*CB Attribute Eval:Brains, Strength, or Technique?

*CB Attribute Eval:Brains, Strength, or Technique? - Printable Version

+- International Simulation Football League (https://forums.sim-football.com)
+-- Forum: Community (https://forums.sim-football.com/forumdisplay.php?fid=5)
+--- Forum: Media (https://forums.sim-football.com/forumdisplay.php?fid=37)
+---- Forum: Graded Articles (https://forums.sim-football.com/forumdisplay.php?fid=38)
+---- Thread: *CB Attribute Eval:Brains, Strength, or Technique? (/showthread.php?tid=20628)

Pages: 1 2

*CB Attribute Eval:Brains, Strength, or Technique? - iStegosauruz - 03-30-2020

[div align=\\\"center\\\"]Background and Methodology –[/div]
I simulated approximately 100,800 games so that you wouldn’t have to.

Well, that’s a bit misleading. I did simulate approximately 100,800 games, however, the data presented in this study will only cover approximately 50,400 games. The sample size was only supposed to be 50,400 games – and I completed the first run of that data several days ago – however Windows decided to be a stupid operating system and wiped half of the data from those games that I needed, meaning I had to completely redo all of the simulations.

About a week ago @`Laser` approached me about potentially looking into the value of strength on cornerbacks. My player is a CB so I was already interested in the efficiency of various attributes on the position so I decided to jump right in and study it. I chose to examine more than strength so that I would have a baseline of what to compare its efficiency too. The other attributes I chose to test were hands, intelligence, and tackling.

As always with my studies I ran batches of simulations for various different groupings. To determine the groupings for this study I had to first determine the baseline control group. I used the TPE tracker to find the attributes for every team’s highest TPE cornerbacks. I realize that in several situations a team may be playing one of these CBs at safety, however I decided that for the purposes of this study those variations were of little importance.

After finding the attribute values for those 20 cornerbacks I averaged them together to find the attribute values for the average CB in the league. The average CB looks like this:

[div align=\\\"center\\\"] [Image: V1nim0J.png]

[/div]

I chose to round the values to the nearest whole number – so values such as Tackle rounded up from 61.85 to 62.

I then calculated the average height and weight for the CBs. In general, it is most efficient for positions to be max height and weight. I averaged the max height and weight for the various CB archetypes and determined that the CBs in my simulations would be 6’2” and 207 pounds.

For simulation purposes I chose to use the Orange County Otters because Laser is one of Orange County’s front office members. This means that Kacey Dream and AJ Lattimer became the guinea pig CBs for the study.

After I had determined all of that I formulated the values at which I would test the performance of these CBs. I would keep all of their attributes at the league average values and raise on for each batch. I chose to test strength at values of 70, 80, 90 and 100; tackling at values of 70, 80, 90, and 100; intelligence and values of 90 and 100; and hands at values of 80, 90, and 100.

I ran four sets of simulations at each variable grouping. Each set was 900 simulations each. Two of the four sets were at home, two of the four sets were on the road. That means each grouping has 3,600 simulations worth of data.

I also determined that I’d look at the impact raising these attributes would have on a team’s overall performance – meaning things such as a team’s winning percentage, average total yards allowed, average rush yards allowed, average pass yards allowed, and average interceptions forced – and also at the impact they have on an individual CB’s performance – meaning things such as average interceptions per game, average passes defended per game, and average tackles per game.

[div align=\\\"center\\\"]Data and Calculations
Hands
Team Statistics[/div]
For the three cornerback archetypes, hands max at between 70 and 85. The average CB in the league has a hand’s value of 74. As would be expected, increasing hands does have an increase in a team’s winning percentage.

[div align=\\\"center\\\"] [Image: RTFsJCn.png]

[/div]

A team’s winning percentage increases from 55.32% at the average CB hand’s value of 74 to 58% at a hand’s value of 100. This is an increase of 2.68%. If the hands value increases to the average cap – so moving from 74 to 80 – a team’s winning percentage increases from 55.32% to 55.67% - a very minor increase of 0.35%.

[div align=\\\"center\\\"] [Image: 856vOOe.png]

[/div]

There was also a consistent increase in the average amount of interceptions the Otters forced per game as the hand’s value of their CBs increased. The control group forced .6775 interceptions per game. At the average hand’s value cap of 80 their CBs were forcing 0.6925 interceptions per game – an increase of 0.015 interceptions per game. At the max possible hand’s value of 100 the Otters were forcing 0.77 interceptions per game – an increase of 0.0925 interceptions per game.

[div align=\\\"center\\\"] [Image: FkZLXQl.png]

[/div]

[div align=\\\"center\\\"]Player Statistics[/div]
The increase in hand’s value caused an increase in the amount of interceptions and passes defended the individual CB’s averaged per game. The increase in average interceptions per game also caused a minor increase in the average amount of defensive touchdowns the CB’s scored per game – more interceptions mean more are run back for touchdowns.

The control group averaged 0.12125 interceptions per game while the average hand’s cap of 80 averaged 0.12625. This is a minor increase of .005 interceptions per game. The max hand value group of 100 averaged 0.16375 interceptions per game – a difference of 0.0425 interceptions per game.

[div align=\\\"center\\\"] [Image: mugp5CF.png]

[/div]

The control group averaged 1.06625 passes defended per game. The average hand’s cap of 80 averaged 1.045 per game – a decrease of 0.2125 from the control group and probably within the margin of error. The max hand value group of 100 averaged 1.1375 passes defended per game – an increase of 0.07125 per game.

[div align=\\\"center\\\"] [Image: e1nrSoD.png]

[/div]

[div align=\\\"center\\\"]Intelligence
Team Statistics[/div]
For the three cornerback archetypes, intelligence maxes at either 80 or 85. The league average CB has an intelligence value of 79. As would be expected, increasing intelligence does have an impact on a team’s winning percentage.

[div align=\\\"center\\\"] [Image: gSozWIW.png]

[/div]

A team’s winning percentage increases from 55.32% at the average CB intelligence value of 79 to 57.94% at an intelligence value of 100. This is an increase of 2.63% - very comparable to the increase in a team’s winning percentage when the hand’s attribute was maxed.

[div align=\\\"center\\\"] [Image: CIaY7rN.png]

[/div]

Opponents also average fewer total yards, pass yards, and rush yards as the hand’s value increases. The control group allowed an average of 334.195 total yards, 226.055 pass yards, and 108.1375 rush yards per game to opponents. At the max intelligence value of 100 the Otters were surrendering 331.6475 total yards, 225.4925 pass yards, and 106.155 rush yards per game to opponents. These are differences of 2.55 total yards, 0.5625 pass yards, and 1.9825 rush yards per game. For intelligence the increase of maxing the value to 100 had a greater impact on the run game than the pass game - the opposite effect maxing hands had. I hypothesize that this is because intelligence has some impact on the ability of a CB to read whether a play is a run or a pass and react accordingly. The higher the intelligence the greater their ability to read and react, meaning they get to the play quicker and are able to contribute in a stop before their opponents advance the ball farther.

[div align=\\\"center\\\"] [Image: 6UQz8vA.png]

[/div]

There were not consistent patterns in the amount of turnovers forced as the intelligence value increased. The max intelligence group - 100 - did force more interceptions per game than the control group, however, the group in the middle - 90 - forced less than the control group. There are only three data points so I do not feel comfortable drawing a conclusion from this data.

[div align=\\\"center\\\"]Player Statistics [/div]
The increase in intelligence value caused an increase in the amount of interceptions the individual CB’s averaged per game. There was also an increase in the average amount of passes defended per game from the control group to the max group - 100 - but the group in the middle - 90 - was the lowest value, so I don’t feel comfortable drawing a conclusion on the impact intelligence has on passes defended from this sample of data.

The control group averaged 0.12125 interceptions per game while the max intelligence group - 100 - averaged 0.13375 interceptions per game. This is an increase of 0.0125 interceptions per game. As with the hand’s value the increase in individual interceptions per game also coincided with an increase in individual defensive touchdowns per game. More individual interceptions mean more opportunities for defensive touchdowns.

[div align=\\\"center\\\"] [Image: YR4sZo9.png]

[/div]

[div align=\\\"center\\\"]Tackling
Team Statistics [/div]
For the three cornerback archetypes, tackle maxes at either 70 or 80. The average CB in the league has a tackle value of 62. Tackle is the first attribute I’ve tested for this study that did not have a noticeable impact on a team’s winning percentage.

[div align=\\\"center\\\"] [Image: P9x5oCn.png]

[/div]

The Otter’s winning percentage was 55.32% in the control grouping. All other groupings were higher than it - the lowest difference being the 80 tackle group and the highest difference being the 90 tackle group. The 80 tackle group had a 55.83% winning percentage - a 0.51% difference. The 90 tackle group had a 56.75% winning percentage - a 1.43% difference. There was not a consistent upward trend, however, meaning these numbers are not only insignificant because of the low increase in percentage but non-conclusive at determining if a CB’s tackle attribute has any substantive impact on a team’s winning percentage.

[div align=\\\"center\\\"] [Image: eRMJ2NF.png]

[/div]

Opponents did consistently average fewer total yards per game and fewer pass yards per game. Rush yards were not substantively impacted. The control group allowed an average of 334.195 total yards per game and an average of 226.055 pass yards per game. Although the values at the variable groupings did not consistently decrease they never increased past the control value. I do not feel comfortable drawing many substantial conclusions from this data, however it does feel like there is a light correlation between a CB’s tackle attribute and the total yards and pass yards a team surrenders. It is a weaker correlation than both a CB’s hand attribute and intelligence attribute.

[div align=\\\"center\\\"] [Image: iUUhFjO.png]

[/div]

There is not a significant impact in the amount of turnovers a team forces as their CB’s tackle attribute scales. Four of the five variable groupings had higher average interceptions forced per game than the control group, however, the 80 tackle grouping had a significantly lower average interceptions forced per game value than the control group. There is the possibility that this is an outlier or a blip in the data, however, I tend to believe that because this is such a large sample of data that even if it is a blip or an outlier that it also shows that there isn’t a strong correlation to assume a relation between the tackle attribute and average interceptions per game.

[div align=\\\"center\\\"] [Image: n0dEMb8.png]

[/div]

The average amount of fumbles a team forces per game has the same issue that the interceptions data has. Two of the four variable groups are greater than or equal to the control group, however the other two variable groups are lower than the control group. I cannot draw any significant conclusions between a CB’s tackle attribute and the average amount of fumbles a team forces per game.

[div align=\\\"center\\\"] [Image: GqHim2i.png]

[/div]

[div align=\\\"center\\\"]Player Statistics [/div]
The increase in tackle value surprisingly did not cause an increase in the average amount of tackles an individual CB got per game. The only individual statistics I tracked that increased with the increases in tackle value was forced fumbles. The control group had individual CBs forcing an average of 0.00875 fumbles per game. This figure gradually increased as the tackle value increased, reaching an average of 0.01125 fumbles per game at both 70 tackle and 80 tackle - the highest tackle value that the current CB archetypes can reach - and an average of 0.015 fumbles per game at 100 tackle.

[div align=\\\"center\\\"] [Image: vcbwQGA.png]

[/div]

[div align=\\\"center\\\"]Strength
Team Statistics [/div]
For the three cornerback archetypes, strength maxes at either 65 or 70. The average CB in the league has a strength value of 57. As would be expected, increasing strength does cause an increase in a team’s winning percentage.

[div align=\\\"center\\\"] [Image: HcLW0KF.png]

[/div]

The Otter’s winning percentage was 55.32% for the control group. All variable groupings steadily increased this percentage. The highest strength value a CB can currently achieve under the current archetypes - 70 - had a 58.75% winning percentage. This is a 3.43% increase from the control group. This is the largest increase in winning percentage any attribute I’ve tested thus far has caused. The highest strength value I tested - 100 - had a 62.81% winning percentage. This is a substantial 7.49% higher than the control group.

[div align=\\\"center\\\"] [Image: QN1XKy9.png]

[/div]

Opponents also consistently average less total yards per game, less pass yards per game, and less rush yards per game as the strength value increased. The control group allowed an average of 334.195 total yards per game, 226.055 pass yards per game, and 108.1375 rush yards per game. The highest strength value a CB can currently achieve - 70 - allowed an average of 326.64 total yards per game, 218.3425 pass yards per game, and 108.2975 rush yards per game. These are differences of 7.56 total yards per game and 7.7125 pass yards per game. Average rush yards allowed per game did increase by 0.16 yards per game, however each out grouping I tested caused consistent decreases in rush yards allowed per game so I think that this just a blip in the data.

At the highest strength value I tested - 100 - the Otters allowed an over of 313.375 total yards per game, 206.2325 pass yards per game, and 107.145 rush yards per game. These are differences of 20.82 total yards per game, 19.8225 pass yards per game, and 0.9925 rush yards per game. These are significant numbers and the biggest change I found as a result of any of the attributes I tested in this study.

[div align=\\\"center\\\"] [Image: B2uz34F.png]

[/div]

Strength is also the first attribute I tested that caused a significant decrease in the opponent’s average yards per pass attempt. I did not calculate adjusted yards per pass attempt, but the rough calculation that takes total pass yards and divides it by attempts is still relevant. Most of the other attributes I tested did cause minor decreases in this metric, however strength caused substantial decreases. The control group allowed 7.1225 yards per pass attempt to opponents while the highest strength value a CB can currently achieve under the designed archetypes - 70 - allowed 6.84 yards per pass attempt. This is a 0.2825 yard per pass attempt difference. The 100 strength grouping I tested allowed a study low 6.445 yards per pass attempt to opponent quarterbacks. This is a 0.6775 yard difference from the control group.

[div align=\\\"center\\\"] [Image: sy3uVtR.png]

[/div]

There were not substantial or consistently changes in the average amount of interceptions forced per game as strength increased. Three of the four variable groups did register a value higher than the control group, however, the highest was the 80 strength grouping while the lowest, and the group to score below the control group, was the 100 strength grouping.

[div align=\\\"center\\\"] [Image: 0f3o1cc.png]

[/div]

Three of the four variable groups did register a value for average fumbles forced per game higher than the control group, however in this metric the 80 strength group scored lower than the control group. The 100 strength group was the highest in this metric, however, which means that there is probably a stronger correlation between strength and fumbles forced then there is between strength and interceptions. Neither of the correlations is particularly strong, however.

[div align=\\\"center\\\"] [Image: jEPx4L3.png]

[/div]

[div align=\\\"center\\\"]Player Statistics [/div]
The increase in strength value only caused a noticeable increase in the individual CB’s average fumbles forced per game. Average passes defended per game did increase, however both the 70 and 80 strength groups were lower than the control while the 90 and 100 strength groups were significantly higher than the control group. I decided not to put much stock in the correlation between strength and average passes defended per game because of this.

[div align=\\\"center\\\"] [Image: aE4ERCX.png]

[/div]

For forced fumbles, the control group had CB’s individually forcing an average of 0.00875 fumbles per game. This metric consistently increased with one small blip in the 80 strength group. At the 70 strength grouping - the highest strength value the current CB archetypes can achieve - the CBs were individually forcing 0.01125 fumbles per game, a 0.0025 fumble per game difference. At the highest strength value I tested - 100 - the CBs were individually forcing an average of 0.0175 fumbles per game, a 0.00875 increase over the control group.

[div align=\\\"center\\\"] [Image: 3IOSvK1.png]

[/div]

[div align=\\\"center\\\"]Conclusions [/div]
There’s one thing I want to note before I get into the conclusions. With a lot of these statistics we’re dealing with small increases - literal quarter of a percent increases in fumbles per game, etc - however in a sample this size and with the amount of range that those figures have to change those increases may look like small numbers but they’re pretty impactful anyway. Keep that in mind.

1. Of the attributes tested - strength, hands, intelligence, and tackling - strength had the largest impact on a team’s success on the field. That makes it one of - if not the most - valuable attribute for cornerbacks to max after speed.
2. Strength has a lot of value to being maxed second because it has such a low cap - 65 or 70. I think there is definitely an argument to potentially increasing that cap for cornerbacks.
3. The attributes tested only have minor impacts on personal statistics. I haven’t tested it but I’d wager that a lot of the individual statistics are heavily influenced by a cornerback’s speed. The greater their ability to be in a play the more individual attributes they’ll accrue.
4. If team’s are looking for cornerbacks who will have the ability to contribute on the field quickly in this upcoming draft they should prioritize those with high speed and high strength.

[div align=\\\"center\\\"]Random Plugs [/div]
1. This was literally the longest study I’ve done both from the sample size - 50,400 sims - but also from the fact I had to completely redo it. Hopefully y’all like it. Feel free to ask questions!
2. As always, my data is open source and can be found here.
3. Graders please give @`Laser` a bonus or a portion of my payout. I’ll let y’all determine what is appropriate. He ran some simulations for me that I ended up using as data validation early on. He also let me constantly pester him for the last week about this study.

*CB Attribute Eval:Brains, Strength, or Technique? - Bigred1580 - 03-30-2020

Holy crap man! Is this a TC multi?!?!?

*CB Attribute Eval:Brains, Strength, or Technique? - Huskies311 - 03-30-2020

(03-30-2020, 06:27 PM)Bigred1580 Wrote:Holy crap man! Is this a TC multi?!?!?

I’m saying. Some of the best work I’ve ever seen in the league

*CB Attribute Eval:Brains, Strength, or Technique? - Bigred1580 - 03-30-2020

(03-30-2020, 05:30 PM)Huskies311 Wrote:I’m saying. Some of the best work I’ve ever seen in the league

Yeah steg, In case you didn’t know that is the best compliment I can think to give someone

*CB Attribute Eval:Brains, Strength, or Technique? - Bayley - 03-30-2020

How can we give you all the money?

*CB Attribute Eval:Brains, Strength, or Technique? - flyeaglesfly29 - 03-30-2020

Awesome work man!

*CB Attribute Eval:Brains, Strength, or Technique? - iStegosauruz - 03-30-2020

(03-30-2020, 05:30 PM)Huskies311 Wrote:I’m saying. Some of the best work I’ve ever seen in the league

(03-30-2020, 05:31 PM)Bigred1580 Wrote:Yeah steg, In case you didn’t know that is the best compliment I can think to give someone

(03-30-2020, 05:35 PM)flyeaglesfly29 Wrote:Awesome work man!

Thanks guys! It was a grind to finish but definitely interesting in the end.

*CB Attribute Eval:Brains, Strength, or Technique? - Memento Mori - 03-30-2020

This is so, so good. As always.

*CB Attribute Eval:Brains, Strength, or Technique? - Vorshayla - 03-30-2020

gonna retire and recreate now

k thanks

*CB Attribute Eval:Brains, Strength, or Technique? - infinitempg - 03-30-2020

how much free time do you have and how do i go about borrowing some