The Problem with Live Cards

Mizzery · 03-10-2020, 11:00 PM

I have another thread where I have shared the results of a three team test I started over the holidays after buying a second license when OOTP 20 went on sale. In a nutshell, the team built on power (high POW for hitters, high MOV for pitchers, bring fences in) has significantly underperformed my OBA team (High CON for batters, push fences out, increase park factors for hits).

I have learned a lot from other posters who in the past have shared their deep dive analysis into the statistics of the game, so I thought I would share my own and how I think the game unintentionally devalues current Live players.

I took all of my hitters across all teams that have more than 1,000 AB at some combination of Diamond and Perfect levels. I know other analysis has focused primarily on the Perfect level only, but I felt there was likely a large universe of players like myself that have teams that drift between Diamond championships and .500 Perfect league results, and while we constantly try to upgrade the rosters, the same team basically plays at both levels.

I also used my own cards rather than the entire universe of hitters at Perfect because I feel there are extreme high end players in Perfect that are generally unattainable for 95% of the teams, especially FTP players, and they could distort some of the results.

What was left was a database of 1.5 million at bats at Diamond and Perfect level, which I ran regression analysis on every possible combination of ratings and results I could think of.

As other posters have shared in the past, the hitting ratings do exactly what they are supposed to (r-squared correlations to follow): Avoid K vs. K% was at .9512, Eye vs. BB% was at .9116, Power vs. HR rate .7989, CON vs. Batting Average at .7112, and GAP vs. 2B and 3B rate at .7023.

But what combination of ratings equal success? WAR can't be used because it includes defensive ratings, and I played with TB% and Runs Created, but ultimately landed back on OPS as an easy benchmark, especially because it is agnostic to the rest of the players in the lineup.

The only hitting rating that showed any kind of decent statistical relationship against OPS was CON, at .3939- every other rating had a r-squared to OPS of .05 or less.

When combining the factors, the best combination I could come up with had a .4997 relationship to OPS- and that was a weighting of CON x 5.1, EYE x 1.6, POW x 1.33, and GAP x 1.

There are many other factors that come into play- quality of pitching the batter faces, park factors, defense, and BABIP, which is the one rating hidden on the cards. But I think there is a strong indication that Contact is the single biggest factor for a hitter's success at the higher levels.

Bringing this back to MLB Live Cards, 2019 had the highest HR rate in MLB history, and the highest K rate. Hitters in today's game have changed their swing angle to add loft on the ball to take advantage of the juiced ball and go for the home run, sacrificing contact and batting average in the process.

In OOTP, I have seen only a handful of players who can sustain results at Diamond and Perfect levels with a CON rating below 60, yet many Live players fall under that mark.

The 93 OVR Joey Gallo has a POW rating of 103 and a CON rating of 47. My Gallo card has a lifetime BA of .202 on 18k ABs before I finally gave up. The 90 OVR Giancarlo Stanton has a POW of 100, CON of 58, and my Stanton card has a lifetime .235 BA on 9500 AB, now firmly on my reserve roster. 92 OVR Mitch Garver, 85 POW, 51 CON. Eugenio Suarez, Pete Alonso, Gary Sanchez, etc. all fall in the same camp.

There are historical cards that underperform as well due to high POW, low CON (I've got a 99 OVR Mike Schmidt with a lifetime .217 BA), but the impact is heavier on the Live side.

Obviously this doesn't apply to every LIVE player- Cody Bellinger and Christian Yelich are examples of LIVE players with high POW and CON, but it does suggest that if the OOTP PT universe is tied to some average historical year to normalize for players from all eras, and if current players continue to drift away from historic hitting patterns, as a group, LIVE players may continue to underperform historic cards, regardless of the OVR rating.

I didn't look at pitchers yet, but other posters who have shared their results around MOV would seem to confirm this theory- as MLB hitters hit more HR, MLB pitchers serve up more HR, lowering their MOV rating and punishing them accordingly in the normalized OOTP PT universe.

rburgh · 03-13-2020, 06:13 AM

If the LIVE cards were given OVR ratings on the same scale as the historicals, there would be very few diamond level LIVE cards. That might scare away some users, who would conclude that baseball was no longer relevant.

But I believe that their batter ratings are simply based on standard deviation calculations for park adjusted HR rates, contact rates, XBH rates, walk rates, and K rates. And since baseball today has a much larger world population to draw from, there are a lot of people with very high-end skills, which kills off the upper tail of the normal distribution curve.

Babe Ruth hitting twice as many homers as any other TEAM was several SD's above the norm. Pete Alonso's 53 HR's doesn't score as well on that scale, since 9 other players last year hit more than 40. But he was still the best power hitter in baseball last year and deserves his 100 OVR POW rating. Compared to guys who had smaller populations, and were compared against fewer players, though, his power rating on an SD basis would probably be 85 or so. That would not sit well with Met fans or the younger demographic here.

Barmy Fungy-Phipps · 03-13-2020, 10:01 AM

They can't possibly be using standard deviation from the current season, so if they are it is from some other arbitrary baseline.

That's actually one of my main statistical concerns - I think it would be more accurate if ALL cards were normalized to a standard deviation for the year they are from.

Rain King · 03-13-2020, 10:08 AM

Quote:

Originally Posted by Barmy Fungy-Phipps

They can't possibly be using standard deviation from the current season, so if they are it is from some other arbitrary baseline.

That's actually one of my main statistical concerns - I think it would be more accurate if ALL cards were normalized to a standard deviation for the year they are from.

Agreed!

stl jason · 03-13-2020, 10:28 AM

I think the best course of action with any cards (live or historical), is to evaluate them based off their individual ratings and ignore the name on the cards and the overall rating (as hard as it might be to ignore the name Babe Ruth or Ty Cobb and not expect them to be superstars, that's how you have to approach PT).

prime example: Rogers Hornsby versus Zack Wheat

100 Rogers Hornsby:
contact=90
gap=83
power=88
eye=74
avoid k=51
speed=19
steal=62
running=60

100 Zack Wheat:
contact=109
gap=97
power=91
eye=56
avoid k=95
speed=55
steal=77
running=95

Wheat's career year in real life was 1922, so it makes a super easy comparison between the two players:

Hornsby 1922:
AB=623
Runs=141
Hits=250
2B=46
3B=14
HR=42
RBI=152
SB=17
CS=12
BB=65
SO=50
BA=.401
OBP=.460
SLG=.722
OPS=1.181
OPS+=207
TB=450

Wheat 1922:
AB=600
Runs=92
Hits=201
2B=29
3B=12
HR=16
RBI=112
SB=9
CS=6
BB=45
SO=22
BA=.335
OBP=.390
SLG=.503
OPS=.891
OPS+=128
TB=302

So for a quick comparison, Hornsby should easily surpass Wheat in Contact, Gap, and Power... slightly better in Eye; and Wheat gets the edge in Avoid K; yet the Wheat card ratings are crazy inflated compared to Hornsby.... this can be explained simply by the fact Wheat was released late in the lifecycle of v20, and was thus given some ridiculous ratings to make him a coveted card... the point is, don't expect the ratings on the card to match the performance of the player in real life 100% of the time.

(and from a quick browse of the v21 historical card list, it appears some of these issues might be addressed by withholding some cards for special releases, which should make it interesting to see how/when they show up)

chazzycat · 03-13-2020, 08:49 PM

I generally agree with your premise, but OPS doesn't seem like the right stat for this kind of analysis. Keep in mind that OPS is a pretty basic stat...it's just OBP plus SLG. OBP includes AVG, and walks. But SLG also includes AVG, and power. So if you think about it, AVG is doubly-represented in OPS compared to the power component or the eye component. So yeah...that's why you're seeing such disparities in the correlations. Because contact is being double counted. I'd suggest wRC+ as a more robust offensive stat.

That all being said...contact is still king in this game. I'm not disputing that. I'm just saying the difference isn't nearly as pronounced as .5 to .05. That kind of makes it seem like ratings other than contact don't matter in the slightest.

Having taken a FTP low-CON team to perfect level, I can confirm other ratings do matter at least a little more than that. I still have six live cards (3 normal, 2 all-stars, one POTM) and they are doing ok. Just don't look at their batting averages. For example my Trout this year is hitting .174! It's a tough league. But he's still an above average hitter (110 wRC+) from all the homers & walks.

Just to say it again though I do agree with your general premise - the league normalization process does generally have a nerfing effect on live cards, due primarily to their lower contact ratings and for pitchers their lower MOV. It's just not quite THAT bad. And the right park factors can mitigate some of it.

Orion · 03-17-2020, 08:54 PM

Quote:

Originally Posted by stl jason

I think the best course of action with any cards (live or historical), is to evaluate them based off their individual ratings and ignore the name on the cards and the overall rating (as hard as it might be to ignore the name Babe Ruth or Ty Cobb and not expect them to be superstars, that's how you have to approach PT).

prime example: Rogers Hornsby versus Zack Wheat

100 Rogers Hornsby:
contact=90
gap=83
power=88
eye=74
avoid k=51
speed=19
steal=62
running=60

100 Zack Wheat:
contact=109
gap=97
power=91
eye=56
avoid k=95
speed=55
steal=77
running=95

Wheat's career year in real life was 1922, so it makes a super easy comparison between the two players:

Hornsby 1922:
AB=623
Runs=141
Hits=250
2B=46
3B=14
HR=42
RBI=152
SB=17
CS=12
BB=65
SO=50
BA=.401
OBP=.460
SLG=.722
OPS=1.181
OPS+=207
TB=450

Wheat 1922:
AB=600
Runs=92
Hits=201
2B=29
3B=12
HR=16
RBI=112
SB=9
CS=6
BB=45
SO=22
BA=.335
OBP=.390
SLG=.503
OPS=.891
OPS+=128
TB=302

So for a quick comparison, Hornsby should easily surpass Wheat in Contact, Gap, and Power... slightly better in Eye; and Wheat gets the edge in Avoid K; yet the Wheat card ratings are crazy inflated compared to Hornsby.... this can be explained simply by the fact Wheat was released late in the lifecycle of v20, and was thus given some ridiculous ratings to make him a coveted card... the point is, don't expect the ratings on the card to match the performance of the player in real life 100% of the time.

(and from a quick browse of the v21 historical card list, it appears some of these issues might be addressed by withholding some cards for special releases, which should make it interesting to see how/when they show up)

Mmmmm Creamy Wheat.

Mizzery · 03-19-2020, 12:19 AM

Quote:

Originally Posted by chazzycat

I generally agree with your premise, but OPS doesn't seem like the right stat for this kind of analysis. Keep in mind that OPS is a pretty basic stat...it's just OBP plus SLG. OBP includes AVG, and walks. But SLG also includes AVG, and power. So if you think about it, AVG is doubly-represented in OPS compared to the power component or the eye component. So yeah...that's why you're seeing such disparities in the correlations. Because contact is being double counted. I'd suggest wRC+ as a more robust offensive stat.

I took your suggestion and reran my correlations, this time using wRC+, and the result was essentially the same. You are correct that using wRC+ weakened the correlation with CON by itself (from .39 using OPS to .26 using RC+), but CON still represented a 3x relationship in performance over EYE, GAP, or POW, which all came in at .07 or below.

When trying to combine the factors into one predictive measure, I once again could not get higher than around .50, and that was only after weighting CON at 4.5x, EYE at 1.75x, and leaving GAP and POW at 1x.

Not saying this is the ultimate in statistical analysis- as I indicated in my OP, I did not take the entire OT universe, just my 1.5 million at bats of it. I do think directionally though, this indicates that the CON rating for hitters is far and away the most significant rating for forecasting offensive performance in the current PT environment.

Billy Beane would be proud that getting on base (CON + EYE) seem to be the most critical hitting ratings, but it reinforces my original point that the PT universe is unlikely to be favorable to many of the current hitting trends in Live cards.

chazzycat · 03-19-2020, 10:00 PM

yep that sounds much closer to the truth. 3x is more reasonable than 10x for sure, but yes, contact is still king!

03-10-2020, 11:00 PM	#1
Mizzery Minors (Single A) Join Date: Feb 2019 Posts: 99	The Problem with Live Cards I have another thread where I have shared the results of a three team test I started over the holidays after buying a second license when OOTP 20 went on sale. In a nutshell, the team built on power (high POW for hitters, high MOV for pitchers, bring fences in) has significantly underperformed my OBA team (High CON for batters, push fences out, increase park factors for hits). I have learned a lot from other posters who in the past have shared their deep dive analysis into the statistics of the game, so I thought I would share my own and how I think the game unintentionally devalues current Live players. I took all of my hitters across all teams that have more than 1,000 AB at some combination of Diamond and Perfect levels. I know other analysis has focused primarily on the Perfect level only, but I felt there was likely a large universe of players like myself that have teams that drift between Diamond championships and .500 Perfect league results, and while we constantly try to upgrade the rosters, the same team basically plays at both levels. I also used my own cards rather than the entire universe of hitters at Perfect because I feel there are extreme high end players in Perfect that are generally unattainable for 95% of the teams, especially FTP players, and they could distort some of the results. What was left was a database of 1.5 million at bats at Diamond and Perfect level, which I ran regression analysis on every possible combination of ratings and results I could think of. As other posters have shared in the past, the hitting ratings do exactly what they are supposed to (r-squared correlations to follow): Avoid K vs. K% was at .9512, Eye vs. BB% was at .9116, Power vs. HR rate .7989, CON vs. Batting Average at .7112, and GAP vs. 2B and 3B rate at .7023. But what combination of ratings equal success? WAR can't be used because it includes defensive ratings, and I played with TB% and Runs Created, but ultimately landed back on OPS as an easy benchmark, especially because it is agnostic to the rest of the players in the lineup. The only hitting rating that showed any kind of decent statistical relationship against OPS was CON, at .3939- every other rating had a r-squared to OPS of .05 or less. When combining the factors, the best combination I could come up with had a .4997 relationship to OPS- and that was a weighting of CON x 5.1, EYE x 1.6, POW x 1.33, and GAP x 1. There are many other factors that come into play- quality of pitching the batter faces, park factors, defense, and BABIP, which is the one rating hidden on the cards. But I think there is a strong indication that Contact is the single biggest factor for a hitter's success at the higher levels. Bringing this back to MLB Live Cards, 2019 had the highest HR rate in MLB history, and the highest K rate. Hitters in today's game have changed their swing angle to add loft on the ball to take advantage of the juiced ball and go for the home run, sacrificing contact and batting average in the process. In OOTP, I have seen only a handful of players who can sustain results at Diamond and Perfect levels with a CON rating below 60, yet many Live players fall under that mark. The 93 OVR Joey Gallo has a POW rating of 103 and a CON rating of 47. My Gallo card has a lifetime BA of .202 on 18k ABs before I finally gave up. The 90 OVR Giancarlo Stanton has a POW of 100, CON of 58, and my Stanton card has a lifetime .235 BA on 9500 AB, now firmly on my reserve roster. 92 OVR Mitch Garver, 85 POW, 51 CON. Eugenio Suarez, Pete Alonso, Gary Sanchez, etc. all fall in the same camp. There are historical cards that underperform as well due to high POW, low CON (I've got a 99 OVR Mike Schmidt with a lifetime .217 BA), but the impact is heavier on the Live side. Obviously this doesn't apply to every LIVE player- Cody Bellinger and Christian Yelich are examples of LIVE players with high POW and CON, but it does suggest that if the OOTP PT universe is tied to some average historical year to normalize for players from all eras, and if current players continue to drift away from historic hitting patterns, as a group, LIVE players may continue to underperform historic cards, regardless of the OVR rating. I didn't look at pitchers yet, but other posters who have shared their results around MOV would seem to confirm this theory- as MLB hitters hit more HR, MLB pitchers serve up more HR, lowering their MOV rating and punishing them accordingly in the normalized OOTP PT universe.

03-13-2020, 06:13 AM	#2
rburgh All Star Reserve Join Date: Nov 2012 Location: Pittsburgh Posts: 733	And also If the LIVE cards were given OVR ratings on the same scale as the historicals, there would be very few diamond level LIVE cards. That might scare away some users, who would conclude that baseball was no longer relevant. But I believe that their batter ratings are simply based on standard deviation calculations for park adjusted HR rates, contact rates, XBH rates, walk rates, and K rates. And since baseball today has a much larger world population to draw from, there are a lot of people with very high-end skills, which kills off the upper tail of the normal distribution curve. Babe Ruth hitting twice as many homers as any other TEAM was several SD's above the norm. Pete Alonso's 53 HR's doesn't score as well on that scale, since 9 other players last year hit more than 40. But he was still the best power hitter in baseball last year and deserves his 100 OVR POW rating. Compared to guys who had smaller populations, and were compared against fewer players, though, his power rating on an SD basis would probably be 85 or so. That would not sit well with Met fans or the younger demographic here.

03-13-2020, 10:28 AM	#5
stl jason Hall Of Famer Join Date: Apr 2006 Posts: 9,943	I think the best course of action with any cards (live or historical), is to evaluate them based off their individual ratings and ignore the name on the cards and the overall rating (as hard as it might be to ignore the name Babe Ruth or Ty Cobb and not expect them to be superstars, that's how you have to approach PT). prime example: Rogers Hornsby versus Zack Wheat 100 Rogers Hornsby: contact=90 gap=83 power=88 eye=74 avoid k=51 speed=19 steal=62 running=60 100 Zack Wheat: contact=109 gap=97 power=91 eye=56 avoid k=95 speed=55 steal=77 running=95 Wheat's career year in real life was 1922, so it makes a super easy comparison between the two players: Hornsby 1922: AB=623 Runs=141 Hits=250 2B=46 3B=14 HR=42 RBI=152 SB=17 CS=12 BB=65 SO=50 BA=.401 OBP=.460 SLG=.722 OPS=1.181 OPS+=207 TB=450 Wheat 1922: AB=600 Runs=92 Hits=201 2B=29 3B=12 HR=16 RBI=112 SB=9 CS=6 BB=45 SO=22 BA=.335 OBP=.390 SLG=.503 OPS=.891 OPS+=128 TB=302 So for a quick comparison, Hornsby should easily surpass Wheat in Contact, Gap, and Power... slightly better in Eye; and Wheat gets the edge in Avoid K; yet the Wheat card ratings are crazy inflated compared to Hornsby.... this can be explained simply by the fact Wheat was released late in the lifecycle of v20, and was thus given some ridiculous ratings to make him a coveted card... the point is, don't expect the ratings on the card to match the performance of the player in real life 100% of the time. (and from a quick browse of the v21 historical card list, it appears some of these issues might be addressed by withholding some cards for special releases, which should make it interesting to see how/when they show up) __________________ Beer League dynasty thread - knock 'em back!

03-13-2020, 10:01 AM	#3
Barmy Fungy-Phipps Minors (Rookie Ball) Join Date: Mar 2015 Posts: 41	They can't possibly be using standard deviation from the current season, so if they are it is from some other arbitrary baseline. That's actually one of my main statistical concerns - I think it would be more accurate if ALL cards were normalized to a standard deviation for the year they are from.

03-13-2020, 08:49 PM	#6
chazzycat All Star Starter Join Date: Dec 2014 Posts: 1,685	I generally agree with your premise, but OPS doesn't seem like the right stat for this kind of analysis. Keep in mind that OPS is a pretty basic stat...it's just OBP plus SLG. OBP includes AVG, and walks. But SLG also includes AVG, and power. So if you think about it, AVG is doubly-represented in OPS compared to the power component or the eye component. So yeah...that's why you're seeing such disparities in the correlations. Because contact is being double counted. I'd suggest wRC+ as a more robust offensive stat. That all being said...contact is still king in this game. I'm not disputing that. I'm just saying the difference isn't nearly as pronounced as .5 to .05. That kind of makes it seem like ratings other than contact don't matter in the slightest. Having taken a FTP low-CON team to perfect level, I can confirm other ratings do matter at least a little more than that. I still have six live cards (3 normal, 2 all-stars, one POTM) and they are doing ok. Just don't look at their batting averages. For example my Trout this year is hitting .174! It's a tough league. But he's still an above average hitter (110 wRC+) from all the homers & walks. Just to say it again though I do agree with your general premise - the league normalization process does generally have a nerfing effect on live cards, due primarily to their lower contact ratings and for pitchers their lower MOV. It's just not quite THAT bad. And the right park factors can mitigate some of it.

03-19-2020, 10:00 PM	#9
chazzycat All Star Starter Join Date: Dec 2014 Posts: 1,685	yep that sounds much closer to the truth. 3x is more reasonable than 10x for sure, but yes, contact is still king!