The
Station Balance - Methodical Analysis forum post by Dark.Revenant has provided useful insights into the power of stations, and it has no doubt shaped discussion about Starsector's stations for years. However, as the analysis was originally made during Starsector 0.9.1a, it is unlikely that the results still hold up as of the current 0.97a-RC11 version, and the use of modded ships, weapons, and fighters necessarily means the analysis' relevance for discussing vanilla balance is limited at best (which makes sense, given the original motivation of Dark.Revenant's analysis was to inform the creation of their own modded stations).
Therefore, I have replicated this analysis for Starsector 0.97a, this time using only vanilla content (i.e., no modded ships, weapons, and fighters). Since orbital station discussion tends to center around which station would be the most powerful in general, and to reduce the time needed to make this
rather-preliminary analysis, only the highest station tiers (which includes the fully-operational Remnant station) will be tested here.
(I will also be replicating the overarching content, style, and format of Dark.Revenant's original analysis, because I think it is neat and, thus, is worth mimicking.)
Test Method- Use a battle size of 400, and all other gameplay settings, like max ships per fleet, are left at vanilla defaults.
- Use the Adversary mod (v6.3.1 at time of writing) containing the faction against which the stations will be tested:
- (Full disclosure: I am the author of this mod. I am also the author of the A Fleet Testing Mod mod (v0.7.1 at time of writing) containing the station tester mission used for this analysis. The mission itself was derived from Dark.Revenant's Interstellar Imperium station tester mission.)
- The main reason is that the Adversary faction includes nearly all vanilla human ships (barring the XIV and LG-only ship variants as well as automated ships) and most vanilla weapons and fighters (excluding LG-only and Omega weapons as well as Remnant fighters), while also not including any modded ships or weapons that would hurt this analysis' relevance to vanilla playthroughs. The closest vanilla faction equivalent would probably be the independent "mercenary" faction, which does a similar thing with having many human ships and weapons, except they still lack several of the human ships (no Pirate and LP variants, no Executor, no Invictus, etc.; this is main reason I did not use them as the benchmark faction), and they also favor phase ships less due to their 4/2/1 ship type split.
- It is worth pointing out that the Adversary's doctrine (which will be described later) is arguably tailored towards station-busting due to prioritizing max ship size and aggression (by contrast, the "everything" faction from the SWP mod only has Ship Size 3 and Aggression 3).
- (I could have created a separate faction solely for this analysis, but I did not wish to spend extra time to make this analysis, and the Adversary faction is about 90% of what I was looking for as a generalized benchmark anyways. Besides, a future, more thorough analysis would ideally look across multiple vanilla factions instead to account for differences in faction doctrine.)
- Spawn a station at 100% CR, with no officer and no autofit. Only the AI has full control of the station.
- The Adversary faction has access to ships, weapons, fighters, and hullmods from all human factions, except for XIV and LG variants. No ships, weapons, or fighters are prioritized, so these will be picked according to default vanilla weights. The faction uses the following doctrine:
- Warships 3 / Carriers 2 / Phase Ships 2
- Officers 3 / Ship Quality 2 / Fleet Size 2 (none of these actually apply here, as the enemy fleet always spawns with a set quality and FP as well as no officers)
- Ship Size 5
- Aggression 5 (i.e., all ships are Reckless)
- Combat Freighter Combat Use Fraction 0
- Autofit Randomize Probability 0
- Select a fleet size, in Fleet Points (FP) according to the following logic:
- First test - start at 320 FP (chosen as it represents the average FP of a Remnant Ordo at max strength, and is also close enough to the original Star Fortress FP results from Dark.Revenant's 0.9.1a analysis.)
- Otherwise - Adjust by a constant number of FP (which is 5 here), going up if the station won and going down if the station lost.
- Spawn a 70% CR Adversary fleet using the doctrine and size described above, with autofit enabled, no civilian ships (e.g., no Drams, Buffalos, etc.), no officers, and 120% quality (resulting in better weapons and no random D-mods) for the enemy side. The enemy starts with the standard 5 command points upon battle start.
- If the fleet's FP does not match the target FP, repeat the previous step up to 1000 times; the fleet with either the same or closest FP to the target FP is chosen.
- Generates a static, no-terrain battle scenario that replicates a campaign station battle around a planet:
- Non-hyperspace
- No asteroids
- Battle map size of 18000x18000 units2 (default size of battle map with no objectives)
- Fixed standoff range of 6000 units
- The enemy is not allowed to retreat, must fight to the last, and must deploy all ships.
- Repeat the test until after a minimum of 10 bouts and a win ratio of 50% is reached between the last 10 bouts (i.e., 5 wins and 5 losses in the last 10 bouts).
- The final score is the sum of the FP of the last 9 bouts plus the FP of what would have been the next bout, divided by 10 and rounded to the nearest whole number.
This is what the average Adversary fleet generated in the mission looks like, if you are curious:
Link to the image showing the stations being tested (huge image, which is why it is not embedded here):
https://i.imgur.com/93zJrau.jpegTo speed up the testing, an in-combat plugin automatically speeds up the game such that the internal game logic is running at 1/30 second intervals at the fastest possible speed, limited by the refresh rate and current FPS (going any faster would effectively destabilize the game's combat engine and make the results much less reliable).
ResultsBout Results (for reference)
These show the last 10 bouts leading to a 50% win ratio for each station tested. An asterisk (*) indicates extra bouts that replace earlier recorded bouts; for example, 3 *'s indicates 3 extra bouts were played beyond the initial 10 bouts.
Low Tech Star Fortress (11 bouts played):
*1. 310 (W)
2. 315 (W)
3. 320 (L)
4. 315 (L)
5. 310 (W)
6. 315 (L)
7. 310 (W)
8. 315 (L)
9. 310 (W)
10. 315 (L)
Average FP: (315+310+315+310+315+310+315+310+315+320) / 10 ~= 314 FP
Midline Star Fortress (34 bouts played):
**1. 290 (L)
**2. 285 (W)
**3. 290 (L)
**4. 285 (W)
*5. 290 (L)
*6. 285 (W)
*7. 290 (W)
*8. 295 (L)
*9. 290 (L)
*10. 285 (W)
Average FP: (290+285+290+285+290+285+290+295+290+285) / 10 ~= 289 FP
High Tech Star Fortress (13 bouts played):
*1. 330 (W)
*2. 335 (W)
*3. 340 (L)
4. 335 (L)
5. 330 (W)
6. 335 (L)
7. 330 (W)
8. 335 (W)
9. 340 (L)
10. 335 (L)
Average FP: (335+340+335+330+335+340+335+330+335+330) / 10 ~= 335 FP
Remnant Station (11 bouts played):
*1. 330 (L)
2. 325 (W)
3. 330 (W)
4. 335 (W)
5. 340 (L)
6. 335 (W)
7. 340 (W)
8. 345 (L)
9. 340 (L)
10. 335 (L)
Average FP: (325+330+335+340+345+340+335+340+335+330) / 10 ~= 336 FP
Low Tech Star Fortress:
314 FPMidline Star Fortress:
289 FPHigh Tech Star Fotress:
335 FPRemnant Station:
336 FPAnalysisCompared to the results from the 0.9.1a analysis, the FP results of all tested stations are noticeably lower, especially for the Remnant Station (336 FP here vs the 370 FP before). In addition, the gap between the buildable stations' FP results is much higher, with the Midline station having the lowest at 289 FP and the High Tech station having the greatest at 335 FP.
Besides random variance and luck (those are always a given), some potential explanations for the differences in FP results compared to the 0.9.1a analysis are:
- Higher ratio of capitals and cruisers in the test fleets. Capitals and cruisers, by virtue of having higher flux capacity, armor, and hull integrity than the smaller ship classes, are more likely to survive getting in range of the station and, therefore, spend more time actually dealing damage to the station. The Reckless personality imposed by the Adversary's doctrine also means the ships will close in on the station for much longer, though this also means carriers, frigates, and destroyers are more likely to die needlessly.
- Less ship/weapon/fighter choices due to a lack of modded content, leading to increased likehood of ships with higher missile counts and use of high-damage strike missiles or bombers. Most notably, test fleets packing the most HE strike missile firepower (e.g, Reapers, Hammers, Hurricanes, Cobra bombers) often did the best against the stations, usually by blowing up important modules earlier in the fight. Conversely, fleets that lacked enough missile firepower tended to be defeated with ease or barely scored a victory against the station.
- New content additions and ship/weapon rebalances since 0.9.1a. For example, the Pegasus and the Invictus generally performed well against the stations due to heavy missile spam and Lidar Array barrages, respectively. Plus, general buffs to ships like the Onslaught, Mora, and Legion, as well as improved ship AI, likely contributed to stations faring a little worse. Finally, while weapons buffs technically do improve station effectiveness (particularly for the Midline's Storm Needler and Hephaestus Assault Gun), these buffs generally favor the ships more, especially with the changes to the Heavy Autocannon, Pulse Laser, Cyclone Reaper Launcher, and more.
Regarding the clear FP disparity between the buildable stations, the High Tech station seemed to win out
despite the default station loadout being surprisingly undergunned (e.g., using Ion Pulsers in two of the available Large slots, and generally emphasizing EMP damage over raw damage). Based on observations, the Mine Strike systems (targeting vulnerable ships anywhere in the map), Gargoyle drones (Paladin PD protecting against missiles, and Fortress Shield generally keeping the drones alive much longer than the Low Tech or Midline drones), and separate shielding (protecting the station with Fortress Shield and allowing the main modules to fire regardless of shield status) all appear to contribute to the High Tech station's apparent dominance over the Low Tech and Midline stations.
On the other hand, the Midline station appeared to struggle against the test fleets, resulting in it having the lowest FP result. From what was observed, the likelihood that the Midline station wins
largely depends on how often the main module is able to freely vent. At low flux, the main module's Large weapon batteries tended to wreck targeted ships, similar to an Invictus's Lidar Array barrage. At high flux, however, the main module only fired its Storm Needler and Graviton Beams, usually preventing it from also firing its Hephaestus Assault Guns and Mjolnir Cannons. Because the enemy fleet will naturally surround the station, the main module ended up staying at high flux levels in most losing battles due to it continuing to fire its kinetic weapons at any and all ships and fighters--which usually failed to finish them off.
Meanwhile, the Low Tech station scored in the middle, between the two other buildable stations. Like what was noted in the 0.9.1a analysis, enemy ships sometimes insisted on firing through the invincible structural spurs, wasting precious ammo and building up soft flux; this behavior likely contributed to few of the station's wins. Other than this
errant behavior still appearing on the 0.97a version, there is not much else to say about the Low Tech station.
Lastly, the Remnant station performed relatively poorly compared to its 0.9.1a analysis result, only besting the High Tech station's FP result by just 1 FP. While the 360 degree constant long-range firepower certainly favored the station against the test fleets, the subpar shields and armor modules also meant it was more vulnerable to heavy strike damage blowing up many of its modules early in the battle, which largely explained most of its losses.
Final ThoughtsWhile I can say the results from this station analysis reveal a lot about how stations are faring in the current 0.97a version, I should point out that this analysis
does not account for other faction doctrines, nor does it account for a (player) fleet supporting the station against the enemy fleets. For example, the Midline station, and the High Tech station to a lesser extent, would likely perform much better when a friendly fleet provides it cover from flanking enemy ships, amplifying its ability to provide supporting firepower. In addition, actual enemy fleets
have officers as well as active admiral skills, not to mention an entirely different fleet doctrine that either helps or hinders them against stations in general. Also, we should not forget about the stations themselves having vastly different loadouts due to faction weapon availability and autofit--anyone who fought against the Persean League's Midline Star Fortress should know that well.
Despite all of what I just said, the results do appear to support the notion of certain stations being better than others, following the High Tech > Low Tech > Midline sentiment that I have often noticed in community discussions both in the forums and outside of it. Whether these results are a cause for concern or simply a random coincidence that can be brushed off as such remains to be seen; a more thorough analysis, with a more robust and replicable methodology that actually tests against the actual vanilla factions, would probably need to be done to say anything more conclusive.
Special Thanks- Dark.Revenant - For the original station analysis and methodology! Also for the II station tester code that I uncerimoniously cribbed to help create my own station tester mission (hopefully you don't mind!).
- Himemiko - For the Java 23/24 conversion kit to vastly improve Starsector performance! Without this, running the tests like would've taken too long for me to actually consider replicating the original station analysis.
Author
Topic: [0.97a] Station Balance - A Replicated Analysis (Read 2366 times)