32 replies to this topic

[Y E O N N E]

Hit the Deck, on 28 September 2016 - 04:41 PM, said:

Just a note, be careful when interpreting those SD/standard deviations. Two equal SD values of two different averages doesn't mean that they have the same spread.

Right. It's just a point value indicating that there is inconsistency. Magnitude merely suggests severity of inconsistency, but there is little relation between the standard deviations of two unrelated sets of data (unrelated in this case being 'Mech A versus 'Mech B ).

It's why I was questioning that normalized standard deviation value. It seems rather meaningless to me, drawing a correlation where none should be drawn.

Edited by Yeonne Greene, 28 September 2016 - 04:45 PM.

[Kubernetes]

This just confirms my belief in the Marauder over the Black Knight. Dem Whammys OP tho.

[Tarogato]

Hit the Deck, on 28 September 2016 - 04:41 PM, said:

Just a note, be careful when interpreting those SD/standard deviations. Two equal SD values of two different averages doesn't mean that they have the same spread.

Yeonne Greene, on 28 September 2016 - 04:44 PM, said:

Right. It's just a point value indicating that there is inconsistency. Magnitude merely suggests severity of inconsistency, but there is little relation between the standard deviations of two unrelated sets of data (unrelated in this case being 'Mech A versus 'Mech B ).

It's why I was questioning that normalized standard deviation value. It seems rather meaningless to me, drawing a correlation where none should be drawn.

If you visit the googledoc, on the All_charts, I have replaced the standard deviations for the normalised stats with the CoVs from the actual scores. This should be a lot more useful. I haven't done it for the tonnage brackets at the bottom or anything, because that's a pain in the arse to do.

[Y E O N N E]

Tarogato, on 28 September 2016 - 05:16 PM, said:

If you visit the googledoc, on the All_charts, I have replaced the standard deviations for the normalised stats with the CoVs from the actual scores. This should be a lot more useful. I haven't done it for the tonnage brackets at the bottom or anything, because that's a pain in the arse to do.

I did see that, actually. Thanks for that!

Kubernetes, on 28 September 2016 - 04:50 PM, said:

This just confirms my belief in the Marauder over the Black Knight. Dem Whammys OP tho.

I've been saying it since forever: MAD-5D > Black Knight. It's really a fantastic 'Mech, very well-tuned and rewards skill without relying on cheese of any sort.

Edited by Yeonne Greene, 28 September 2016 - 05:35 PM.

[Ted Wayz]

Thanks. No real surprises however.

[El Bandito]

Jman5, on 28 September 2016 - 10:48 AM, said:

I can't help but wonder how much of these scores are due to straight imbalance and how much is from the number of people competing in each leaderboard.

For example, the Orion was an uncompetitive leaderboard. I was watching it in the first day and it took a long time until it even filled up to 75 players. The end result was a predictably low scoreboard.

The archer and catapult are pretty similar in design and playstyle. Yet the catapult leaderboard beat the pants off the archer leaderboard. If this was pre-resizing you could make a quirk argument but the catapult got its hitpoint quirks nerfed hard. In fact the Archer blows it out of the water in this regard.

Is the archer a truly bad missile-mech or is it simply an underplayed one?

Being underplayed is a reliable sign that is considered as a bad mech. In fact, public's poor opinion of it after it was released was the whole reason why I was obsessed with my Archers and have such a good WLR with the 5W. I wanted to prove to people that it can work, though Archer is very limited in function. Its super low positioned cockpit combined with its easy to hit torsi hitboxes basically screws the chassis over the hardest.

Which is why I wanted to see if people would support high-mounted Archer cockpit to give new lease on the little used 70 tonner, in this thread. http://mwomercs.com/...archer-cockpit/
Amazing how some people want the Archer to stay screwed, due to lore reasons.

Edited by El Bandito, 28 September 2016 - 07:19 PM.

[Y E O N N E]

The ARC-5W is the only one that is interesting to me. Trolling around with 9x SRM2 appeals to my tastes.

Unfortunately, it's about go get upstaged by the Huntsman.

[Rock Roller]

Yeonne,

Any chance you can change the title to "Proof that the Victor needs buffed"?

Honestly, great wotk and thank you.

[Y E O N N E]

Rock Roller, on 28 September 2016 - 07:07 PM, said:

Yeonne,

Any chance you can change the title to "Proof that the Victor needs buffed"?

Honestly, great wotk and thank you.

Wat, this isn't my thread and that's not my work. Tarogato is the one you should be thanking.

[Rock Roller]

Damn Vodka!!!

[627]

Just want to throw in that these events had very different weight class distribution on the battlefield. You saw way more medium mechs and few others in that event while the assault event was basically "spot the non-assault" with 8+ Assaults per side.

With so much more potential damage on the field, it was easier to get higher scores in the assault event.

This doesn't screw the whole data and comparing mechs within a weight class is still a good indicator but comparing different weight classes is not without issues.
If you look at it this way, it is even more interesting where the cheetah, crow and hunchback IIC landed. They got high scores but without the "potential damage" buff from the assault event.

[Tahawus]

http://mwomercs.com/...ost__p__5414473

My alternate approach to analyzing the data. It generally supports Taragato's results, vut uses a more quantitative approach.

Edited by Tahawus, 29 September 2016 - 06:03 AM.

[KodiakGW]

My poor Dragons. The only heavy in the bottom 10. *sniff*

Also glad I refunded the Victor Mastery Pack I bought on sale.