dev diary #78 Home Field Advantage Tweaks

[Rocketcan]

Hello Coaches!

 

Welcome to another Dev Diary for NewSim on SimFBA!

 

In v0.3.0, we introduced Homefield Advantage (HFA) to the simulation engine. At the time, it was purely a function of player discipline: undisciplined players could be affected by over 10 points in attributes (dropping some players 3 or more tiers of skill), while undisciplined players had the potential to not be affected at all.

 

After a few games, this was *seriously* OP, to the point where we took the very drastic step of nerfing it mid-season. This lowered the maximum drop to attributes to 7 points, rather than 10. This helped somewhat, but there were still too many home wins (about 73%). The real-life number is closer to 63%.

 

So we went into this offseason with a high priority attached to redoing HFA. Here is what we came up with.

 

HFA Now

 

The way that HFA works now takes the previous model as a base and expands upon it by taking into account stadium sizes. For now (and likely for the foreseeable future), attendance will not matter. We will assume that all fans of these schools are diehard and will fill the stands, come hell or high water.

 

Stadium size will now cap the amount of drop that a player can see from playing on the road. Small stadiums will still have an effect, but it will be a bit smaller than the largest stadiums. Any stadium that has a capacity less than or equal to 25,000 fans will have the lowest cap (only about a maximum of 2 points drop), whereas any stadium with a capacity greater than or equal to 75,000 fans will have the maximum cap (about 7 points).

 

So here is a step-by-step walkthrough of what happens to a player visiting another team:

• The sim checks against the player's discipline rating to see what the maximum debuff to their attributes will be this game. Higher discipline means a lower maximum debuff, and lower discipline means a maximum higher debuff. Disciplines below a certain threshold have a maximum of 7. Disciplines above a certain threshold have a max debuff of 0 (they are unaffected).
  
• The sim checks against the stadium capacity formula to see what the cap to the debuff is.
  
• If the stadium debuff cap is *lower* than the maximum debuff for the player, the player's debuff is set to the stadium cap. This means that an undisciplined player playing in a less harsh environment might not be affected as much as they would be in a very harsh environment.
  
• Everything up until this point is just to determine the *range* of possible debuffs, *NOT* the actual debuffs themselves.
  
• Once the maximum debuff is established, the sim will roll to see what the debuff is. This roll will be anywhere from zero up to the maximum debuff for the player determined in previous steps. Importantly, this means that just because a player is undisciplined, that doesn't mean they will *always* be bad on the road. They just have the potential to have very bad games on the road, but they could be unaffected if they roll well.
  

Test Runs

 

Now that we have gone over how the system works, let's take a look at how it fares testing.

 

 

As you can see here, last season we had about a 73.33% home win rate. Over the 5 test seasons that I ran, we averaged around 58%. This is even slightly under the target, but I think it is enough of a factor, with realistic origins to be worth it while also having a bit more parity than real life.

 

That is it for this dev diary. I apologize for the length, but I hope it was worth it. Comment with any questions you have and I'll be sure to answer them as best as I can!

 

Until next time,

 

Rocketcan

[Ezaco]

And the people rejoiced.

[Minnow]

How long does it take to compute 5 years of test seasons?

[Rocketcan]

It's about 30 minutes per test season, so about 2.5 hours. Closer to 3 though because I didn't always notice when the season was over to run the data collection script.

[smackemz]

In the future will there be a way to expand some of the smaller stadiums?

[Rocketcan]

Great question. If so, very far future. Part of me says no though. Could be a layer too deep or really skew things too far from IRL.

 

Short answer: not for a long time, if at all.