• Hello Guest. We are upgrading the server's XenForo versions. This has to be done in stages. 1.5.18 ---> 1.5.23 ---> 2.0.0 ---> 2.0.12 ---> 2.1.0. I will likely upgrade each one and leave it for a day or so to see if there's any kinks. When moving from 1.x to 2.x, our add-ons will cease to work. Much of the functionality of said add-ons is now native within XenForo 2.x, so I don't think we'll miss much if anything. IF we are, we can try out new 2.x XF add-ons.

OOTP Statistical Analysis/Nerding Thread

NML

Well-Known Member
Yep, adjusted mine down since mine is for runs and urs is for WAR, which is almost exactly a 10:1 ratio

[xtable]
{tbody}
{tr}
{td}VARIABLE{/td}
{td} NML {/td}
{td}TRAVIS {/td}
{/tr}
{tr}
{td}CON{/td}
{td}.118{/td}
{td}.112{/td}
{/tr}
{tr}
{td}GAP{/td}
{td}.010{/td}
{td}.023{/td}
{/tr}
{tr}
{td}POW{/td}
{td}.056{/td}
{td}.055{/td}
{/tr}
{tr}
{td}EYE{/td}
{td}.044{/td}
{td}.035{/td}
{/tr}
{tr}
{td}K'S{/td}
{td}.028{/td}
{td}.033{/td}
{/tr}
{/tbody}
[/xtable]

Pretty close across the board, although I do have an extra variable that you didn't when it came to Batting WAR.

---

Correct me if I'm wrong, but OOTP 20/80 ratings are compared to ALL players, not just WBL, correct?
 

Travis7401

Douglass Tagg
Community Liaison
Yeah, those are statistically identical, any difference is just going to be based on sample size and noise from stadium/park factors.
 

OU11

Pleighboi
Utopia Moderator
The one I was always interested in was contact but specifically how young players with different potentials in Ks/Eye affected their final contact ratings. I didn't get adderall until I didn't care anymore but I (and @Orlando too) had a hypothesis that if their Ks potential was high and their contact potential was average they would end up with above average contact because whatever the hidden rating is was often the highest mover during the development phase.
 

NML

Well-Known Member
Would we expect all leagues on 18 to act the same way? Obviously each league (or even years within a league) will have minor changes based on what talent level is available, but I'm curious as to whether I could run the same sort of test within another league, simply for the data.

The issue with testing almost any of that type of hypothesis is a lack of data I have on our league. Even with just the last ten years of stats and scouting across all levels, that would be fairly easily to test.
 

NML

Well-Known Member
That's what I figured.

A quick guess I have based on my data is that we have a downswing in batting talent - based on what I saw with the wRAA compared to the data set (which I'm not sure where it's from, late 50's is my guess).

Every team had a positive WAR for pitching, ranging from 3 to 22. But several teams had negative for batters and I think the highest was Dam with 18 even though they have multiple MVP-level players.
 

Orlando

Well-Known Member
Utopia Moderator
Yeah I’m pretty sure we have dynamic talent turned on so things will swing from pitcher heavy to batter heavy. They shouldn’t change the weight of 80/20 ratings, but it does change overall ratings (starzzz). A guy with 80 power has 80 power regardless of league and era. That said, those ratings aren’t exact and vary from scout to scout.
 

NML

Well-Known Member
But wouldn’t 80 power vary based on the other players? i.e. the top 2.2% (or 80 rating) of power hitters may hit 50 homers during one year but may hit 65 the next decade

In addition, if you had an 80 power hitter vs 50 stuff pitchers compared to 70 stuff, that would alter things as well

Edit: or is 10 a standard deviation? For some reason I thought it was 15
 

doh

THANK YOU Dermott McHeshi
This is based off minimal research-- but I believe we are in a swing with better defensive players vs. offensive players.

Team DE was horrendous for years and in the 50s from just quick looking league BABIP was usually .305-.308 range. Now it's in the .295 range. Also I think owners are picking more defensive players.
 

Orlando

Well-Known Member
Utopia Moderator
But wouldn’t 80 power vary based on the other players? i.e. the top 2.2% (or 80 rating) of power hitters may hit 50 homers during one year but may hit 65 the next decade

In addition, if you had an 80 power hitter vs 50 stuff pitchers compared to 70 stuff, that would alter things as well

Edit: or is 10 a standard deviation? For some reason I thought it was 15
The results will vary, but their rating does not.
 

NML

Well-Known Member
On the back of 10 years of sim stats with help from Lando and OU, I did more research, this time focused on wRC+ instead of wRAA. The reason for this was I wasn’t getting park factors in the equation and I was worried with some of the extreme parks that we have. In addition, it normalizes everything compared to the rest of the league, so it should eliminate most of the ebbs and flows of the talent pool we have.

On the batters side, this bumped my r2 from .584 to .69 (giggitty). The coefficients were mostly the same, though Gap got a nice bumped compared to both mine and Travis’s ratings. The rankings of importance were mostly the same, with contact first and foremost. Power continues to be a decent second, roughly half the importance of contact. Then eye, then gap (which was previously bottom), then previous wRC+, and finally avoid K’s.

This is pretty intuitive - the ability to make contact drives everything else, as does how hard you can hit it (just to a lesser extent). Then, it’s are we able to take pitches to get walks and work the count to get a better pitch to hit, as well as our ability to turn those singles into doubles and triples. Previous wRC+ can gives us an indication if someone is a perennial over/underachiever. Avoid K’s can lead to some productive outs or forcing errors, but is certainly the least valuable skill.

As some sort of reference, if a player gets a 5 point bump in Avoid K, you can expect them to be 1% better. A ten point bump in contact would make them an impressive 15% better - which works out to be around an extra ten runs (or 1 WAR).

Continuing this to pitchers, unfortunately I was not able to get as high of a r2. I’m stuck at .486. I have a feeling that if xFIP was more readily available, I could accomplish this. So I’m sure that I can calculate this and get more accurate results.

Still, nearly .5 isn’t terrible, and we can still get fairly accurate predictions with this.

Most interesting is that movement was more predictive than stuff.
 

Travis7401

Douglass Tagg
Community Liaison
I always found movement to be most important. I think stuff is important as well, but there is a lot more "noise" in that stat because the # of good pitches matters. For instance, it is harder to control for people setting pitchers as RPs and getting higher stuff ratings, etc.
 

NML

Well-Known Member
So I've been doing some research into prediction intervals - for this purpose, mostly, with how much certainty can we predict that a team will make the playoffs? I'm still a step away from that; I can get a prediction interval on wins for a team, but it's hard to translate that to playoffs specifically. Once I have full team predictions, that will be a little easier.

A note about this: it doesn't include injuries. I haven't found a way to account for those yet, which is skewing the numbers since this is strictly based off of predicted production vs actual production. If we assume that injuries evenly affect each team, each year (which is completely false), then this would hold true. But, I don't have the research to predict injuries, nor do I want to go into the process of finding back-ups for each team. So let's go with that faulty assumption and move on. Anyways, here's what I've found so far:

We can get within 10 wins of a team's record about 94% of the time. Another way to put that, a team will finish with their predicted wins, minus 10, 97% of the time. In other words, if a team is predicted to finish 20 wins above another team, that is a more or less a lock to happen. So if we set the barrier of making the playoffs at 82 wins, a team with a predicted record of 92-60 will make the playoffs almost always. Because of our high percentage of teams making the playoffs relative to the number of teams, this makes sense.

How about those long shots? It works the other way. If a team is predicted to win 72 games, they will finish above .500 only 3% of the time. If a team is predicted to win 65 games, they have a .0038% chance of finishing with 82 wins or more. For those interested, the standard deviation of a season is about 6.36 wins - so roughly 68% of the time, a team will finish within six wins of their prediction.

The unfortunate limitation of this is 1) movement of players, such as adding a free agent, trading, promoting, etc and 2) injuries. That means that the actual variance is significantly higher. However, I think this does have an interesting point - if ur team is predicted to be a bottom-half team, like around 70 wins, ur chances of making the playoffs without significant injuries to the other middle of the road teams (plus none to you) OR a move to improve, is significantly unlikely.
 

TonyGin&Juice

Magic City Refugee
Not including injuries makes everything afterward pointless.

Last season my sim model actually predicted a NDR WS win. I deleted it and never talked about it because I didn’t want to believe @Travis7401 would finally have a comeback to my “I got two WS and you ain’t got shit” posts.
 

NML

Well-Known Member
I just don’t think I could accurately predict injuries. Even if I had good research into how often a fragile player gets hurt, for example, people use those players in different ways that will limit/increase those injuries

And I don’t really feel like going through and setting depth charts for each team. Maybe next year.
 
It's also trickier because players aren't just blanket fragile; They have proneness in various areas which can changes things a lot. A SP with back proneness high enough to make him fragile may miss a month per season, but one with arm proneness is missing entire seasons at a time
 

Travis7401

Douglass Tagg
Community Liaison
Not including injuries makes everything afterward pointless.

Last season my sim model actually predicted a NDR WS win. I deleted it and never talked about it because I didn’t want to believe @Travis7401 would finally have a comeback to my “I got two WS and you ain’t got shit” posts.
I don't believe you! Nothing could predict how we shocked the world by winning it all!
 

Travis7401

Douglass Tagg
Community Liaison
I think the regular season results are generally pretty insulated from injury and reflect quality depth throughout the entire roster. That's why I think my MODEL generally did well predicting my regular season win totals. Everyone has injuries at some point and teams without depth get hit hardest. I think the playoffs tend to be the opposite, and highlight star power rather than depth, therefore an injury to a star player can really hurt your playoff chances more than it would in the regular season.

I think that difference explains why a lot of 4 seeds win the ship in this league when they might not have a great overall roster, but have some amazing star players and avoid injury to those players during the playoffs. I lost several playoff series where I clearly had the better TEAM throughout the season, but when you looked at the best 3-4 starting pitchers, best 2 BP pitchers, and starting lineups I actually was at a disadvantage compared to the "4 seed."

The exception to my injury/regular season thing is when you do have some decent depth but just happen to lose all the players from the same position, including your depth. I've had that happen before and I believe it happened @Karl Hungus 's OF last year. When you have an injury or two and you have to start some SCRUB player, that's a depth problem. When you have 5 injuries in the OF at one time, that's just bad luck.
 

Karl Hungus

Here to fix the cable
I think the regular season results are generally pretty insulated from injury and reflect quality depth throughout the entire roster. That's why I think my MODEL generally did well predicting my regular season win totals. Everyone has injuries at some point and teams without depth get hit hardest. I think the playoffs tend to be the opposite, and highlight star power rather than depth, therefore an injury to a star player can really hurt your playoff chances more than it would in the regular season.

I think that difference explains why a lot of 4 seeds win the ship in this league when they might not have a great overall roster, but have some amazing star players and avoid injury to those players during the playoffs. I lost several playoff series where I clearly had the better TEAM throughout the season, but when you looked at the best 3-4 starting pitchers, best 2 BP pitchers, and starting lineups I actually was at a disadvantage compared to the "4 seed."

The exception to my injury/regular season thing is when you do have some decent depth but just happen to lose all the players from the same position, including your depth. I've had that happen before and I believe it happened @Karl Hungus 's OF last year. When you have an injury or two and you have to start some SCRUB player, that's a depth problem. When you have 5 injuries in the OF at one time, that's just bad luck.

I feel like my team has some good players, but starting three AAA outfielders for two months was brutal. Morale goes in the tank, and the return of the starters isn't enough to counteract that.
 
My team is specifically built for the playoffs with 3 great starters and relievers who can go multiple innings. My batters 5-9 are consistently terrible, my 4th and 5th starters are subpar (but great relievers in the postseason), and the last few pen arms aren't as good, so I can't hover much further from .500. I think going back to the SL Championship year (where we beat Fax) we haven't won or lost more than 3 games by 3+ runs in the playoffs, which off the top of my head is at least a 25 game sample size.

NML's team that beat me in the playoffs had an even more extreme version of this formula which was humorous to watch. But yeah, playoff success from 'underdogs' definitely is more than just sample size luck
 

Travis7401

Douglass Tagg
Community Liaison
Well it certainly influenced my latest build! I set a goal to get the 3 best SPs and 2 best RPs I could, rather than just throwing a bunch of rigger pitchers out there, as is tradition.
 

TonyGin&Juice

Magic City Refugee
I just don’t think I could accurately predict injuries. Even if I had good research into how often a fragile player gets hurt, for example, people use those players in different ways that will limit/increase those injuries.
That's very true. Sorry if my response came off like a HOT TAEK, that wasn't my intent. I actually had more written out but it didn't save once the Motorhead started playing in my drum room and the sticks started flying.

Injuries are very hard to predict but I'd say they are probably one your best indicators of what to expect for team for a season. One of the reasons Wooly's team of cheaters did so well is that he set them all to durable and turned the injuries down to 0% probability. Sure, those guys were also all edited to "STUD" status but they being durable was how he won so many games. In looking back at some of my teams to see where things went to shit (other than trading away first rounders for pitchers that turned into bums and handing out stone cold stupid extensions) the 2061 season was where we really started falling apart. That team hit really well and had very solid pitching but the loss of our major defensive star CF Eric de Keizer the prior season really killed us defensively. The guy we replaced him with had a great bat but wasn't a CF and with No Glove Golf Mind in LF we lost a lot of games we should have won by not having any defense in 2/3 of the outfield.

I think that difference explains why a lot of 4 seeds win the ship in this league when they might not have a great overall roster, but have some amazing star players and avoid injury to those players during the playoffs. I lost several playoff series where I clearly had the better TEAM throughout the season, but when you looked at the best 3-4 starting pitchers, best 2 BP pitchers, and starting lineups I actually was at a disadvantage compared to the "4 seed."
That 2059 Miami team that bet HELS had the good fortune of Niek Ploeg, Luis Reyes, and Randy Gadd all being relatively health for the playoffs and the rotation having three legit lefty starters and a shutdown closer/stopper that wasn't worn out or injured. Looking back at that playoff roster I'm kind of shocked we scored as many runs as we did given how absolutely LOATHSOME that bottom third of the order was in the WS. I mean, does anyone remember @Travis7401 having a meltdown when BUM ASS Jean Foucault came in to play CF when one of the decent guys got injured?
 

TonyGin&Juice

Magic City Refugee
It wasn't so much that he came in, it was that he came in and played like a GOD :laughing:
Sure you aren't thinking of SL Shampionship MVP Hermann 'Digger' Wideen? That guy was the definition of playing the best when the lights were the brightest. Career WBL regular season triple slash of .256/.302/.418/.720 vs a playoff triple slash of .360/.389/.512/.901. Almost got left off the playoff roster but I feeling his personality with that team might be magic. Turned out I was right.
 

NML

Well-Known Member
Did some bad math earlier - standard deviation is actually 5.07 wins. So, about 68% of time we are actually within five games of predicted record. Not extremely different from an application stand point, but figured I'd share anyways.

Also ran numbers on last year's WAR by position. Here's the count of players who finished with above a 2 WAR by position:

[xtable=217x@]
{tbody}
{tr}
{td=92x@}C{/td}
{td=125x@}11{/td}
{/tr}
{tr}
{td}1B{/td}
{td}8{/td}
{/tr}
{tr}
{td}2B{/td}
{td}8{/td}
{/tr}
{tr}
{td}3B{/td}
{td}10{/td}
{/tr}
{tr}
{td}SS{/td}
{td}6 {/td}
{/tr}
{tr}
{td}LF{/td}
{td}6{/td}
{/tr}
{tr}
{td}CF{/td}
{td}10{/td}
{/tr}
{tr}
{td}RF{/td}
{td}6{/td}
{/tr}
{tr}
{td}DH{/td}
{td}2{/td}
{/tr}
{/tbody}
[/xtable]

I'm kind of amazed we only had 70 players finish with above 2 WARS WON. 2 WAR is sort of a generally accepted "starter" quality player. If we bump it to 4 and label these as potential all-stars, we get this:

[xtable=217x@]
{tbody}
{tr}
{td=92x@}C{/td}
{td=125x@}3{/td}
{/tr}
{tr}
{td}1B{/td}
{td}5{/td}
{/tr}
{tr}
{td}2B{/td}
{td}3{/td}
{/tr}
{tr}
{td}3B{/td}
{td}3{/td}
{/tr}
{tr}
{td}SS{/td}
{td}4{/td}
{/tr}
{tr}
{td}LF{/td}
{td}3{/td}
{/tr}
{tr}
{td}CF{/td}
{td}3{/td}
{/tr}
{tr}
{td}RF{/td}
{td}3{/td}
{/tr}
{/tbody}
[/xtable]

That's a total of 27, and it's very well spread out.

I was mostly curious if there was any position that was under/over saturated with talent, but that doesn't appear to be the case. Since this is based off of their listed position, I'm guessing a couple of those CFers are actually playing corner. Catcher seems to be a bit talent rich right now, while not surprisingly, shortstop seems to have the least.
 

Travis7401

Douglass Tagg
Community Liaison
Haha I don’t remember. It must’ve been in the season thread. That was like three analysis ago doe
It was last year:

WORLD LEAGUE
Amsterdam: 95-67
False Bay: 88-74
Lisbon: 86-76
Dublin: 85-77
Helsingborg: 80-82
United Kingdom: 80-82
Seoul: 79-83
North Dakota: 75-87

TOPIA LEAGUE
Moscow: 95-67
Las Vegas: 89-73
Kabul: 86-76
Buffalo: 77-85
Cairo: 72-90
Little Rock: 72-90
Miami: 71-91
Istanbul: 65-97


Then I lol'd at you and called my shot! "Lol at NDR last when we’re gonna win the ship."

My BTT tool predicted 90 wins, fwiw. Can't remember where I actually finished? 95 or something?
 

NML

Well-Known Member
Like I said, the analysis done then is nothing similar to what I have now. I'm pretty sure I was vastly underestimating rookies, of which you had a bunch. Also it was a snap shot of the season at one point, so any trades you made or any injuries you had were not included.

Looks like I still had you at 75 wins, it's just that varsity is so stack that meant last
 

NML

Well-Known Member
I mean it was literally done on some random sim in like mid-April lol. No one was supposed to make big assumptions about their team based off of that haha. I'll do one this year after the first sim of the season when we have accurate teams.

Hammers and Riggers over-performed, FB's entire team died, but otherwise I don't think I was that far off on most teams. I think Kabul had a lot of injuries in that sim so they were lower than they would be otherwise.