Zg1s has the right idea. The strategy guide on the wiki covers this stuff in quite a bit of detail for a generalized guide https://valeofeternity.wiki.gg/wiki/2_P ... tegy_Guide
Lots of concepts summarized there including the tame/sell and pick decisions throughout the game. Your discovery that holding high scoring cards early as being game deciding is totally correct. A common 2p example is choosing to tame/sell/pass Gi-rin going second in Round 1, which will often simply add 16 to your score, add nothing, or add 16 to your opponent's score. Each of those is a huge difference.
Early Generator
Your first good stone generator should be valued quite highly, especially in the early game. We all know how large the difference between having versus not having a stone generator is. I would Round 1 first pick Undine Queen over Eternity nearly every time (with a few exceptions) because the stones are more valuable. There are so many good scoring cards in the game. I would gladly keep Eternity Round 1, but it is much more replaceable than Undine Queen. If my opponent gets a lucky R1P1 UQ and there are no other generators I would at least be content that I get Eternity plus my preferred remaining card as compensation. UQ would have slight odds to start, but if Rounds 2-4 give me a strong generator in return and not much for them the odds can quickly swing into my favor (and often does).
Once you have your first GOOD stone generator, make sure you don't get too greedy on stone generators and give up strong scoring cards. You should be trying to put yourself in a solid/balanced position in the early game. This means having a combination of stone generation AND points in hand for later or at least a good variety of threats in hand that will make it difficult for opponents to block you if you were unable to stock up on good generic scoring. I regularly punish players that get too greedy with their stone generation by cutting them off scoring in the mid & late game. They then generate way too many stones and have nothing productive to do with those stones so they are effectively wasted. Remember that the whole point of stones is to spend them to score and that this game has surprisingly few ways to greatly reward a player for setting up crazy stone generation early.
Sand Timers
Sand timer points and late game instant scoring can be compared to each other and properly evaluated quite quickly with practice. Since we are talking early game Zg1s is correct to generalize that sand timer cards score better than instants because they will have more rounds to trigger. Nessie is a simple example since it scores 2 points per round. It starts off very strong and drops off by 2 points every round, plus additional drop off depending on your desire to use Dragons.
A funny comparison here since they are pretty much polar opposite cards is Burning Skull versus Eternity as Round 1 picks. Many players would be much more threatened by the Eternity. If you are looking at them in terms of points versus stones they are actually quite similar. I have had a few recent opportunities to Round 1 first pick Burning Skull with a 4 value sell to summon it immediately and score. The result was that Burning Skull scored 24 points for a total cost of 11, divided into all the small payments. I can adjust my whole game plan around Burning Skull since it is providing up to 24 points. This greatly lowers the bar for my other scoring stuff for the whole game, allowing me to "go smaller" and still reach 60. Burning Skull has its own set of synergy pieces (1-stone generators), many of which might be low value for the opponent which allows me to pick them up easily and "fix" the 1-stone requirement problem. From experience I know that juggling stones for early game Burning Skull isn't that difficult and the points scored is impressive so I am glad to use it, even without synergy
Instants
For Eternity it is quite opposite in how you are playing it, but the same general idea of already having "points in hand" applies. It is a probable 16/20 points with many strong synergy pieces. Some cards can pay for Eternity on their own and it forms high scoring combos with Scorch or Aeris, making those cards major threats for the opponent. There are also 3 other family diversity scoring cards that can tag along well with Eternity to benefit from the same setup. It will take some work to collect different card families, summon them first, then reach 12 mana to summon Eternity but from experience players know that Eternity is quite a strong overall card since it is actually quite doable in practice. Since it is an instant scoring effect players have all game to plan around setting it up and summoning it, making Eternity quite a solid early game pick.
Interactive Pick Choices
In 2P every card that you do not pick is going to your opponent so they is quite a bit of value in picking the scoring cards even if they are not immediately useful. An Earth example is given here since I like Earth An example would be an opponent first picking Basilisk on Round 2 with Sand Giant revealed too. It would be good to prioritize Sand Giant against a player with an early Basilisk. If they pick Basilisk and get Sand Giant back, that is a very strong round for them so you need to be getting a REALLY strong card pair in return to make that a worthwhile exchange. For Sand Giant I specifically say tame, not sell because: 1) The card has enough potential for you even with Basilisk gone. 2) Because Mimic exists selling Sand Giant is always risky because Mimic can return Sand Giant, which could swing a whole game outcome on its own (and has many times). Sometimes a Mimic player can even Sand Giant twice in 1 game by removing it, especially with amazing stones from Basilisk so Sand Giant could represent a 40 point threat if Mimic arrives.
In 2P I am quite aggressive with my defensive picks since I am playing a competitive 2P game every choice is Zero Sum. I am carefully tracking everything that is happening for both players the entire game and consider all available information when I make my decisions. My pick choices are therefore based on what card pairings will give the largest net benefit to myself. As summarized in the 2P strategy guide this often means first picking a card that is valuable to BOTH players and when picking second, usually passing them back what they want the least. I evaluate cards based on their baseline/vacuum strength plus a plus/minus adjustment for each player based on synergy with what players already have, what's gone (discard pile), and what is remaining in deck, sometimes considering strategy clashing (anti-synergy) as well. Cards that are picked defensively will often sell well if they are no use to you.
One example to tie some things together
It is Round 2 and I have a Nessie (R1P2) summoned already and have a Valkyrie (R2P1) and Behemoth (R1P2) in hand (great start!), (1)(1) stones remaining. My opponent most notably has a Scorch in hand and (6)(1)(1)(1) in stones, ready to pick up something to summon.
Round 3 my opponent first picks a strong generator (Snail Maiden), putting them in a much better position and leaves me with some solid choices: Gi-rin, Troll, and Surtr, all pretty solid for them. Troll is not great for me so I think about how I will use Gi-rin and Surtr first to finalize my decision. In this position I make the uncommon decision to sell Gi-rin on Round 3 to develop my board with Valkyrie, which is projected to score me 12+ points (2 points x 6 rounds), looking close to 18 points by my estimation.
Selling Gi-rin early feels bad because it is giving up 16 points, but at least I was able to deny it from my opponent for general scoring AND a solid Scorch target. I was content to sell nearly any other card in this position to get the Valkyrie out, Gi-rin just happened to be there. Since I already have Behemoth in hand and got to tame Surtr this round I feel like I am in a solid, balanced position with many points available, don't need Gi-rin.
Let's evaluate my projected points position, assuming a Round 8 finish:
Round 3, (3/60) points (1+nessie)
Nessie +12 points (6*2, no dragon ever)
Valkyrie +18 points (something like 2, 2, 3, 3, 4, 4 = 18 projected)
Surtr +6 points at worse, ideally 8 or 10 by summoning late
Behemoth +12 points expected.
Total points available (3+12+18+6+12) = 51/60 points available by Round 3, wow. This makes it clear that the Gi-rin is certainly not needed. Also, how much space does this plan require? Only 4 space? That's great! Plenty of room to add stone generators or additional scoring cards. It would be almost impossible to miss 60+ points by Round 8 unless the Valkyrie is quickly discarded via Boulder.
Lots of concepts summarized there including the tame/sell and pick decisions throughout the game. Your discovery that holding high scoring cards early as being game deciding is totally correct. A common 2p example is choosing to tame/sell/pass Gi-rin going second in Round 1, which will often simply add 16 to your score, add nothing, or add 16 to your opponent's score. Each of those is a huge difference.
Early Generator
Your first good stone generator should be valued quite highly, especially in the early game. We all know how large the difference between having versus not having a stone generator is. I would Round 1 first pick Undine Queen over Eternity nearly every time (with a few exceptions) because the stones are more valuable. There are so many good scoring cards in the game. I would gladly keep Eternity Round 1, but it is much more replaceable than Undine Queen. If my opponent gets a lucky R1P1 UQ and there are no other generators I would at least be content that I get Eternity plus my preferred remaining card as compensation. UQ would have slight odds to start, but if Rounds 2-4 give me a strong generator in return and not much for them the odds can quickly swing into my favor (and often does).
Once you have your first GOOD stone generator, make sure you don't get too greedy on stone generators and give up strong scoring cards. You should be trying to put yourself in a solid/balanced position in the early game. This means having a combination of stone generation AND points in hand for later or at least a good variety of threats in hand that will make it difficult for opponents to block you if you were unable to stock up on good generic scoring. I regularly punish players that get too greedy with their stone generation by cutting them off scoring in the mid & late game. They then generate way too many stones and have nothing productive to do with those stones so they are effectively wasted. Remember that the whole point of stones is to spend them to score and that this game has surprisingly few ways to greatly reward a player for setting up crazy stone generation early.
Sand Timers
Sand timer points and late game instant scoring can be compared to each other and properly evaluated quite quickly with practice. Since we are talking early game Zg1s is correct to generalize that sand timer cards score better than instants because they will have more rounds to trigger. Nessie is a simple example since it scores 2 points per round. It starts off very strong and drops off by 2 points every round, plus additional drop off depending on your desire to use Dragons.
A funny comparison here since they are pretty much polar opposite cards is Burning Skull versus Eternity as Round 1 picks. Many players would be much more threatened by the Eternity. If you are looking at them in terms of points versus stones they are actually quite similar. I have had a few recent opportunities to Round 1 first pick Burning Skull with a 4 value sell to summon it immediately and score. The result was that Burning Skull scored 24 points for a total cost of 11, divided into all the small payments. I can adjust my whole game plan around Burning Skull since it is providing up to 24 points. This greatly lowers the bar for my other scoring stuff for the whole game, allowing me to "go smaller" and still reach 60. Burning Skull has its own set of synergy pieces (1-stone generators), many of which might be low value for the opponent which allows me to pick them up easily and "fix" the 1-stone requirement problem. From experience I know that juggling stones for early game Burning Skull isn't that difficult and the points scored is impressive so I am glad to use it, even without synergy
Instants
For Eternity it is quite opposite in how you are playing it, but the same general idea of already having "points in hand" applies. It is a probable 16/20 points with many strong synergy pieces. Some cards can pay for Eternity on their own and it forms high scoring combos with Scorch or Aeris, making those cards major threats for the opponent. There are also 3 other family diversity scoring cards that can tag along well with Eternity to benefit from the same setup. It will take some work to collect different card families, summon them first, then reach 12 mana to summon Eternity but from experience players know that Eternity is quite a strong overall card since it is actually quite doable in practice. Since it is an instant scoring effect players have all game to plan around setting it up and summoning it, making Eternity quite a solid early game pick.
Interactive Pick Choices
In 2P every card that you do not pick is going to your opponent so they is quite a bit of value in picking the scoring cards even if they are not immediately useful. An Earth example is given here since I like Earth An example would be an opponent first picking Basilisk on Round 2 with Sand Giant revealed too. It would be good to prioritize Sand Giant against a player with an early Basilisk. If they pick Basilisk and get Sand Giant back, that is a very strong round for them so you need to be getting a REALLY strong card pair in return to make that a worthwhile exchange. For Sand Giant I specifically say tame, not sell because: 1) The card has enough potential for you even with Basilisk gone. 2) Because Mimic exists selling Sand Giant is always risky because Mimic can return Sand Giant, which could swing a whole game outcome on its own (and has many times). Sometimes a Mimic player can even Sand Giant twice in 1 game by removing it, especially with amazing stones from Basilisk so Sand Giant could represent a 40 point threat if Mimic arrives.
In 2P I am quite aggressive with my defensive picks since I am playing a competitive 2P game every choice is Zero Sum. I am carefully tracking everything that is happening for both players the entire game and consider all available information when I make my decisions. My pick choices are therefore based on what card pairings will give the largest net benefit to myself. As summarized in the 2P strategy guide this often means first picking a card that is valuable to BOTH players and when picking second, usually passing them back what they want the least. I evaluate cards based on their baseline/vacuum strength plus a plus/minus adjustment for each player based on synergy with what players already have, what's gone (discard pile), and what is remaining in deck, sometimes considering strategy clashing (anti-synergy) as well. Cards that are picked defensively will often sell well if they are no use to you.
One example to tie some things together
It is Round 2 and I have a Nessie (R1P2) summoned already and have a Valkyrie (R2P1) and Behemoth (R1P2) in hand (great start!), (1)(1) stones remaining. My opponent most notably has a Scorch in hand and (6)(1)(1)(1) in stones, ready to pick up something to summon.
Round 3 my opponent first picks a strong generator (Snail Maiden), putting them in a much better position and leaves me with some solid choices: Gi-rin, Troll, and Surtr, all pretty solid for them. Troll is not great for me so I think about how I will use Gi-rin and Surtr first to finalize my decision. In this position I make the uncommon decision to sell Gi-rin on Round 3 to develop my board with Valkyrie, which is projected to score me 12+ points (2 points x 6 rounds), looking close to 18 points by my estimation.
Selling Gi-rin early feels bad because it is giving up 16 points, but at least I was able to deny it from my opponent for general scoring AND a solid Scorch target. I was content to sell nearly any other card in this position to get the Valkyrie out, Gi-rin just happened to be there. Since I already have Behemoth in hand and got to tame Surtr this round I feel like I am in a solid, balanced position with many points available, don't need Gi-rin.
Let's evaluate my projected points position, assuming a Round 8 finish:
Round 3, (3/60) points (1+nessie)
Nessie +12 points (6*2, no dragon ever)
Valkyrie +18 points (something like 2, 2, 3, 3, 4, 4 = 18 projected)
Surtr +6 points at worse, ideally 8 or 10 by summoning late
Behemoth +12 points expected.
Total points available (3+12+18+6+12) = 51/60 points available by Round 3, wow. This makes it clear that the Gi-rin is certainly not needed. Also, how much space does this plan require? Only 4 space? That's great! Plenty of room to add stone generators or additional scoring cards. It would be almost impossible to miss 60+ points by Round 8 unless the Valkyrie is quickly discarded via Boulder.