• Log InLog In
  • Register
Liquid`
Team Liquid Liquipedia
EDT 00:55
CEST 06:55
KST 13:55
  • Home
  • Forum
  • Calendar
  • Streams
  • Liquipedia
  • Features
  • Store
  • EPT
  • TL+
  • StarCraft 2
  • Brood War
  • Smash
  • Heroes
  • Counter-Strike
  • Overwatch
  • Liquibet
  • Fantasy StarCraft
  • TLPD
  • StarCraft 2
  • Brood War
  • Blogs
Forum Sidebar
Events/Features
News
Featured News
Code S Season 1 - RO12 Group A: Rogue, Percival, Solar, Zoun12[ASL21] Ro8 Preview Pt1: Inheritors16[ASL21] Ro16 Preview Pt2: All Star10Team Liquid Map Contest #22 - The Finalists22[ASL21] Ro16 Preview Pt1: Fresh Flow9
Community News
RSL Revival: Season 5 - Qualifiers and Main Event4Code S Season 1 (2026) - RO12 Results02026 GSL Season 1 Qualifiers25Maestros of the Game 2 announced92026 GSL Tour plans announced15
StarCraft 2
General
Code S Season 1 (2026) - RO12 Results Code S Season 1 - RO12 Group A: Rogue, Percival, Solar, Zoun Team Liquid Map Contest #22 - The Finalists Blizzard Classic Cup @ BlizzCon 2026 - $100k prize pool MaNa leaves Team Liquid
Tourneys
RSL Revival: Season 5 - Qualifiers and Main Event GSL Code S Season 1 (2026) SC2 INu's Battles#15 <BO.9 2Matches> WardiTV Spring Cup SEL Masters #6 - Solar vs Classic (SC: Evo)
Strategy
Custom Maps
[D]RTS in all its shapes and glory <3 [A] Nemrods 1/4 players [M] (2) Frigid Storage
External Content
The PondCast: SC2 News & Results Mutation # 523 Firewall Mutation # 522 Flip My Base Mutation # 521 Memorable Boss
Brood War
General
Pros React To: Leta vs Tulbo (ASL S21, Ro.8) [BSL22] RO16 Group A - Sunday 21:00 CEST [BSL22] RO16 Group B - Saturday 21:00 CEST RepMastered™: replay sharing and analyzer site BW General Discussion
Tourneys
[BSL22] RO16 Group Stage - 02 - 10 May Escore Tournament StarCraft Season 2 [Megathread] Daily Proleagues [ASL21] Ro8 Day 2
Strategy
Fighting Spirit mining rates Simple Questions, Simple Answers What's the deal with APM & what's its true value Any training maps people recommend?
Other Games
General Games
Daigo vs Menard Best of 10 Stormgate/Frost Giant Megathread Nintendo Switch Thread Dawn of War IV Diablo IV
Dota 2
The Story of Wings Gaming
League of Legends
G2 just beat GenG in First stand
Heroes of the Storm
Simple Questions, Simple Answers Heroes of the Storm 2.0
Hearthstone
Deck construction bug Heroes of StarCraft mini-set
TL Mafia
Vanilla Mini Mafia Mafia Game Mode Feedback/Ideas TL Mafia Community Thread Five o'clock TL Mafia
Community
General
US Politics Mega-thread European Politico-economics QA Mega-thread Russo-Ukrainian War Thread 3D technology/software discussion Canadian Politics Mega-thread
Fan Clubs
The IdrA Fan Club
Media & Entertainment
[Manga] One Piece Anime Discussion Thread [Req][Books] Good Fantasy/SciFi books Movie Discussion!
Sports
2024 - 2026 Football Thread McBoner: A hockey love story Formula 1 Discussion
World Cup 2022
Tech Support
streaming software Strange computer issues (software) [G] How to Block Livestream Ads
TL Community
The Automated Ban List
Blogs
Sexual Health Of Gamers
TrAiDoS
lurker extra damage testi…
StaticNine
Broowar part 2
qwaykee
Funny Nicknames
LUCKY_NOOB
Iranian anarchists: organize…
XenOsky
Customize Sidebar...

Website Feedback

Closed Threads



Active: 1615 users

[Dota 2] The Winner's advantage

Blogs > etofok
Post a Reply
1 2 3 4 Next All
etofok
Profile Blog Joined March 2012
138 Posts
Last Edited: 2015-04-09 16:47:03
April 08 2015 21:40 GMT
#1
I'll start from far away to give you a better perspective.



Here is the imaginary scrollbar that any developer at pretty much any strategy game have to somehow balance:

[image loading]


If you allow early snowballs -> fast deathballs, the late-game part of your game becomes irrelevant, which is not what you want. And vice versa.

For a long time, IceFrog's had the intention to speed up the game - which he still does with a great success, unfortunately it led to this:

[image loading]
State of the game at TI4



And shortly after TI4 IceFrog adjusted the bar. Slightly.
[image loading]


I mean he had to: if you remember how huge was the outcry of the community about early push starts a.k.a. “Pick early game heroes and roll through your opponents” it can persuade any person that something is indeed wrong with the game.


So here it comes: 6.82 - the patch where you get a radiance on Spectre at 40 mins and win the game become you won a skirmish that made 25k gold swing. That was adjusted rather quickly because the issue was as apparent as mud on one’s brand new white shoes.

[image loading]

in other words
[image loading]



The Comeback Mechanic is aimed to provide to the losing side a much better chance to catch-up, providing a "clutch" game as the result. Along with other changes like glyph replenishment and flat-out nerfs, it prevents snowballiness (to an extent) meaning you play a more "even" match throughout the game.


After several tweaks of TCM everyone finally became quite content with the results of it: after all the logic behind it is actually okay and makes total sense. It hasn't been apparent how the change might still negatively affect the game after all those number tweaks, however today we can easily see the issue(s) that emerged with the advent of such mechanic: the ability to endlessly defend a high-ground because you can’t get as ahead as it is required to being able to break the high-ground and to finally make usage of your advantage.


[image loading]
representation of my thesis


High-ground is notably hard to push onto - that’s the defender’s advantage right there. It is kind of supposed to be like that, because this is the barracks or even the throne we are talking about.
The advantage is actually so drastic it might sometimes be bigger than, let’s say for our imaginary situation - 30k gold. Due to several reasons it is almost impossible to exceed beyond a certain amount of “advantage” as the winner side and thus that advantage is practically negated as you ascend onto enemy’s HG.


[image loading]
well memed my friend


In terms of difficulty the fight will be “equal” meaning you are supposed to trade “evenly”.
However when you trade 5 heroes for 5 heroes, the defender gets way more gold out of it due to rubber band, and pushing into the defender becomes nearly impossible, therefore teams avoid this situation in the first place by farming instead and not actually pushing / trying to end the game because the risk is not worth the effort.

[image loading]
Sometimes you can't push even if you want. The real flaw is when your enemies can't push either.

This is not because of The Comeback Mechanic alone.
The problem is that the losing side still gets "enough" gold and experience from lane creeps pushing into them to be able continue their item progression, while abandoning all the map control whatsoever.


[image loading]
when map control look like this the game is close to be done



This is the most important part of a strategy game - map control. In Starcraft when a player gets contained he is going to lose the game eventually, because he’ll run out of resources. If you've ever played TvT you should know what I mean. Check out this wiki page to read some more if you aren't familiar with the concept.

However because in Dota there are lane creeps that give gold every 30 seconds regardless where they are killed - be it at their spawn point or at your barracks line - this situation is weird to say the least: there can’t be enough contain to actually contain.


After a certain point the winning side’s cores reach their maximum potential - farming don’t make them stronger anymore, whilst enemy still has a room to grow. Therefore the “real” Networth of the winning team stops being relevant to the items they have, because that gold stops being put into items.

With time, that gap between teams only shrinks even if the gold/exp discrepancy increases on the graphs. Add to this the fact that the defender’s advantage is still there AND the risk of a fail push skyrockets due to potentially disastrous rubberband gold/exp transfers.


[image loading]
I'm honestly so proud of these pictures


It all comes down to map control, or rather how pointless it might be in some specific situations.


Let’s list what does having full map control provide:
1) Roshan - aegis, cheese.
2) Access to both jungles and ancients.
3) Runes.

If I were to suggest something in hopes to avoid the main problem [that is giving up on map control and still getting resources]:


1) Jungle creeps give more gold with time / with Roshan kills / as you sacrifice your aegis or cheese to the shopkeeper / whatever reason.

Point is, with the entire map control should come bigger dominance than it is right now to exceed “the gap” [that consists of the risk of rubber band + defender’s advantage + actual networth difference] allowing your support heroes to acquire late-game items for themselves which ultimately will lead to victory faster than 2 hours of resets. “Useless gold” is no longer present, but being transferred straight into support slots such as hex, bkb, halberd, diffusal(!) or even damage items.


Update: The side with map control is going to farm both the jungles plus Roshan and thus farming 3-4 times as fast items that actually matter. Lane Creeps do scale, while Jungle Creeps do not, meaning support heroes are going to have much easier time farming them.

On average there are 533 gold in one Jungle (wiki), so ideally ~1100 gold at the moment. However the amount of Lane Creeps is increasing with time and thus the amount of gold they give away. Therefore Jungles are worth less in comparison = controlling jungles worth less as the game progresses. I want to put higher emphasis on jungle creeps: that way in a 2 hour game the winner side with complete map control will get more gold out of the map so "the gap" will be increasing instead of shrinking.



2) Sellable cheese with increasing cost similar to bounty runes.

Same idea basically - to get additional items for the winning team to increase “the gap” and to finally break their defenses. I mean something really ridiculous like 4k for a cheese at 80 minutes might be actually a great thing for the game.


The Roshan, being a contesting point, gets way higher emphasis in the late-game where aegis and cheese value is usually goes down because your cores run out of slots.

The losing team knowing that will be drawn out of their base to contest this. As you might already understood - the change should lead to exactly this if you want to address the problem I'm stating.


Update - F.A.Q.

1) Make Aegis a buff/aura like in LoL
I had the same idea, but disregarded it almost instantaneously because I wanted to make support heroes stronger and more fun to play in such circumstances rather than buffing a 6 slotted Medusa.

2) Rubber band has to go!
Keep in mind, the rubber band doesn't come into play if nothing happens or the winning team finally wins

3) Lane creeps dying in proximity of their HG are worth less gold
This change essentially destroys their barracks once you get rid of tier 2s which can be done relatively early in the game. That said you remove half of the objectives from the game to contest for: Tier 3s, Tier 4s, 3 lanes of barracks and throne along with negating all the defender's advantage (you just don't push anymore) and making the game ridiculously boring to play.

4) Buyback to resurrect with 50% hp / mana
It also affects all the stages of the game and several heroes such as Spectre and Furion, while doesn't really address the problem of not pushing / pushing, only changing the dynamic of a fight. I would prefer asking a pro about this one.

5) Remove Cooldown on buyback
This encourages core heroes to endlessly farm while supports of both teams stay afk


Summary: If a team abandons all map control whatsoever for a long period of time their disadvantage should be increasing.


Thanks for reading.
[Dota 2] The Winner's Advantage by etofok

Correct my grammar and suggest better wording if you have time.
You can read some other my posts there if you're interested.
The king, the priest, the rich man—who lives and who dies? Who will the swordsman obey?
MrCon
Profile Blog Joined August 2010
France29748 Posts
April 08 2015 21:59 GMT
#2
Nothing much to say except : very good
TechSc2
Profile Blog Joined September 2010
Netherlands554 Posts
April 08 2015 22:20 GMT
#3
First of all i would say this is a great write-up

Second of all ,i completely disagree with your solution. Your solution would only make the game go longer, since every single gold on the defenders side is worth much more then on the attackers side, so you need to increase every single jungle creep + ancients every single minute as much as that difference in actual gold worth.

I think a more simple solution would be in order, to not overcomplicate things, such simple things like decreasing T3 tower hp, make highground sieging items a little bit cheaper ( heart, vangaurd, pipe ) and simply nerfing the highground specialists ability that makes them the HG specialist ( snipers schrapnel slow, headshot slow ).

The current meta is one of physical damage, and to stay alive you need to do alot of damage and have lifesteal, so buying more then 1 defensive item actually makes your survivability worse then another damage item
Twitch.tv/TechGTV / Twitter.com/TechGTV
ahswtini
Profile Blog Joined June 2008
Northern Ireland22212 Posts
April 08 2015 22:20 GMT
#4
oh shit graphs
"As I've said, balance isn't about strategies or counters, it's about probability and statistics." - paralleluniverse
RoyGBiv_13
Profile Blog Joined August 2010
United States1275 Posts
April 08 2015 22:48 GMT
#5
5/5 graphs and sniper hate
Any sufficiently advanced technology is indistinguishable from magic
Tephus
Profile Joined May 2011
Cascadia1754 Posts
April 08 2015 22:55 GMT
#6
You need a lesson on graphs from parallel universe. These make too much sense.
AdministratorTeam Liquid VP of Esports
etofok
Profile Blog Joined March 2012
138 Posts
Last Edited: 2015-04-08 23:15:09
April 08 2015 22:58 GMT
#7
Your solution would only make the game go longer, since every single gold on the defenders side is worth much more then on the attackers side, so you need to increase every single jungle creep + ancients every single minute as much as that difference in actual gold worth.


Update: The side with map control is going to farm both the jungles plus Roshan and thus farming 3-4 times as fast items that actually matter. Lane creeps do scale, while Jungle Creeps do not, meaning support heroes are going to have much easier time farming them.

On average there are 533 gold in one Jungle (wiki), so ~1100 gold. However the amount of Lane Creeps is increasing with time and thus the amount of gold they give away. Therefore Jungles are worth less in comparison = controlling jungles worth less. I want to put higher emphasis on jungle creeps: that way in a 2 hour game the winner side with a complete map control will get more gold out of the map so "the gap" will be increasing instead of shrinking.

I'll put this as update, ty.
The king, the priest, the rich man—who lives and who dies? Who will the swordsman obey?
mnck
Profile Joined April 2010
Denmark1518 Posts
April 09 2015 00:15 GMT
#8
I love the idea of being able to sell the cheese for a large amount of gold lategame. I have also thought that killing rosh loses meaning late and that it should be worked to reduce long games. Something like making Aegis number 3-5 stronger than the first 3. I'd love to see an aegis that scaled in a way that there is no way a 6 slotted carry doesn't pick it up. Maybe something like +2000 hp or maybe make it a buff/aura like in LoL
@Munck
Denia1
Profile Joined January 2011
148 Posts
April 09 2015 00:27 GMT
#9
I think Rosh already gives a lot. Making him even more important doesn't seem like a great solution, however scaling the jungles might be a good idea. One thing your analysis ignores is that the winning team also gets more gold from lane creeps because they are able to farm them pretty much anywhere, and so no lane creeps are 'lost' killed by an enemy creep giving no gold or exp, because no hero is there.

The game right now definitely feels better than a year ago in terms of fun during gameplay, even though I liked Faceless Void in pubs a little bit more than I like Troll Warlord, and I liked the more 'roaming' meta supports had at the time. However, the high ground/comeback mechanics are still a little bit too much. Maybe controlling key points on the opponents side could lead to some spawning of extra good catapults that only go for structures, or you could just buy them. Situations where both team just sit back, too afraid to do anything should certainly be avoided.
Bomber, MC, Jaedong, Scarlett, Grubby, DeMuslim, fy, Super, n0tail, Illidan, Universe
etofok
Profile Blog Joined March 2012
138 Posts
April 09 2015 01:08 GMT
#10
make it a buff/aura like in LoL


Yeah, I had the same idea, but disregarded it almost instantaneously because I wanted to make support heroes stronger and more fun to play in such circumstances rather than buffing a 6 slotted Medusa.
The king, the priest, the rich man—who lives and who dies? Who will the swordsman obey?
TheTenthDoc
Profile Blog Joined February 2011
United States9561 Posts
April 09 2015 01:42 GMT
#11
I wonder what impact modifying the gold rubberband to ignore tower and roshan gold would have. I think that might also help, as right now towers are almost liabilities.
ahw
Profile Blog Joined April 2011
Canada1099 Posts
April 09 2015 02:00 GMT
#12
solid graphs 10/10

if you are thinking outside the box (stuff besides tweaking rubber band numbers), I think a penatly for being down outer towers is probably the best bet. Something like double the fortification cooldown, or -2 seconds duration, or something like that. Basically, forfeiting your outer towers gives up map control and makes base defense a touch tougher.

This would still make high ground holds viable and possible to pull off, but not having a reliable fortification would make base breaking builds doable.
Comeh
Profile Blog Joined July 2008
United States18919 Posts
April 09 2015 02:20 GMT
#13
I came here for the graphs. I stayed here for the graphs.
ヽ(⌐■_■)ノヽ(⌐■_■)ノヽ(⌐■_■)ノヽ(⌐■_■)ノヽ(⌐■_■)ノヽ(⌐■_■)ノDELETE ICEFROGヽ(⌐■_■)ノヽ(⌐■_■)ノヽ(⌐■_■)ノヽ(⌐■_■)ノヽ(⌐■_■)ノヽ(⌐■_■)ノヽ(
PhoenixVoid
Profile Blog Joined December 2011
Canada32747 Posts
April 09 2015 02:47 GMT
#14
Great write-up backed with 5/5 graphs and images.
I'm afraid of demented knife-wielding escaped lunatic libertarian zombie mutants
BluemoonSC
Profile Blog Joined November 2010
SoCal8910 Posts
April 09 2015 03:06 GMT
#15
a very interesting read. i think i agree with all of your points.

mid game lineups suffer because of what you mention. you almost have to draft with the intent of being pushed into a corner waiting for a good high ground engagement.
LiquidDota Staff@BluemoonGG_
Buckyman
Profile Joined May 2014
1364 Posts
Last Edited: 2015-04-09 04:27:34
April 09 2015 04:25 GMT
#16
On April 09 2015 06:40 etofok wrote:
However because in Dota there are lane creeps that give gold every 30 seconds regardless where they are killed - be it at their spawn point or at your barracks line - this situation is weird to say the least: there can’t be enough contain to actually contain.


There is a way around this, but nobody actually does it.

1) Shove mid continuously at their base so they can't leave as 5 without being obvious about it.
2) Farmed carry takes ancients periodically, but otherwise stands ready to assist the team
3) A support freezes the offlane where the enemy T2 used to be while transitioning towards a carry or aura-wagon.
4) The position 3 is in his own jungle, pulling his own creeps into a stacked small camp to keep the safe lane away from the enemy base. He has Boots of Travel to join the frozen/shoved lane if there's a fight.
5) The remaining two heroes occupy the enemy jungle, farming it while being ready to fight if the enemy goes on mid during the shove or the lane-freezing support.

Result: The high ground defenders have 1 lane of income, compared to 2-3 plus a full jungle. And the team with map control is positioned so that there's no obvious way to initiate an advantaged fight.
Jaeger
Profile Joined December 2009
United States1150 Posts
April 09 2015 05:02 GMT
#17
horadric cube, requires one 5k component from each secret shop, merges with aegis and any 2 items to form 1 item with combined attributes of both items.
https://www.dotabuff.com/players/8137911
TheEmulator
Profile Blog Joined July 2010
28100 Posts
April 09 2015 05:08 GMT
#18
Nice post!
Administrator
Nitrohell
Profile Joined April 2015
1 Post
April 09 2015 05:54 GMT
#19
Good post.

Really like the idea of jungle creeps scaling with time, about cheese though, instead of making it worth more gold later, why not actually buff the aegis and cheese the more times you kill Roshan.

Cheese for example, after xth Roshan kill, Cheese can now be used on allies and have x range, which increases with more kills. This way the non 6-slotted players can pick cheese and then give a burst of heal to their carry when he needs it in a team fight.

With Aegis, after xth Roshan kill, Aegis now provides its restore buff in an x range area around the dying player, it restores 10% hp and mana per second for 5 seconds, buff is removed if the hero takes any damage for that duration.

Roshan gold bounty could also scale with time/times killed, that way teams will have an extra incentive to kill Rosh later, since it will benefit the supports and less farmed heroes more.
Baozi
Profile Joined March 2011
United States1191 Posts
April 09 2015 05:57 GMT
#20
✔ Graphs
✔ Starcraft references
✔ Dank memes
✔ Well-reasoned and informed analysis with proposed solutions
✘ Srapnel, ha-ha hee-hee

4/5 nice post, commended
"Universe is very spacey, we called him space man. He made a lot of space." - Arteezy
1 2 3 4 Next All
Please log in or register to reply.
Live Events Refresh
Replay Cast
00:00
2026 GSL S1: Ro12 Group A
Liquipedia
[ Submit Event ]
Live Streams
Refresh
StarCraft 2
NeuroSwarm 268
ProTech137
PiLiPiLi 32
StarCraft: Brood War
Mind 231
ZergMaN 16
Icarus 8
League of Legends
JimRising 694
Counter-Strike
Stewie2K676
Other Games
summit1g7254
WinterStarcraft578
C9.Mang0565
monkeys_forever471
RuFF_SC2152
ViBE59
kaitlyn22
ToD21
Organizations
Other Games
gamesdonequick1280
Dota 2
PGL Dota 2 - Main Stream83
StarCraft 2
Blizzard YouTube
StarCraft: Brood War
BSLTrovo
[ Show 12 non-featured ]
StarCraft 2
• OhrlRock 3
• AfreecaTV YouTube
• intothetv
• Kozan
• IndyKCrew
• LaughNgamezSOOP
• Migwel
• sooper7s
StarCraft: Brood War
• BSLYoutube
• STPLYoutube
• ZZZeroYoutube
League of Legends
• Stunt367
Upcoming Events
Replay Cast
4h 6m
RSL Revival
5h 6m
Classic vs GgMaChine
Rogue vs Maru
WardiTV Invitational
6h 6m
Percival vs Shameless
ByuN vs YoungYakov
IPSL
11h 6m
Ret vs Art_Of_Turtle
Radley vs TBD
BSL
14h 6m
Replay Cast
19h 6m
RSL Revival
1d 5h
herO vs TriGGeR
NightMare vs Solar
uThermal 2v2 Circuit
1d 9h
BSL
1d 14h
IPSL
1d 14h
eOnzErG vs TBD
G5 vs Nesh
[ Show More ]
Patches Events
1d 19h
Replay Cast
2 days
Wardi Open
2 days
Afreeca Starleague
2 days
Jaedong vs Light
Monday Night Weeklies
2 days
Replay Cast
2 days
Sparkling Tuna Cup
3 days
Afreeca Starleague
3 days
Snow vs Flash
WardiTV Invitational
3 days
GSL
4 days
Classic vs Cure
Maru vs Rogue
GSL
5 days
SHIN vs Zoun
ByuN vs herO
OSC
5 days
Replay Cast
5 days
Escore
6 days
The PondCast
6 days
WardiTV Invitational
6 days
Replay Cast
6 days
Liquipedia Results

Completed

Escore Tournament S2: W5
WardiTV TLMC #16
Nations Cup 2026

Ongoing

BSL Season 22
ASL Season 21
CSL 2026 SPRING (S20)
IPSL Spring 2026
KCM Race Survival 2026 Season 2
KK 2v2 League Season 1
Acropolis #4
SCTL 2026 Spring
RSL Revival: Season 5
2026 GSL S1
BLAST Rivals Spring 2026
IEM Rio 2026
PGL Bucharest 2026
Stake Ranked Episode 1
BLAST Open Spring 2026
ESL Pro League S23 Finals
ESL Pro League S23 Stage 1&2
PGL Cluj-Napoca 2026

Upcoming

BSL 22 Non-Korean Championship
CSLAN 4
Kung Fu Cup 2026 Grand Finals
HSC XXIX
uThermal 2v2 2026 Main Event
Maestros of the Game 2
2026 GSL S2
Stake Ranked Episode 3
XSE Pro League 2026
IEM Cologne Major 2026
Stake Ranked Episode 2
CS Asia Championships 2026
IEM Atlanta 2026
Asian Champions League 2026
PGL Astana 2026
TLPD

1. ByuN
2. TY
3. Dark
4. Solar
5. Stats
6. Nerchio
7. sOs
8. soO
9. INnoVation
10. Elazer
1. Rain
2. Flash
3. EffOrt
4. Last
5. Bisu
6. Soulkey
7. Mini
8. Sharp
Sidebar Settings...

Advertising | Privacy Policy | Terms Of Use | Contact Us

Original banner artwork: Jim Warren
The contents of this webpage are copyright © 2026 TLnet. All Rights Reserved.