03-15-2020, 07:29 PM
(This post was last modified: 03-15-2020, 11:01 PM by iStegosauruz.)
[div align=\\\"center\\\"]Background:[/div]
I simulated 25,199 total games so no one else would have to. Several days ago, there was a debate in the NSFL General Discord Chat about the value of offensive lineman. The claim was made that a human offensive tackle at 300TPE was as valuable as a 550TPE bot offensive tackle. I decided to investigate the validity of this claim.
The NSFL salary cap is $80 million per season. The average NSFL team spends 25.81% of their cap space - $20.65 million a season – on offensive lineman bots. Based off data from the TPE and stats databases there are only three active offensive linemen in the league. One – Edmund Beaver – plays for the Austin Copperheads while the other two – Brave Ulysses and Givussafare Rubbe – play for the New Orleans Second Line. Beaver is a 535TPE lineman making $2 million this season. Ulysses is a 727TPE lineman making $3 million this season. Rubbe is a 709TPE lineman making $2 million this season.
All three active linemen make less than the average bot a team purchases – a 550TPE lineman which costs $4.5 million a season. If a team can get the same production out of a 300TPE human offensive tackle and pay them equivalently to what teams are currently paying active human offensive linemen - $2.3 million on average – they can save $2.2 million per lineman and $4.4 million per season compared to the cost of a bot offensive tackle.
[div align=\\\"center\\\"]Methodology:
[/div]
I only focused on offensive tackles because the rules stipulate that a team’s bot offensive tackles can only weigh 310 pounds. A human offensive tackle can weigh 340 pounds. I chose to use the Orange County Otters as my control team because they have the best representation of the average bot offensive line in the league. They currently pay for five bot offensive linemen with 550TPE apiece. Their tackles are 6’7” and 310 pounds – about league average for bot offensive tackles.
I ran two sets of 100 game simulations at seven different TPE/Weight levels against every one of the other nine teams for both home and away. This means that were 200 total games played with the Otters at home at every TPE/Weight combination against every other team in the league and then 200 total games played with the Otters away at every TPE/Weight combination against every other team in the league. The Otters thus played 1400 total games at home against all nine teams and 1400 games away.
The TPE/Weight combinations were divided as such:
[div align=\\\"center\\\"][/div]
The control group is the 550TPE/310lbs group.
[div align=\\\"center\\\"]Results:[/div]
The first simulations pitted the Otters as the home team against every other team in the league. I ran two sets of 100 simulations at each TPE/Weight point to avoid the simulation crashing. The runs with the Otters as the home teams resulted in the following:
[div align=\\\"center\\\"][/div]
When both sets of data were combined the control group won 72.56% of their games at home. As expected, once the weight was increased from 310lbs to 340lbs on the control amount of TPE the win percentage increased by 2.83% to 75.39% total. What is interesting is that all variable TPE/Weight combinations won a greater percentage of their games at home than the control group. The lowest variable combination was 200TPE and 340lbs. This group won 73.15% of their games at home – 0.59% more than the control group.
The Otters could conceivably have two human offensive tackles at the maximum weight of 340 pounds with TPE as low as 200 and still have as high of a win percentage at home as they do with their two 550TPE bots that cost them $4.5 million per season apiece.
These results represented graphically:
[div align=\\\"center\\\"][/div]
The second simulations pitted the Otters as the away team against every other team in the league. Once again, I ran two sets of 100 simulations at each TPE/Weight point to avoid the simulation crashing. The runs with the Otters as the away team resulted in the following:
[div align=\\\"center\\\"][/div]
When both sets of data were combined the control group won 49.58% of their games as the away team. Only one variable combination – 350TPE and 340lbs – was greater than this at a 49.61% winning percentage. Strangely, the greatest gap in winning percentage between the control group and a variable combination was when the variable combination was 550TPE and 340lbs. That group won 46.72% of their games – 2.86% less than the control group. On whole, however, the gaps in winning percentage are not statistically significant.
These results represented graphically:
[div align=\\\"center\\\"][/div]
I then combined the home and away data sets into one combined set of data. This means that each grouping now had 1800 total games in the each run respectively.
[div align=\\\"center\\\"][/div]
The control group had a total win percentage of 61.06%. Of the six variable groups three had higher winning percentage, one had the same winning percentage, and two had lower winning percentage. The gap between the control and the highest was the 250TPE 340lbs group which had a winning percentage of 61.86% - a 0.8% gap. The gap between the control and the lowest was the 200TPE 340lbs group which had a winning percentage of 60.60% - a .46% gap. On whole, all variable groups performed about as well as the control group. There was not a noticeable gap in winning percentage.
These results represented graphically:
[div align=\\\"center\\\"][/div]
[div align=\\\"center\\\"]Conclusions 1.0: [/div]
Teams can get almost equivalent production out of a human offensive tackle weighing 340 pounds with 200TPE as they would a bot offensive tackle weighing 310 pounds with 550TPE. If these human offensive linemen would be paid equivalent to the current rate a human offensive lineman is paid – about $2.3 million – teams could save $4.4 million per season by having human offensive tackles.
It is always feasible these human tackles would be paid less as well. At 200TPE they would be 457TPE less than the current average human offensive lineman. This logic can also be flipped and they could always make more than the $2.3 million current average. As long as they make less than the $4.5 million teams currently pay for the 550TPE bot offensive tackles it is still beneficial for the team.
[div align=\\\"center\\\"]Deeper Dive: [/div]
Since I had data from 25,199 simulated games, I also decided to look at the statistical performance at each variable combination when it came to fumbles and yards per game.
The data collection looks the same for this as it did for the win percentage analysis. Two runs of 100 games at each variable combination for both home and away.
I first looked at fumbles:
[div align=\\\"center\\\"]
[/div]
Although there is a trend of increasing fumbles lost percentage for the variable combinations as TPE decreases in the simulations where the Otters were the away team, it is not particularly statistically significant. The gap only reaches 1.09% at its highest point. When looking at all the runs the percentages balance out again and there doesn’t appear to be a strong trend for the variable combinations having a higher percentage of fumbles lost than the control group.
After looking at fumbles I looked at yards per game to see if the teams performed as well offensively at each variable group.
[div align=\\\"center\\\"][/div]
In general, when looking at total yards per game the variable combinations generally had higher total yards per game, rush yards per game, and pass yards per game at the 550TPE, 450TPE, and 350TPE marks. The yards per game numbers for these categories began to even out to the control group around the 300TPE and 250TPE variable combination groups. On whole, however, there is not a particularly large gap in the performance of any of the variable combination groups when compared to the control group. The team performed very similarly across all simulations with all variable groups.
Graphical representation of total yards per game across groupings:
[div align=\\\"center\\\"][/div]
Graphical representation of rush yards per game across groupings:
[div align=\\\"center\\\"][/div]
Graphical representation of pass yards per game across groupings:
[div align=\\\"center\\\"][/div]
[div align=\\\"center\\\"]Conclusions 2.0: [/div]
1. Teams should be encouraging more new players to pursue being offensive linemen – particularly offensive tackles – because they get similar performance out of an offensive tackle weighing 340 pounds with 200TPE when compared to the average bot offensive tackle that weighs 310 pounds and has 550TPE.
2. Teams can save on average $4.4 million per year against the cap if these human offensive tackles are paid the current league average for offensive tackles. This would take the average team spend on offensive linemen bots from 25.81% of the cap to between 20.86% and 22.51% of the cap depending on how you account for the teams with human offensive tackles currently.
3. There is not a statistical drop off in performance as it relates to fumbles, total yards per game, rush yards per game, or pass yards per game from a team who has bot offensive tackles weighing 310 pounds with 550TPE to a team who has human offensive tackles weighing 340 pounds with some amount of TPE all the way down to 200TPE.
[div align=\\\"center\\\"]Random Plugs: [/div]
1. This took an immeasurable amount of hours. I've been up until 5am the last few days running the sims.
2. All data can be found here: https://drive.google.com/open?id=1CA67GXPv6...hzPPSoTVGl50gMF
3. If you like this stuff check out my boy Forty's post on the value of home field and intelligence and how those relate to penalties. You can find it here.
4. The reason its 25,199 and not a clean 25,200 like its supposed to be is because the sim glitched a few times from things like program popups on my computer to me accidentally moving the box out of position for the autoclicker. It wasn't worth redoing it over a few issues.
I simulated 25,199 total games so no one else would have to. Several days ago, there was a debate in the NSFL General Discord Chat about the value of offensive lineman. The claim was made that a human offensive tackle at 300TPE was as valuable as a 550TPE bot offensive tackle. I decided to investigate the validity of this claim.
The NSFL salary cap is $80 million per season. The average NSFL team spends 25.81% of their cap space - $20.65 million a season – on offensive lineman bots. Based off data from the TPE and stats databases there are only three active offensive linemen in the league. One – Edmund Beaver – plays for the Austin Copperheads while the other two – Brave Ulysses and Givussafare Rubbe – play for the New Orleans Second Line. Beaver is a 535TPE lineman making $2 million this season. Ulysses is a 727TPE lineman making $3 million this season. Rubbe is a 709TPE lineman making $2 million this season.
All three active linemen make less than the average bot a team purchases – a 550TPE lineman which costs $4.5 million a season. If a team can get the same production out of a 300TPE human offensive tackle and pay them equivalently to what teams are currently paying active human offensive linemen - $2.3 million on average – they can save $2.2 million per lineman and $4.4 million per season compared to the cost of a bot offensive tackle.
[div align=\\\"center\\\"]Methodology:
[/div]
I only focused on offensive tackles because the rules stipulate that a team’s bot offensive tackles can only weigh 310 pounds. A human offensive tackle can weigh 340 pounds. I chose to use the Orange County Otters as my control team because they have the best representation of the average bot offensive line in the league. They currently pay for five bot offensive linemen with 550TPE apiece. Their tackles are 6’7” and 310 pounds – about league average for bot offensive tackles.
I ran two sets of 100 game simulations at seven different TPE/Weight levels against every one of the other nine teams for both home and away. This means that were 200 total games played with the Otters at home at every TPE/Weight combination against every other team in the league and then 200 total games played with the Otters away at every TPE/Weight combination against every other team in the league. The Otters thus played 1400 total games at home against all nine teams and 1400 games away.
The TPE/Weight combinations were divided as such:
[div align=\\\"center\\\"][/div]
The control group is the 550TPE/310lbs group.
[div align=\\\"center\\\"]Results:[/div]
The first simulations pitted the Otters as the home team against every other team in the league. I ran two sets of 100 simulations at each TPE/Weight point to avoid the simulation crashing. The runs with the Otters as the home teams resulted in the following:
[div align=\\\"center\\\"][/div]
When both sets of data were combined the control group won 72.56% of their games at home. As expected, once the weight was increased from 310lbs to 340lbs on the control amount of TPE the win percentage increased by 2.83% to 75.39% total. What is interesting is that all variable TPE/Weight combinations won a greater percentage of their games at home than the control group. The lowest variable combination was 200TPE and 340lbs. This group won 73.15% of their games at home – 0.59% more than the control group.
The Otters could conceivably have two human offensive tackles at the maximum weight of 340 pounds with TPE as low as 200 and still have as high of a win percentage at home as they do with their two 550TPE bots that cost them $4.5 million per season apiece.
These results represented graphically:
[div align=\\\"center\\\"][/div]
The second simulations pitted the Otters as the away team against every other team in the league. Once again, I ran two sets of 100 simulations at each TPE/Weight point to avoid the simulation crashing. The runs with the Otters as the away team resulted in the following:
[div align=\\\"center\\\"][/div]
When both sets of data were combined the control group won 49.58% of their games as the away team. Only one variable combination – 350TPE and 340lbs – was greater than this at a 49.61% winning percentage. Strangely, the greatest gap in winning percentage between the control group and a variable combination was when the variable combination was 550TPE and 340lbs. That group won 46.72% of their games – 2.86% less than the control group. On whole, however, the gaps in winning percentage are not statistically significant.
These results represented graphically:
[div align=\\\"center\\\"][/div]
I then combined the home and away data sets into one combined set of data. This means that each grouping now had 1800 total games in the each run respectively.
[div align=\\\"center\\\"][/div]
The control group had a total win percentage of 61.06%. Of the six variable groups three had higher winning percentage, one had the same winning percentage, and two had lower winning percentage. The gap between the control and the highest was the 250TPE 340lbs group which had a winning percentage of 61.86% - a 0.8% gap. The gap between the control and the lowest was the 200TPE 340lbs group which had a winning percentage of 60.60% - a .46% gap. On whole, all variable groups performed about as well as the control group. There was not a noticeable gap in winning percentage.
These results represented graphically:
[div align=\\\"center\\\"][/div]
[div align=\\\"center\\\"]Conclusions 1.0: [/div]
Teams can get almost equivalent production out of a human offensive tackle weighing 340 pounds with 200TPE as they would a bot offensive tackle weighing 310 pounds with 550TPE. If these human offensive linemen would be paid equivalent to the current rate a human offensive lineman is paid – about $2.3 million – teams could save $4.4 million per season by having human offensive tackles.
It is always feasible these human tackles would be paid less as well. At 200TPE they would be 457TPE less than the current average human offensive lineman. This logic can also be flipped and they could always make more than the $2.3 million current average. As long as they make less than the $4.5 million teams currently pay for the 550TPE bot offensive tackles it is still beneficial for the team.
[div align=\\\"center\\\"]Deeper Dive: [/div]
Since I had data from 25,199 simulated games, I also decided to look at the statistical performance at each variable combination when it came to fumbles and yards per game.
The data collection looks the same for this as it did for the win percentage analysis. Two runs of 100 games at each variable combination for both home and away.
I first looked at fumbles:
[div align=\\\"center\\\"]
[/div]
Although there is a trend of increasing fumbles lost percentage for the variable combinations as TPE decreases in the simulations where the Otters were the away team, it is not particularly statistically significant. The gap only reaches 1.09% at its highest point. When looking at all the runs the percentages balance out again and there doesn’t appear to be a strong trend for the variable combinations having a higher percentage of fumbles lost than the control group.
After looking at fumbles I looked at yards per game to see if the teams performed as well offensively at each variable group.
[div align=\\\"center\\\"][/div]
In general, when looking at total yards per game the variable combinations generally had higher total yards per game, rush yards per game, and pass yards per game at the 550TPE, 450TPE, and 350TPE marks. The yards per game numbers for these categories began to even out to the control group around the 300TPE and 250TPE variable combination groups. On whole, however, there is not a particularly large gap in the performance of any of the variable combination groups when compared to the control group. The team performed very similarly across all simulations with all variable groups.
Graphical representation of total yards per game across groupings:
[div align=\\\"center\\\"][/div]
Graphical representation of rush yards per game across groupings:
[div align=\\\"center\\\"][/div]
Graphical representation of pass yards per game across groupings:
[div align=\\\"center\\\"][/div]
[div align=\\\"center\\\"]Conclusions 2.0: [/div]
1. Teams should be encouraging more new players to pursue being offensive linemen – particularly offensive tackles – because they get similar performance out of an offensive tackle weighing 340 pounds with 200TPE when compared to the average bot offensive tackle that weighs 310 pounds and has 550TPE.
2. Teams can save on average $4.4 million per year against the cap if these human offensive tackles are paid the current league average for offensive tackles. This would take the average team spend on offensive linemen bots from 25.81% of the cap to between 20.86% and 22.51% of the cap depending on how you account for the teams with human offensive tackles currently.
3. There is not a statistical drop off in performance as it relates to fumbles, total yards per game, rush yards per game, or pass yards per game from a team who has bot offensive tackles weighing 310 pounds with 550TPE to a team who has human offensive tackles weighing 340 pounds with some amount of TPE all the way down to 200TPE.
[div align=\\\"center\\\"]Random Plugs: [/div]
1. This took an immeasurable amount of hours. I've been up until 5am the last few days running the sims.
2. All data can be found here: https://drive.google.com/open?id=1CA67GXPv6...hzPPSoTVGl50gMF
3. If you like this stuff check out my boy Forty's post on the value of home field and intelligence and how those relate to penalties. You can find it here.
4. The reason its 25,199 and not a clean 25,200 like its supposed to be is because the sim glitched a few times from things like program popups on my computer to me accidentally moving the box out of position for the autoclicker. It wasn't worth redoing it over a few issues.