that stage, so accuracy is getting more important.

 

Now, you also need to keep in mind that time-costs for C,D,M are universal, but your estimates based on them - are not. Your estimate need to be based on your ability to shoot an As on that distant target. In other words - you need to know how fast you can shoot accurately on various distances. That will drive your estimates and decisions for a given stage. E.g. GMs split time shooting As on a partial 30y target is less than D-shooter's one.

[lstange]

7 minutes ago, euxx said:

you know GMs time for classifiers

To get from HHF to time you need points. You can make assumptions about how many points GMs shoot, but that may not be exact (and will also depend on division).

[euxx]

1 hour ago, lstange said:

To get from HHF to time you need points. You can make assumptions about how many points GMs shoot, but that may not be exact (and will also depend on division).

 

Both HHFs and max points are known for all classifiers. PractiScore Competitor shows them.

[lstange]

1 hour ago, euxx said:

max points are known

But that's not what all GMs shoot, and it's not always necessary to shoot all alphas for a GM-level run. On average, a 95%+ classifier has 90% of available points. And it differs by classifier.

[euxx]

1 hour ago, lstange said:

But that's not what all GMs shoot, and it's not always necessary to shoot all alphas for a GM-level run. On average, a 95%+ classifier has 90% of available points. And it differs by classifier.

 

So, you have %5 margin for a errors. That error margin can be also converted to time-cost.

PS: yes, you don't have to have get all As to get a GM HFs, but you can also get GM HFs with a slower time when shooting all As.

[euxx]

Btw, if I'm not mistaken, uspsa uses 95% for HHFs

[lstange]

1 hour ago, euxx said:

Btw, if I'm not mistaken, uspsa uses 95% for HHFs

USPSA HQ did not explain exactly how HHFs were defined, but they roughly correspond to 99.5th percentile of the empirical hit factor distribution. At least for production.

[euxx]

2 hours ago, lstange said:

USPSA HQ did not explain exactly how HHFs were defined, but they roughly correspond to 99.5th percentile of the empirical hit factor distribution. At least for production.

 

You can verify that by looking at their match results.

[MikeBurgess]

17 hours ago, euxx said:

Btw, if I'm not mistaken, uspsa uses 95% for HHFs

I believe the process they said hey were using (and I verified it on some of the new 18 series classifiers) was take the average of the top 10 scores on that classifier, for older classifiers they supposedly removed obvious outliers then averaged the top 10 but I believe some of that was conditional as L10, Limited and SS sometimes have the same HHF but I seriously doubt the data sets are the same. 

 

 

[euxx]

32 minutes ago, MikeBurgess said:

I believe the process they said hey were using (and I verified it on some of the new 18 series classifiers) was take the average of the top 10 scores on that classifier, for older classifiers they supposedly removed obvious outliers then averaged the top 10 but I believe some of that was conditional as L10, Limited and SS sometimes have the same HHF but I seriously doubt the data sets are the same. 

 

I'm just saying that times you see for not-yet-shot classifier stages in PractiScore Competitor app are based on 95% of classifier score and not 100%.

I'm getting base HHFs from uspsa (unofficially), so that is the same data as they are using and to me it doesn't matter how they got them.

[MikeBurgess]

9 minutes ago, euxx said:

 

I'm just saying that times you see for not-yet-shot classifier stages in PractiScore Competitor app are based on 95% of classifier score and not 100%.

I'm getting base HHFs from uspsa (unofficially), so that is the same data as they are using and to me it doesn't matter how they got them.

got it thanks

[lstange]

On 11/13/2019 at 4:37 PM, euxx said:

I disagree on taking Cs for a distant targets. The HHF likely will be lower for that stage, so accuracy is getting more important.

I looked at GM-level classifier runs in Production, and there is a statistically significant positive correlation between HHF and points shot divided by maximum available points. I excluded fixed time, as their HHFs are not directly comparable.

[AlexMT]

If I may add one PoW from Rob's side ..... "In a match I would never make up D on stage with HF 7-10 and C on stage with HF 3-5" assuming Major scoring. If we transpose that to minor ......

[euxx]

16 hours ago, lstange said:

I looked at GM-level classifier runs in Production, and there is a statistically significant positive correlation between HHF and points shot divided by maximum available points. I excluded fixed time, as their HHFs are not directly comparable.

 

Is that for the same classifier? I still don't see how that disprove my point. Each shooter has his own style and cadence. So, some can shoot faster to compensate for lost points and with 5% margin window for GM scores they can do that. I'm just saying they can also shoot slower but more accurate to make GMs HHF.

[lstange]

3 hours ago, euxx said:

Is that for the same classifier?

No, all Virginia Count and Comstock classifiers combined. Each dot is one run. X axis is high hit factor, so results for the same classifier are lined up vertically.