Types Of Bleeds And Pushes In Gwent


In the preceeding chapter of the tutorial (Bleeding & Pushing or How To Play Round 2?) the general structure and economy of Round 2 in Gwent: The Witcher Card Game was explained. A natural follow-up would be to look into pushes and bleeds occuring in practice and discuss the ideas behind them. We would start from section devoted to pure push and then move to bleeding.

To keep text shorter we would keep following convention:

P1 = player with initiative in R2

P2 = player with second say in R2, possibly defending the push.

What Push May Do?

Let’s see what could be achieved by push outside bleeding.

  • Shortening R3

Each card below 7 played by P2 in Round 2 shortens Round 3 by 1. If P1 manages to get out all of P2 cards then Round 3 would be 3 cards long.

An engine (card getting value over time) gaining +2 points per turn is worth 20+ points in a long round 3, but only 6+ points in a 3 card Round 3. Moreover, if P2 deck is heavily engine oriented, then engines would compete for value; if there are three +2 per turn engines in P2 hand in short R3, then value added to base power is only +4 on average – even if first engine may play on par with pointslam counterparts, the next two will play below power vs provision curve

Pointslam vs Engines is a classical example of situation where R2 push with the purpose of shortening the round should be applied (usually only win-con, have a look at World Masters 1 final game), but not exclusive. Another one is a matchup where first say is favored for the engine part of the deck, but at the same time last say is favored for pointslam/control part. It is then advisable to trade engines in R2 with first say, so that this part doesn’t matter too much in Round 3.

An exemplary matchup would be a mirror of non-devotion Nature’s Gift Symbiosis decks with Harald Gord and tall punish (Gwent 7.2). The side which sets up Symbiosis engines first would be able to remove opponent ones with extra value on Rebukes. At the same time, last say is favored outside engines+Rebukes part because of Gord+tall punish deadlock. The side with initiative would then try to trade engines (hard to call it a bleed even, as trades are symmetrical), while keeping last say for R3.

  • First say abuse

It is not a rare event that Round 2 push doesn’t eye any particluar trades, nor aims at manipulating Round 3 structure, but simply first say is strongly prefered in a given matchup.

Such situation commonly occurs in reactive engines mirrors (for example Stockpile Siege), or in matchups of decks where board setup is required before playing reactive payoff (Crimes Syndicate, Symbiosis…). P1 simply wants to seize as many points as possible thanks to first say and considers passing only if first say advantage vanished somehow.

  • Abusing Hand Quality Advantage

While the decision to 2:0 often arises spontanously during R2 push, sometimes the hand power invites playing for a win in the round right from the start.

According to ‘The Rule Of 16 vs Card Drawing Probability‘ in R2 there would be about 7/10 strongest cards in opponent’s hand on average. If playing no tutor, there would be 67% chance opponent has strongest card, 44% for 2 cards, 28% for 3 (to check more numbers look at GM&P6: Tutors). If your crucial package is completed, then you have a decent chance to 2:0 just because of probabilities. To expect a 2:0, your hand should have at least equal points to opponent’s hand with extra card (sidenote: the higher point gap between golds and bronzes in the meta, the more often such a brute force push is successful)

Proven that the hand is really outstanding and opponent couldn’t make use of (double)last say in R2, little could go wrong even if opponent defends the push at even cards. If so, then they committed probably just as much as you did.

Missing the opportunity of Hand Quality Advantage Abuse would probably result in P2 relatively improving their hand in R3 way more than P1

  • Tempo Abuse

If P1 deck has advanatage of tempo plays over P2, then P1 may try to achieve card advantage with a tempo pass after a few powerplays. Such technique is not very common as usually requires heavy commitment and then payoff from having double last say. It was sometimes used in Elf + Poison hybrid deck in earlier Homecoming; Elves package tempoed opponent out, while Poisons with double last say got perfect targets with no threat of purify. 

  • Pushing Against Leader

Pushing against leader means shortening R3 and careful choice of cards kept in hand in order to make opponent’s leader awkward in a short round. Such play is especially possible against reactive leaders after winning on even; in a shorter Round 3 it would be easier to avoid giving P2 good targets. Examples of leaders potentially struggling in a short round with first say are Blood Scent, Reckless Flurry or Enslave.

  • Push For Carryover / Against Carryover

Sometimes it is advanatgous for P1 to play into Round 2 just to establish carryover. Ciri:Nova or Vandergrift are exemplary cards (Gwent 10.9). If losing the card, carryover gained should in general case exceed Ciri:Nova value to gain any reasonable advantage (Getting Ahead And Raw CA Value). Another type of carryover to consider are echo cards like Aerondight or Amphibious Assault. Trading any of these with a non-echo gold without losing a card would be obviously advantageous.

Push could also be played to deny carryover gained by opponent in the case of drypass. Classical example here would be the threat of Squirrel played on Melusine in Elves vs Self-wound matchup (Gwent 10.9).

  • Desperado

When matchup is strongly unfavored in a long R3 and at the same there is no real chance to achieve anything concrete with R2 push, then P1 may go for Desperado.

P1 would push in R2 and put pressure on P2. The reasoning is that R2 is harder to play than R3 – balance between tempo and commitment has to be found. P1 only win-con is P2 making mistakes.


  • Getting out a single card

The conditions to get out a single card are described in Bleeding & Pushing or How To Play Round 2 in ‘R2 Position’ section. In general there are two types of single cards worth getting out in most matchups:

  1. Scenarios/Powerful Engines
  2. Source of all points

From technical point of view there is no big difference between Scenarios and Strong Engines – both of these play for low tempo and are crucial long round cards. Type (1) is very common in Gwent 10.9 reality – let’s take Scenarios played by Guerilla Harmony or Royal Inspiration Knights, or Roland Bleinheim in Lined Pockets. Due to low tempo of (1), decks playing them are especially prone to bleed and usually like to fight for round control in Round 1 to avoid defending the bleed.

The best P2 could achieve is getting more points than P1 in the same number of cards played, so that (1) gets saved for Round 3 (probably 4 cards long).

Depending on how fast (1) is, different tempo would be optimal from P1. The moment (1) gets dropped is usually correct time to pass. For P1 it would be optimal to have a very small lead at this point (thanks to little commitment), or have a lead bigger than (1) + average opponent’s card.

The category (2) is essentially different from (1). Cards from (2) are distinguished by powerful instant value, usually gained at expense of excessive provision cost or long development. Examples of (2) are Alzur with Orbs of Insight chain or Simlas into 4x/5x Waylay. The fundamental difference in approach is that P1 would usually just try to force (2), while not caring about its reach – card would be lost anyway, but opponent would lose the source of point which is worth even more.

For (2) to be really worth bleeding, either the garnish defending it must be very weak, or it must play 8 or more points above the final cards you are ready to commit in R2 (‘Ciri:Nova’ visualization, as in the preceeding chapter) . It usually means 20+ points.

P2 could consider a mulligan of (1) to prevent getting it out. The risk is always 2:0. Similar action with (2) card is less conceivable, as missing the main source of points would more likely result in lost round. Also the situation when (2) gets bled out is often not as clear-cut bad as (1) proven that P1 made some commitments during bleed. 

  • Disrupting a (combo) packet

When P2 keeps in hand a packet of N R3-deciding synergistic cards, the bleeding would be rather easy to execute. If the packet size is equal to 3 or bigger, then P1 doesn’t need to play the last card to disturb the packet, even when not ahead. It is one of the reasons why 3+ card combos are so rare in competitive Gwent.

While in the case of getting out a single card, P1 should have assessed if points in hand would be enough to get (1)/(2) out, in the case of combos the disrupt is almost guaranteed. The only real risk of failure is in the case of N=2 and opponent getting ahead so much that the last card is unable to close the point gap.

Another practical threat is P2 playing tricky and mulliganing out some combo pieces, possibly leaving them at flexible access with tutors.

A Gwent 10.9 example of combo deck is Koshchey. In general, combo decks need round control and strong Preparation Team even more than ‘getting out a single card’ targets from the previous point.

  • Payoff Split

If P2 runs a crucial card scaling with the number of units on the board, then P1 may consider playing around it by splitting long R3 into short R2 and R3. The technique is very similar to ‘Shortening R3’ in Push section, the only difference is playing against one particular card rather than whole opposing archetype.

Payoff Split is a short bleed, ending even before the target card could be played for decent value; it shouldn’t be mistaken with getting card out, although may result in target being played for suboptimal value.

If at the risk of losing card, the payoff must be game deciding factor, worth 20+ points (Gwent 10.9). Examples of number of units payoff cards in Gwent 10.9 are Triss: Meteor Shower, Schirrú, Wild Boar of the Sea, Geralt:Igni or Geralt:Yrden. Note that decks playing those type of payoff cards would struggle to bleed opponent themselves.

  • Quality (Value) Bleed

A negative of abusing hand quality advantage. P1 hand is very weak, maybe only 1/2 golds are in hand in R2. Even if long R3 would normally be favored for P1, the chances to improve hand enough are too low. What P1 could then consider is simply trading the bronzes with opponent’s cards. Opponent probably tried to make hand as strong as possible during mulligans and as presented in The Rule Of 16 vs Card Drawing Probability,  would have about 7 golds in hand (unless commited sth in R1).

Quality bleed makes sense only if R2 hand is really weak so that playing really deep with good trades is possible. In a 10 cards R2, trade payoff would start on average only below 7 cards. Then P1 at best would be able to go down to 1 card, so that 2 best golds would be saved in P2; trading the last card is usually unreal. P1 then loses the card, but would have probably much better topdecks in a 4 vs 5 Round 3. To get real quality advantage in R3, P2 must topdeck only bronzes and P1 only golds, which is rather unlikely, unless R2 hand of P2 was close to perfect. Quality bleed from card-down position then is rather a tool to get more chances to save the game than a reasonable way of winning games. Unless there are other factors which benefit P1 from shortening the round, or the last P1 card is (2) – source of all points, quality bleed will more often fail than work.

The situation changes if starting position is a win-on-even. Then P1 may even consider purposeful worsening of their hand to ensure good trades only if going deep enough is guaranteed. Because of this technique, good Gwent players sometimes accept losing on even right on 7 cards rather than 5; longer round makes it harder to achieve perfect quality bleed.

  • Bleeding Against Hand

Perhaps the most mastermind kind of all bleeds. There may be no casual bleeding targets in opponent’s hand, but just the fact of R2 being played make P2 sequencing awkward and cards do not find optimal value. P1 then must predict the shape of P2 hand to start the bleed.

Sequencing issues may also strengthen normal target bleeding. Imagine that you bleed card down against Knights. Vandergrift is still left to be played, and your deck runs Muzzle. Opponent’s hand also probably contains Maiden’s Shield and Damsel in Distress. To not get Vandergrift stolen, opponent would have to commit one of those cards as 3rd to last. Then, unless capable of generating enough point gap, P1 would simply pass. It is possible Muzzle wasn’t even in P1 hand, just P2 known the decklist and played around possible threat!

  •  Provoking A Play

Getting a card out is not always about establishing enough point lead. Sometimes just a threat of doing so in the future is sufficient.

For example Warriors SK lack tools other than leader+Eist combo to deal immediately with tall enough greedy engines. If such engine is accessible in R2, then it could provoke immediate leader use, which is worth 20+ points. P1 may then pass and similar engines would go through in a long enough R3. Point trade was worth more than extra card.

Another classical example is bleeding against Double Cross leader. If P1 hand with leader is stronger than P2 without leader, then P1 simply 2:0 if P2 hoards leader ability. If hand gives no real chance to surpass P1 leader, P2 would have to use leader at some point; the later use, the higher risk of P1 playing around Double Cross by leaving useless cards in hand.


I think the read is long enough at this point and the main topic of various types of pushes and bleeds have been described thoroughly, so time for a break. The last part of the cycle would deal with bleeding tips and general considerations. To be continued…