*New Archetypes Proposal - Wide Receiver

*New Archetypes Proposal - Wide Receiver - Printable Version

+- [DEV] ISFL Forums (http://dev.sim-football.com/forums)
+-- Forum: Community (http://dev.sim-football.com/forums/forumdisplay.php?fid=5)
+--- Forum: Media (http://dev.sim-football.com/forums/forumdisplay.php?fid=37)
+---- Forum: Graded Articles (http://dev.sim-football.com/forums/forumdisplay.php?fid=38)
+---- Thread: *New Archetypes Proposal - Wide Receiver (/showthread.php?tid=26738)

*New Archetypes Proposal - Wide Receiver - Swanty - 10-13-2020

I planned to write a Part 2 of the introduction article here outlining some suggested builds for each position, but it quickly became apparent when testing that this would end up being too much for one article. So instead, I'll focus on one position at a time.

The aim of this study is to come up with 5 archetypes that satisfy the following:

All viable to play as, with similar performance overall but ideally specialised to certain game situations
Each one very clearly unique, with wider attribute ranges
Somewhat representative of the real-life spread of players at that position (though the sim often prevents this)

First off, I'm taking a look at Wide Receivers. In the last article, we looked at the usage statistics which showed that Speed is the dominant choice, with over 60% of all ISFL WRs picking it over the other 3 options. I ran some tests to see how they hold up.

For all the testing below, I simmed as COL

, always playing away and playing against all 13 other ISFL teams equally. The Balanced playbook was the only one used (as it splits roughly equally between the 5 formations) and the pass ratio was turned up a bit to get more significant passing results. I created 3 new identical WRs who were maxed versions of each archetype, and played them in fixed positions across the formations (LWR, RWR, FL).

Here's the results of the existing archetype testing:

Obviously win rate is the ultimate stat that we should care about, but it is extremely volatile and we would need a much higher sample size to get conclusive results. Also, the wide receivers' attributes are a small part of what determines a win. So instead, let's look at Yards Per Attempt. Our effective sample size here is in the tens of thousands of passes attempted and the wide receiver build is a huge part of it, so 742 is more than enough games to make meaningful comparisons. To illustrate that low variance, note how the Home YPA results (which should theoretically be identical) sit in a window of 0.1 yard across the 4 builds.

Looking at the results, it's pretty clear that Speed is the best build, averaging 0.25 YPA more than Route Runner, while the other two lag behind quite significantly.

Let's check out the individual stat lines:

For both LWR/RWR, it seems that it's always better to be a Speed receiver, both in terms of yardage and TDs. They get fewer catches, but that makes sense when they're making more yards per catch and scoring more often. Interestingly, it's a little less clear cut for those playing at FL.

As it currently stands, you could make a case for Red Zone being more useful in a situation where a short catch is needed more than a long completion, but that's not really possible to plan for in the sim. In any case, it's tough to justify choosing anything other than Speed when creating your player.

So onto alternatives. When planning out and testing these, I kept the Speed archetype and tried to come up with 4 alternatives that could keep up with it. It quickly become obvious that this was not going to be possible with any feasible build that had just 90 speed. As such, I went with 2 builds with 95 speed and 2 with 100. This sort of violates #2 in the rule above, but having anything below 95 speed seems to be a death sentence, even with all other attributes maxed.

Before we dive into what makes up each of the proposed archetypes, here are the results (definitely need someone else to name them, creativity is not my strong suit):

The range of YPA for the 5 archetypes is just 0.10 yard which compares favourably to 0.73 from before. Looking at individual stats, the bands are much narrower with different archetypes more suited for different situations. For the 3 x 100 speed builds, Athlete and Speed are pretty much indistinguishable. Speed might be better at FL and Athlete better at LWR, but those would need to be tested per playbook/formation combination. Pocket Rocket is also comparable at most positions but scores touchdowns more often at LWR.

At the 2 x 95 builds, Target Man catches the ball more reliably and is the best FL archetype overall, but Physical is probably the better option for scoring touchdowns at the wider positions. Both of these would also probably be more appropriate as a TE fill-in than the speed ones.

Now onto the actual builds:

SPEED
Max Height: 6'3
Max Weight: 225
(MAX: 65) Strength: 25
(MAX: 90) Agility: 65
(MAX: 80) Intelligence: 40
(MAX: 100) Speed: 70
(MAX: 90) Hands: 55
(MAX: 100) Endurance: 60

This one is unchanged and is the benchmark which we have tried to replicate with the others.

POCKET ROCKET
Max Height: 5'11
Max Weight: 200
(MAX: 70) Strength: 35
(MAX: 90) Agility: 60
(MAX: 95) Intelligence: 55
(MAX: 100) Speed: 70
(MAX: 100) Hands: 60
(MAX: 80) Endurance: 55

These players are small but fast. Their small stature makes them a more difficult target to throw to, so their attribute caps and starting TPE are generally higher to compensate. Endurance can't be raised as high as other positions.

ATHLETE
Max Height: 6'1
Max Weight: 220
(MAX: 70) Strength: 25
(MAX: 100) Agility: 70
(MAX: 75) Intelligence: 30
(MAX: 100) Speed: 70
(MAX: 80) Hands: 55
(MAX: 100) Endurance: 60

Slightly shorter than Speed, these players have higher Agility and Strength at the expense of Hands and Intelligence.

TARGET MAN
Max Height: 6'6
Max Weight: 225
(MAX: 75) Strength: 40
(MAX: 75) Agility: 40
(MAX: 70) Intelligence: 40
(MAX: 95) Speed: 50
(MAX: 100) Hands: 55
(MAX: 100) Endurance: 60

The Target Man is an attractive option for a QB. Their natural height advantage means they come down with the ball more often than their peers. This built-in bonus means their starting TPE and caps are the lowest of all WRs.

PHYSICAL
Max Height: 6'4
Max Weight: 235
(MAX: 85) Strength: 50
(MAX: 90) Agility: 55
(MAX: 90) Intelligence: 55
(MAX: 95) Speed: 55
(MAX: 95) Hands: 50
(MAX: 80) Endurance: 50

The Physical WRs have a nice weight advantage to help them win the 1 on 1s, but the 95 speed and the lack of height compared to Target Man means they need a strong boost across the rest of the attributes to compete with everyone else.

Comparing the variety of wide receivers that we'll end up with in terms of attributes, we've definitely improved the uniqueness and variety available through archetypes:

In conclusion, speed is king and there's not really anything we can do with low speed WRs to make them competitive. But by placing all at 95/100 speed, we've come up with 4 new archetypes that should compete equally overall with Speed receiver. If these archetypes were to be implemented, it would certainly shift the run/pass ratio more towards the latter, increasing passing yards across the league. Though with 60% of the league currently at Speed, it wouldn't be a huge change.

Interested to hear your thoughts!

RE: New Archetypes Proposal - Wide Receiver - Modern_Duke - 10-13-2020

Sorry if you mentioned it and I missed it, but are they all relatively equal in terms of tpe needed to max?

RE: New Archetypes Proposal - Wide Receiver - Swanty - 10-13-2020

Ah sorry I didn't mention. This is how much TPE it takes to max (ignoring the irrelevant stats like blocking etc):

Physical: 1110
Speed & Target Man: 1115
Athlete: 1150
Pocker Rocket: 1180

Original archetypes for comparison:
Route Runner: 1100
Speed: 1115
Redzone: 1140
Possession: 1145

RE: New Archetypes Proposal - Wide Receiver - siddhus - 10-13-2020

Very nice and interesting work. Interested to see what you come up with the other positions

RE: New Archetypes Proposal - Wide Receiver - ReverendOReily - 10-13-2020

"Pocket rocket" is an incredible name, don't let anyone tell you otherwise

RE: New Archetypes Proposal - Wide Receiver - slate - 10-13-2020

(10-13-2020, 08:39 AM)Swanty Wrote: Obviously win rate is the ultimate stat that we should care about, but it is extremely volatile and we would need a much higher sample size to get conclusive results.

If only there were an open source tool you could use to generate really large sample sizes Wink

Great stuff Swanty! Really love the diversity of playstyles you've managed to achieve here while obeying the 95+ speed constraint and improving balance.

RE: New Archetypes Proposal - Wide Receiver - timeconsumer - 10-13-2020

Thoughts:

I've never seen height or weight have an effect on WR performance in this sim so you might want to vet that.

85 strength and 95 speed might result in a need to redo TE archetypes as at that point you really make them weaker, especially something like the vertical TE.

LWR vs RWR vs Slot has a lot to do with the playbooks. They each have a different amount of targets by position etc. Balanced targets the LWR the most excluding the two-TE formation where it targets RWR more, and in spread Slot gets the most targets. But this target % is highly variable based upon the number of receptions and the hands stat of the receiver.

We need to be careful using efficiency metrics as the yardstick here. While it's an easy way to look at things we need to remember that efficiency by design is going to change based upon target volume. More receptions, speed drops. More yards, speed drops. Pretty soon your player is crawling along the field by catch 9. A player that is designed to catch more balls and get more first downs (red zone etc) is going to be less efficient, that doesn't make them worse. A player that is designed to drop more passes is going to be more efficient, even with the drops, that's fine, it doesn't make them better necessarily.

Edit: Me again. The sample size is really confusing me. I'm trying to figure out overall win % (as thats more important than home vs away imo) and if your methodology is that you played the other 13 teams an equal amount of times but I'm coming up with 57.08 games per team? And why is the sample size twice the size for your reworked archetypes, which this is 114.15 games per team? As it stands, control win rate was exactly 50% for 3 archetypes and 50.9% for route runner.

RE: New Archetypes Proposal - Wide Receiver - Modern_Duke - 10-13-2020

Something tc said made me relook at this. The purpose of this was to provide more different options, but based on the results you’ve made 5 of the exact same receiver. Like at lwr, 5 catches, 98 yards, 0.7 td no matter which archetype you choose. Shouldn’t the idea be to have one option that catches more balls but at a lower average, one that catches fewer balls but at a higher average, etc?

RE: *New Archetypes Proposal - Wide Receiver - Swanty - 10-14-2020

(10-13-2020, 03:13 PM)Modern_Duke Wrote: Something tc said made me relook at this. The purpose of this was to provide more different options, but based on the results you’ve made 5 of the exact same receiver. Like at lwr, 5 catches, 98 yards, 0.7 td no matter which archetype you choose. Shouldn’t the idea be to have one option that catches more balls but at a lower average, one that catches fewer balls but at a higher average, etc?

The purpose of this was to provide variety of options indeed, but options that can be all be considered viable for a new player to choose. They're not exactly the same as you can see from the last image. There is far more variety in builds, they just average out to perform more equally. Our current archetypes are more similar in terms of build but wildly different in terms of output.

Of course it would be ideal to come up with archetypes which are not only all viable, but which perform better in different situations. Right now, we have 3 archetypes which perform almost identically in terms of completion percentage (Poss, RR, Speed), but Speed completely outclasses the other two in terms of yards per catch. So there's no real game situation that we can control through strategy where the former two are better, or even equal, to Speed.

In trying to balance them, we do see completion rate differences. Target Man then Physical have the best 2, then Speed and Athlete are equal, then Pocket Rocket has the worst. Their yards per catches are more even (17.3-18.7) than the 3 current archetypes, but that's necessary to make them all options.

As TC alluded to above, this study is very broad, using only the Balanced playbook to achieve a much more equal, average performance than before. And because the end results are so close together, it's likely that each playbook/formation combination is going to have one given archetype perform the best, while that archetype will need to be worse in another situation to make it balance.

No doubt this will get tested to death if released, and teams might realise that Athlete works best at LWR for Spread but is terrible at RWR on Power for example, but I've got no intention of trying to define the metagame for teams. This was intended to provide 5 archetypes that could legitimately be selected by a new player and allow them to be competitive, minimising the risk of being asked a few seasons in to switch over to Speed and ruin the immersion they've created with their player.

(10-13-2020, 02:35 PM)timeconsumer Wrote: Thoughts:

I've never seen height or weight have an effect on WR performance in this sim so you might want to vet that.

It does. WRs would be wise to max theirs out!

(10-13-2020, 02:35 PM)timeconsumer Wrote: 85 strength and 95 speed might result in a need to redo TE archetypes as at that point you really make them weaker, especially something like the vertical TE.

This makes sense, especially if people start playing the Physical archetype at TE. Would likely need to be tested to death alongside archetype changes for all other positions.

(10-13-2020, 02:35 PM)timeconsumer Wrote: LWR vs RWR vs Slot has a lot to do with the playbooks. They each have a different amount of targets by position etc. Balanced targets the LWR the most excluding the two-TE formation where it targets RWR more, and in spread Slot gets the most targets. But this target % is highly variable based upon the number of receptions and the hands stat of the receiver.

Yep, agreed. No doubt even though the average works out even, some of these archetypes will be better in certain playbooks. But it's nice to know that all archetypes will have certain playbook/position combinations out there that work for them. That's for the teams to figure out in my opinion.

(10-13-2020, 02:35 PM)timeconsumer Wrote: We need to be careful using efficiency metrics as the yardstick here. While it's an easy way to look at things we need to remember that efficiency by design is going to change based upon target volume. More receptions, speed drops. More yards, speed drops. Pretty soon your player is crawling along the field by catch 9. A player that is designed to catch more balls and get more first downs (red zone etc) is going to be less efficient, that doesn't make them worse. A player that is designed to drop more passes is going to be more efficient, even with the drops, that's fine, it doesn't make them better necessarily.

That makes sense, and might merit testing again with different tempos to see if one archetype hits that cap a lot earlier than others.

(10-13-2020, 02:35 PM)timeconsumer Wrote: Edit: Me again. The sample size is really confusing me. I'm trying to figure out overall win % (as thats more important than home vs away imo) and if your methodology is that you played the other 13 teams an equal amount of times but I'm coming up with 57.08 games per team? And why is the sample size twice the size for your reworked archetypes, which this is 114.15 games per team? As it stands, control win rate was exactly 50% for 3 archetypes and 50.9% for route runner.

What do you mean by overall win %? In this study, I was the away team the whole time as I believe that takes into account more attributes like Endurance, as when playing at home it doesn't really play a part. It also allowed me to keep the win% around 50% as that allows for more visible variance in win %. And yes the sample size number wasn't exactly spot on. I think it was 57 for some teams and maybe 58 for one team.

RE: *New Archetypes Proposal - Wide Receiver - timeconsumer - 10-14-2020

(10-14-2020, 02:03 AM)Swanty Wrote: What do you mean by overall win %? In this study, I was the away team the whole time as I believe that takes into account more attributes like Endurance, as when playing at home it doesn't really play a part. It also allowed me to keep the win% around 50% as that allows for more visible variance in win %. And yes the sample size number wasn't exactly spot on. I think it was 57 for some teams and maybe 58 for one team.

That part wasn't very clear that you were always the away team. I'd say you'd be best served by testing for both home and away as we want to see how things match up to a "real-world" scenario. We play both home and away games during the season so why eliminate one of those on a hunch that it will be more accurate? Also I feel like looking at sim results between two different things they should both have the same sample size or at least very similar, not a difference of 100%.