Bug fix and improvements for calculateContractDifficulty() #5064
Conversation
Properly uses average BV across all Meks instead of random ones. Corrected ally ratio for House command, best guess magic number for Liaison for now.
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #5064 +/- ##
=========================================
Coverage 10.42% 10.43%
- Complexity 6033 6036 +3
=========================================
Files 953 953
Lines 134106 134121 +15
Branches 19434 19437 +3
=========================================
+ Hits 13986 13990 +4
- Misses 118772 118781 +9
- Partials 1348 1350 +2 ☔ View full report in Codecov by Sentry. |
| rollingCount += weight; | ||
| } | ||
|
|
||
| return totalBattleValue / rollingCount; | ||
| if (campaign.getCampaignOptions().isUseGenericBattleValue()) { | ||
| return ((double) totalBattleValue) / totalGBV; |
There was a problem hiding this comment.
Would you be able to explain why we're dividing BV2 by GBV here? I'm not suggesting it's wrong to do so, but I'm trying to get a sense of what returns we're expecting, so I can factor it into my future work.
There was a problem hiding this comment.
The basic assumption is that difficulty depends on the ratio of player (+allies) BV2 to enemy BV2.
Since FG3 matches GBV it matters less what BV2 the average Mek got, but rather how many points of BV2 it gets per point of GBV.
It's less extreme for the enemies and allies that should be using fairly balanced RATs, but a player achieving an average of ~2500 BV2 per unit entirely with Medium or lighter Clan Meks will face a much weaker OpFor than one achieving the same BV2 average with IS Assault Meks, because the latter gives the OpFor a much larger GBV budget to use.
There was a problem hiding this comment.
I think I can see the logic! :)
To confirm: when we communicate challenge to the player: if I’m displaying half a ‘skull’ per 20% variance between player forces and enemy that will still work with this approach? Or will I need to rethink?
As an illustration, we’re going to be shifting to display skulls as a way of illustrating difficulty. 2.5 skulls would be roughly balanced forces. 2 skulls would have the player force roughly exceed the enemy by about 20%, while 3 skulls would have the enemy power approximately 20% more powerful than the player. And so on, with each half-skull equating to roughly 20% variance.
There was a problem hiding this comment.
Generally yes.
For forces with similar weight distribution and therefore similar GBV, the output of the method shouldn't change other than what the bug fix did.
But I have not run the numbers, so I can't say for sure.
I think I understand what you're doing.
0 skulls = 0% of player BV2 (actually up to 10%)
2.5 skulls = 100% (+-10%)
5 skulls = 200%
That seems like a usable range.
Though with the combination of higher tech and higher skill levels, it's entirely possible for an OpFor to go far beyond 200%. I don't think the difference between 300% and 400% would matter, the player should lose either way, but you could use the reciprocal for the upper half of the scale so it covers all possible BV ratios.
As in 3 skulls mean the player's BV2 is 20% lower than the enemy's and so on, with 4.5 skulls meaning around 500% enemy BV (the player having 80% less) and 5 skulls being the same edge case as 0 skulls, one side having 10 or more times the average BV of the other.
That's a design choice though, more granularity in the range that should actually matter vs being able to differentiate between two extremely bad ideas.
| // TODO estimate the LIAISON ratio by going through each combat lance and | ||
| // getting the actual average (G)BV for an allied heavy/assault mek. | ||
| double allyRatio = switch (getCommandRights()) { | ||
| case INDEPENDENT -> 0; // no allies |
There was a problem hiding this comment.
Now that I can see your code I'm getting a much better idea of what you're intending and I think the use of magic numbers here is perfectly reasonable. While I imagine we probably could derive a 'correct' ratio with a bunch of math, that really doesn't seem worth the effort.
There was a problem hiding this comment.
Yeah, I put that in the same category as including salvage or taking the average across all unit types instead of just meks.
Could we? Yes.
Would it be more accurate? Yes.
Would it actually change the difficulty that is displayed? Probably not.
There was a problem hiding this comment.
100% agree. When you were discussing it on discord I completely misunderstood and thought you were saying that we should use magic numbers instead of determining player power
Properly uses average BV across all Meks instead of random ones.
Corrected ally ratio for House command, best guess magic number for Liaison for now.
This should fix the displayed contract difficulty changing semi-randomly when going back and forth between contracts in the contract market, and generally improve the estimate.
This does not yet estimate the difficulty for Liaison command as accurately as it could, but it's better than what we had before and can be improved later.
@IllianiCBT requested mention