Analysis of Stats Using 1.21

Matrix Games Forums

Forums Register Login Photo Gallery Member List Search Calendars FAQ

My Profile Inbox Address Book My Subscription My Forums Log Out

Analysis of Stats Using 1.21

View related threads: (in this forum | in all forums)

Logged in as: Guest

Users viewing this topic: none

Printable Version

All Forums >> [Current Games From Matrix.] >> [Sports] >> PureSim Baseball >> Analysis of Stats Using 1.21

Page: [1]

Message

<< Older Topic Newer Topic >>

Analysis of Stats Using 1.21 - 1/27/2006 6:00:18 PM

BryanK

Posts: 140
Joined: 8/21/2005
Status: offline

I can put together slides if there's interest, but I'm noticing the same general trends in v1.21 as I did in all my earlier work. The most glaring problem is that homerun production begins to ramp up suddenly around the 10th year of the league in fictional associations. Combined with prior testing, I think this effect occurs regardless of the year the association begins play.

I haven't tested an association using real players.

Shaun, do you need slides or some other quantification?

Post #: 1

RE: Analysis of Stats Using 1.21 - 1/27/2006 6:44:24 PM

KG Erwin

Posts: 8981
Joined: 7/25/2000
From: Cross Lanes WV USA
Status: offline

Bryan, is this a sudden jump in HRs, or a gradual increase? Is it connected to an increase in power ratings in rookies or developing players?

It would be interesting to see if this occurs when using real players over a 10-year span.

(in reply to BryanK)

Post #: 2

RE: Analysis of Stats Using 1.21 - 1/29/2006 1:07:36 AM

BryanK

Posts: 140
Joined: 8/21/2005
Status: offline

Here's a quick slide on the HR numbers. Everything looks to be inline for the real players. (Great job Shaun!) The association using modern players tracks perfectly. The one starting in 1900 is high in the early years but comes back down. My hypothesis is that the mix of real and fictional players in the associaiton's early days drives the variance. It just seems a natural fit since the fictional associaiton's numbers, which are quite high, overlap with the 1900 real-player association through that association's early years.

Based on my prior work with 1.20 and earlier, I believe that fictional associations start off well, but that something with new player generation causes the stats to deviate as time passes. I have other slide shows on the board somewhere which should speak to this in greater detail.

Not to be a bit of a downer... but if the fictional assoctions produce crazy power numbers, the game looses much of its appeal...

(in reply to KG Erwin)

Post #: 3

RE: Analysis of Stats Using 1.21 - 1/29/2006 1:28:04 AM

puresimmer

Posts: 2299
Joined: 7/24/2005
Status: offline

Interesting. I'm actually really happy with the real player's engine at this point, and to be honest even the fictional players is not as much of a downer as I think it looks on that graph.

I'll look more at the fictional stuff. At its worst in your chart the difference is about .35 HR per game, so I'm sure with a few tweaks I can improve it even more. Its a delicate balance to have one engine produce realistic stats by time period for both real and fictional players along with aging and retirement algorithms.

I'll do even more work on PS 2006 to focus on improving this! Thanks so much for the really well thought out analysis.

_____________________________

Developer, PureSim Baseball

(in reply to BryanK)

Post #: 4

RE: Analysis of Stats Using 1.21 - 1/29/2006 1:32:55 AM

BryanK

Posts: 140
Joined: 8/21/2005
Status: offline

Thanks Shaun :)
This is the kind of stuff I do all day long at work with financial figures and such... applying the same thiking to Puresim seemed only natural ;)

Do my thoughts about initial players in a fictional assocations vs. the ones generated later on make any sense?

< Message edited by BryanK -- 1/29/2006 4:10:00 AM >

(in reply to puresimmer)

Post #: 5

RE: Analysis of Stats Using 1.21 - 1/29/2006 7:13:54 PM

BryanK

Posts: 140
Joined: 8/21/2005
Status: offline

I let the real player association run overnight to see how the stats would be affected once the game ran out of real players to import. My suspicions were confirmed, as HR production consistently increased each year. Levels in 2015 were more than 50% higher than they were in 2004.

Real players are dead-on accurate... it's the fictional ones that get screwy...

(in reply to BryanK)

Post #: 6

RE: Analysis of Stats Using 1.21 - 1/30/2006 10:39:05 PM

Amaroq

Posts: 1100
Joined: 8/3/2005
From: San Diego, California
Status: offline

quote:

ORIGINAL: BryanK
Do my thoughts about initial players in a fictional assocations vs. the ones generated later on make any sense?

Absolutely - and I appreciate you doing the analysis.

I'm not convinced that the 1900's association is quite ideally shown there - the last five years look about right, but that feels like too short a sample to be statistically significant with accuracy.

I'd prefer to see you start in 1901 (the first year with 16 teams), and go on through 40-50 years, to see what that looks like.

(in reply to BryanK)

Post #: 7

RE: Analysis of Stats Using 1.21 - 1/31/2006 5:36:55 AM

BryanK

Posts: 140
Joined: 8/21/2005
Status: offline

I can donate a few more clockcycles to the cause ;)
I'll set something up before I go to bed and let it run through when I get back from work tomorrow. For the record, everything I've done thus far has been with a 16 team league.

(in reply to Amaroq)

Post #: 8

RE: Analysis of Stats Using 1.21 - 2/6/2006 6:31:53 AM

BleacherBum

Posts: 56
Joined: 8/21/2005
Status: offline

I haven't run a career sim since last Summer, but I recall seeing leagues evolve in this manner. And after looking into it, I determined that one of the biggest contributing factors was the CPU GM's inability to recognize and develop pitching talent. In the first 5 or so years of a new fictional league, the pitchers are acquired in the initial draft and are used appropriately. But as these guys get old and retire, the next wave of pitchers just doesn't meet major league standards. The CPU would commit various offenses, including overusing very young pitchers in the majors, keeping aging vets around too long, and not properly developing the best prospects. If you notice that your pitching staff shines above the rest of the league, you should investigate this possibility further. If you are running a league of all CPU run teams, then you should also be suspicious of this management flaw.

(in reply to BryanK)

Post #: 9