Tetris Battle update

[Alexsweden]

The title says it all i think, beta version of marathon and sprint is added to Tetris Battle, a clear try to merge the two tetris games available in facebook. Getting closer to what TF is.

Official statement
[spoiler]"We been hard at work trying to bring out the newest game features as soon as possible.
This topic is about the new features currently added into the game.
There are two new modes in Tetris Battle, Marathon and Sprint.
Marathon - Single player game with 15 levels. Score as high as possible within 15 levels.
Sprint - Single player 40 lines game. Clear 40 lines as fast as possible.
You can find those games also in Tetris Friends. We added these two modes in Tetris Battle so you don't have to go two places to play all these modes.
They are in beta form, so do expect more features in the future.
For new users who just started playing Tetris Battle today, those two modes including Sprint 4p and Battle 6p are locked until they meant the unlocking requirements.
For users who played our game before, Sprint and Marathon are locked until they meant the unlocking requirements, while Battle 2P, Battle 6P, and Sprint 4P will be unlocked."


http://www.facebook.com/topic.php?uid=1304...&topic=5712
[/spoiler]
Sprint seems good and is better than the other tetris game on facebook as there is some tuning at least, not much else to say. Gives 5exp and 10gold for a successful run independently of your time.

Marathon is good too but very different from TFs marathon (in fact it is closer to TF Survival).  Marathon awards experience and coins equal to the level you achieved. You need to clear at least 40 lines to get any reward, that is level 7 with 5 lines to go for the next level which will give you 7 gold and exp. Edit: think they have updated it, I have now received coins at a loss at level 3 with only 15k points..

The major changes that I found are:
1 In the end of each level you cannot do a tst to get more points as the count continues on to the next level. e.g. you have 1 line left in level 1 and make a double, now you reach level 2 but it counts as you would already have cleared 1 line there.
Edit: In fact if you do a move which goes over two levels it clears lines from the level above but gives you the points as if it was cleared in the level below. So in the example above you would gain 300 points for the double not 600 had it been cleared fully in level 2.

2 Each line clear may it be spins or whatever is counted exactly as then number you cleared. Tsd taking away two lines from the count, tss taking one and any number of combo still only takes away one, two, three or four lines depending on how many lines you do clear, how far along the combo you are is irrelevant to the count.

3 Each level have different number of lines to be cleared to come to the next level as compared to TF. This is of course natural as the system for counting lines is different.
Table with lines per level and scoring:
[spoiler]

Code Select Expand


Level            Lines
1                  5            
2                  5                
3                  5                
4                  5                
5                  5                
6                 10                
7                 10                
8                 10                
9                 10                
10                10                
11                15                
12                15                
13                15                
14                15                
15                15                

Points:
Single: 100* level
double 300*level
tripel 500*level
tetris 800*level
T-spin Tripel 1200*level
T-spin Duble 800*level
T-spin single 300*level (does not break b2b)
Natural T-spin single 500*level
T-spin null 100*level
combos add 50*level for each increment
Bravo not awarded
b2b awarded normally
harddrop gives 2*drop height softdrop gives 1
[/spoiler]
4 The game also seems faster than other marathon games. In TB at level 15 the speed is about the same as for 15 in TF. However the permitted actions before a piece locks is much lower. In TB it seems to be around 6 and in TF its 15 or 16. This makes the game much harder to play when trying to follow a certain system, ST stacking for example.  I believe this is good as it makes it much more interesting as it is now harder to get a "perfect game".


How to play marathon to maximize scores then?

T-spin zeroes and T-spin tripels
[spoiler]Let's say you use a T for a tsz, you get 100points and have now used a bag which creates 2.8lines and as the tsz does not clear anything you would either have use the next bag to do another tsz and then a tetris alt. combo down but this would not be very efficient. Another way to do it would be to in bag 1 do a tsz(100p) and in the next do a tst(1200p) this would clear 3 out of you now 5.6 lines( 2 * 2.8 ) which gives you 2.6 lines left. If this is done over 150 lines or 53.6 bags you can do 27 sets of tsz-tst and then you are left over with 70.2 lines which has to be cleared with combo or tetrises. 70.2 lines gives 17 tetrises with 2.2lines left over which you can do a single and then a tst with as in the end you can extend your bag count of 53.6 by stacking high on the sides to get the right pieces then do a by that way clearing 152 lines.
This would give you the following points neglecting back to back and any level multiplier:
100 + 1200 * 27 (tsz + tst) = 35 100p
17 * 800 (the tetrises to clear the leftovers) = 13 600
100 + 1200 (single and tst - final use of the "wasted" 2.2 lines) = 1300
Maybe 2tsz can be added as you can stack high in the end without clearing anything. (+ 2 * 100)
In total: 35 100+13 600+1200=50 000points. Edit: and with b2b around  50 000 * 1.5 = 75000 points

Final score: 725 000* (see explanation under the last spoiler tag.)

The single added in the end would of course ruin b2b and hence it would be ideal to clear a single at the start and then start stacking. This would shift the whole setup one set ahead which would also lead to that some line clears that were in-between two levels ended up above overall increasing the score.[/spoiler]
Only T-spin tripels
[spoiler]If you only make tst on the other hand you will have 150lines / 3 = 50tst. This gives you 50 * 1200 = 60 000 points. If we start of the game with and tsd we shift the whole thing so that we end up with clearing 152 lines which is legit. So we can add another 800p t the score giving us 60 800 in total. Edit: b2b 91200

Final score: 881 600*

We would also end up in "waste" T's that could theoretically be used to make tsz to give even more points but that would probably not be possible if using Massi4h's Infinite TST.
There might be a problem with using Massi4h's Infinite TST for 150 lines too, But Massi4h has a video showing him doing 32 tst with that system (2 last ones used 180 and cannot be done on TB) so it can be done for a long while at least. I believe this to be the best way to play if it can be done for 150 lines. If you could start out with and tsd it would be even better for this.
Here is the video:
http://harddrop.com/videos/id_618/title_34...ssi4h%E2%80%99/
[/spoiler]
ST stacking
[spoiler]In ST stacking you have 4 tsd for every tetris with an extra tetris every fifth tetris. 4tsd and a tetris clear 12 lines and are worth 5 * 800 = 4000 points. Doing five of these would clear 60lines and the extra tetris would add 4 more. This you can do twice to end up at 128 lines and with 22 lines left you clear 8tsd and a tetris with two lines left for use of a single and then a tst or a tsd.
This would give 4000 * 5 + 800 (five sets of 4tsd and a tetris plus the added tetris) = 20 800
This can be done twice: 20 800 * 2 = 41 600
Add the 8tsd and the tetris: 41 600 + 9*800 = 48 800
Finally add the single and the tst:  48800 + 100 + 1200 = 50100 points. Edit: b2b 75150

Final score: 726 450*

But we have to take into consideration that in fact you don't have to do 4tsd and then a tetris, by stacking high and not doing a tetris one might be able to not do the last tetris at all but do 2tsd instead.

As stated before the single added in the end would of course ruin b2b and hence it would be ideal to clear a single at the start and then start stacking. This would shift the whole setup one set ahead which would also lead to that some line clears that were in-between two levels ended up above, overall increasing the score.

This is a way to start the game so that you not only do you gain that single I was mentioning earlier but you also start a b2b so that your first tsd gives 1200 instead of 800.
[fumen]110@7eKBfIihBLfB+gB9jBncBiHPIAo3MoDFbUwEOQPAAL?aBliBNYB[/fumen]
Here is a video of it in action

It seems like I lost around 170k at the mistakes done at evel 13-14. But since I could start up a new ST stack I saved myself and got a great score anyway. Had I played perfectly there maybe I would had 800k.

I stated that the maximum score would be around 726k. However since the points from different levels are not static my approximation of 10* score from level 1 does not seem to suffice. Also this is without points from hard and softdrops. My TSS at the start did not only iniciate a B2B. It also ensured that had I had an perfect ST stack I would have not only had all 15 lines at level 15 to make use of the *15 multiplyer for all 15 lines but with TSD's i would have ended up at 1 line away from goal letting me do a TSD at the end (dont do TST's). Also if you watch my replay you will see that for all levels with 10 lines this method ensures that you get the maximum of each and every one of these as you will start at 10 lines on each level hence you will not have any lines cleared between two levels and therfore lose points on that. When level 11-15 kick in you have 15 lines each level and so every second level you will end up with 1 line below and every second 1 line above a level line when you do a TSD. best would be if you would have 1 line below at level 15 as this gives you "extra lines" at the higest level and score multiplier. And as it turns out the TSS does exactly this. So there is no doubt If you are serious about doing ST at marathon you should start out with my TSS at the start as this is the best way to play ST that we know of as of yet.

New reccord

This is a normal almost perfect ST stack. I did not do the TSS at the start since it is not something you can do with every opening bag, also it oftentimes makes the rest of your stack harder to control. This could be due to that I don't put down two T's at the right side stack or something when using this opening. Anyways, it is a pain to play with so I decided to make a try at a normal ST stack.

The difference between the two is quite small before the higher levels, so if you are nowhere near a perfect ST game then dont bother doing the TSS - ST stacking. Just do the normal one.

[/spoiler]
4w with TZ0s
[spoiler]A..... B ... C   ...D........E........F*
1   100   100   20400   136   197200
2   150   250   24150   161   233450
3   200   450   27900   186   269700
4   250   700   31650   211   305950
5   300   1000   35400   236   342200
6   350   1350   39150   261   378450
7   400   1750   42900   286   414700
8   450   2200   46650   311   450950
9   500   2700   50400   336   487200
10   550   3250   54150   361   523450
11   600   3850   57900   386   559700
12   650   4500   61650   411   595950
13   700   5200   65400   436   632200
14   750   5950   69150   461   668450
15   800   6750   72900   486   704700
16   850   7600   76650   511   740950
17   900   8500   80400   536   777200
18   950   9450   84150   561   813450
19   1000   10450   87900   586   849700
20   1050   11500   91650   611   885950
21   1100   12600   95400   636   922200
22   1150   13750   99150   661   958450

A is the amount of combo
B is the points for that particular combo
C is the sum of all combos up until and including the current combo
D is the final sum - C * 150 / A + (54*100). The 5400 points come from perfect usage of T-spin zeroes.
Edit: E is points per line
Edit2: F is the real final score*

Below you have non important stuff really, did it for my own sake mostly
Keep in mind that some values are a bit above the perfect value as 150 / A (number of times a combo of 'A' magnitude can be done in 150 lines) in many cases gives a non integer number and so gives you a .xx value of the C value and as the value of C grows exponential the value is higher than it actually should be.

E.g. a combo of 20 can be done 7.5 times. A full combo gives 11 500 points but a 0.5 combo - which is 10 gives only 3250 which is only 0.28 of the full combo. But the calculation gives you 11 500 * 0.5 which is 5755 and so the value is 2505points too high.

Of course I am assuming that this is the real values for the combo, haven't checked it all yet. It is true for the first couple of values though.
[/spoiler]
Also do remember that playing faster and keeping the stack low will give you more points for harddroping the pieces a longer distance. As it gives 2 points per square on its way down and we use approx. 54 bags to clear all 150 lines required this could give us and extra 14 364 points if all 378 pieces were to be harddroped the maximum distance of 19 rows (I haven't been able to harddrop the whole 20 rows even by spamming so i assume 19 is the maximum)
Numbers on how much you can gain in using harddrops:
[spoiler]

Code Select Expand

A     B     C  
---------------------
1    2    756
2    4    1512
3    6    2268
4    8    3024
5    10    3780
6    12    4536
7    14    5292
8    16    6048
9    18    6804
10    20    7560
11    22    8316
12    24    9072
13    26    9828
14    28    10584
15    30    11340
16    32    12096
17    34    12852
18    36    13608
19    38    14364
--------------------
A is the number of lines dropped
B is the number of points gained
C is the total value of all 378 pieces (54 bags * 7)
[/spoiler]
*
[spoiler]I tried to figure out how you could calculate the real maximum score in an easy way. I added an excel attachment that shows that if you multiply the values obtained from pretending that you play all 150 lines at level 1 by 10 (9.666.. really) then you get a pretty good match for what the value would really be.  

The reason to why 10 is the right number is that I calculated how much points you would get from doing only non combo singles. The final real score would then be 145 000 points. I then compared this to the value gotten from multiplying 100 * 150 - that is the value obtained from pretending that you play the whole game at level 1.
145 000 / 15000 = 9.666..
A problem with this is that as soon as you do the same calculation with non singles you have to take into account that sometimes a level that has a "x line goal" the number of lines cleared at that particular level may be up to x+/- 3 as. For example if you do two tetrises at the start of the game you have cleared 8 lines and all these lines are treated with the level 1 multiplier.

note: I am assuming that the level multiplier also affects the b2b - haven't checked that yet though.

So when using this method remember that it is not exact, we can only hope that these variations cancel out in the long run giving pretty precise results anyway.[/spoiler]

[Kitaru]

I find it a little sad that official games are so far gone that what used to be sensible, standard Marathon behaviors now actually require additional explanation.

The weighted line counts are somewhat interesting though. Also, I'd double-check that t-spin single, as it's probably still 400 to fit the 400*lines t-spin formula. (See: T-Spin Triple = 1200 base points). I would do it myself, but I'm having connection problems I don't care to fix at this hour.

[Ravendarksky]

So basically you want to do Tspin triples constantly and balance it out with a tetris so your last three lines of each level are a tspin triple?

Are b2bs rewarded? Also what is the scoring for combos?

[Alexsweden]

I find it a little sad that official games are so far gone that what used to be sensible, standard Marathon behaviors now actually require additional explanation.

The weighted line counts are somewhat interesting though. Also, I'd double-check that t-spin single, as it's probably still 400 to fit the 400*lines t-spin formula. (See: T-Spin Triple = 1200 base points). I would do it myself, but I'm having connection problems I don't care to fix at this hour.

nope, tss is 300*level. But as you said it would make more sense if it were 400
Edit: ts zero gives 100*level
Edit2: tss is gives the same as a normal single and a tszero which in a way makes sense.

So basically you want to do Tspin triples constantly and balance it out with a tetris so your last three lines of each level are a tspin triple?

Are b2bs rewarded? Also what is the scoring for combos?

Dont know what you mean about last three lines of each level, you get just as much points in the beginning as in the end of each level, just as long as you dont cross the border.
b2bs are rewarded and it seems normal a bsb tsd gives 1200 instead of 800.
combos add 50*level for each increment

[Rosti_LFC]

I'm too far out of the loop to see the implications purely from the score mechanics, but is the best tactic still the dumbass single-single combo stuff, or are techniques like ST-stacking more plausibly optimal?

[mippo]

As far as I can see Alex says lines cleared count as lines (so TSD = 2) which is more similar to TF Survival, so ST stacking is the way to go.

In TF Marathon tsds count as 6 or 7 lines, that's why the single-single combo was best.
Haven't tried it yet though.

[Alexsweden]

I'm too far out of the loop to see the implications purely from the score mechanics, but is the best tactic still the dumbass single-single combo stuff, or are techniques like ST-stacking more plausibly optimal?

Yes ST stacking is probably good here, as mippo said it is like TF survival in gameplay.

Edit: 250th post

[Paul676]

surely just as many tspin zeros as possible would be the perfect tactic... if they take nothing from the line count

[mippo]

I tried to play it but you need to have 10 or 12 friends playing TF battle? Maybe I gotta add some people from here sigh...

[Alexsweden]

Is that an requirement to play it?, lucky me had em from the start

surely just as many tspin zeros as possible would be the perfect tactic... if they take nothing from the line count

First I also thought that it sounded like a good idea, however if you want to use Ts for T-spin zeros you sometimes do this at the cost for another t-spin :/
Let's do some math

All points below are treated as if they were gained at level 1 without b2b.
T-spin zeroes and T-spin tripels
[spoiler]Lets say you use a T for a tsz, you get 100points and have now used a bag which creates 2.8lines and as the tsz does not clear anything you would either have use the next bag to do another tsz and then a tetris alt. combo down but this would not be very efficient. Another way to do it would be to in bag 1 do a tsz(100p) and in the next do a tst(1200p) this would clear 3 out of you now 5.6 lines( 2 * 2.8 ) which gives you 2.6 lines left. If this is done over 150 lines or 53.6 bags you can do 27 sets of tsz-tst and then you are left over with 70.2 lines which has to be cleared with combo or tetris. 70.2 lines gives 17 tetrises with 2.2lines left over which you can do a single and then a tst with as in the end you can extend your bag count of 53.6 by stacking high on the sides to get the right pieces then do a by that way clearing 152 lines.
This would give you the following points neglecting back to back and any level multiplier:
100 + 1200 * 27 (tsz + tst) = 35 100p
17 * 800 (the tetrises to clear the leftovers) = 13 600
100 + 1200 (single and tst - final use of the "wasted" 2.2 lines) = 1300
maybe 2tsz can be added as you can stack high in the end without clearing anything. (+ 2 * 100)
In total: 35 100+13 600+1200=50 000points. Edit: and with b2b around  50 000 * 1.5 = 75000 points

The single added in the end would of course ruin b2b and hence it would be ideal to clear a single at the start and then start stacking. This would shift the whole setup one set ahead which would also lead to that some line clears that were in-between two levels ended up above overall increasing the score.[/spoiler]

Only T-spin tripels
[spoiler]If you only make tst on the other hand you will have 150lines / 3 = 50tst. This gives you 50 * 1200 = 60 000 points. If we start of the game with and tsd we shift the whole thing so that we end up with clearing 152 lines which is legit. So we can add another 800p t the score giving us 60 800 in total. Edit: b2b 91200
We would also end up in "waste" T's that could theoretically be used to make tsz to give even more points but that would probably not be possible if using Massi4h's Infinite TST.
There might be a problem with using Massi4h's Infinite TST for 150 lines too, But Massi4h has a video showing him doing 32 tst with that system (2 last ones used 180 and cant be done on TB) so it can be done for a long while at least. I believe this to be the best way to play if it can be done for 150 lines. If you could start out with and tsd it would be even better for this.
Here is the video:
http://harddrop.com/videos/id_618/title_34...ssi4h%E2%80%99/
[/spoiler]

ST stacking
[spoiler]In ST stacking you have 4 tsd for every tetris with an extra tetris every fifth tetris. 4tsd and a tetris clear 12 lines and are worth 5 * 800 = 4000 points. Doing five of these would clear 60lines and the extra tetris would add 4 more. This you can do twice to end up at 128 lines and with 22 lines left you clear 8tsd and a tetris with two lines left for use of a single and then a tst or a tsd.
This would give 4000 * 5 + 800 (five sets of 4tsd and a tetris plus the added tetris) = 20 800
This can be done twice: 20 800 * 2 = 41 600
Add the 8tsd and the tetris: 41 600 + 9*800 = 48 800
Finally add the single and the tst:  48800 + 100 + 1200 = 50100 points. Edit: b2b 75150

But we have to take into consideration that in fact you don't have to do 4tsd and then a tetris, by stacking high and not doing a tetris one might be able to not do the last tetris at all but do 2tsd instead.

As stated before the single added in the end would of course ruin b2b and hence it would be ideal to clear a single at the start and then start stacking. This would shift the whole setup one set ahead which would also lead to that some line clears that were in-between two levels ended up above overall increasing the score.

This is a way to start the game so that you not only do you gain that single I was mentioning earlier but you also start a b2b so that your first tsd gives 1200 instead of 800.
[fumen]110@7eKBfIihBLfB+gB9jBncBiHPIAo3MoDFbUwEOQPAAL?aBliBNYB[/fumen]
[/spoiler]

Well this is all good in theory (i hope) but in reality the game comes close to 20G in the end which makes some setups difficult to perform. ST stacking however works 'well' in 20G as you can oftentimes stack so that you have a high middle. Also we have to take into account how easily these are achieved. Both ST and the infinite TST is both very doable, ST being the easier one. A freeform tst with tsz however is more difficult to pull of so I don't expect it to be useful until someone invites a system.


Combos together with Tszeros would be very interesting though

[Rosti_LFC]

Seems the optimal strategy is to string only triples until the 20G makes that impossible (because looking at some of the placements required, I think it is impossible) and to then go to ST stacking from then on.

[Alexsweden]

yepp

[perfectclear]

What would the score be if one were to stack a side four wide using as many t spin 0's as possible in stacking and as many t spin singles as possible in downstacking?

[Alexsweden]

A..... B ... C   ...D........E
1   100   100   20400   136
2   150   250   24150   161
3   200   450   27900   186
4   250   700   31650   211
5   300   1000   35400   236
6   350   1350   39150   261
7   400   1750   42900   286
8   450   2200   46650   311
9   500   2700   50400   336
10   550   3250   54150   361
11   600   3850   57900   386
12   650   4500   61650   411
13   700   5200   65400   436
14   750   5950   69150   461
15   800   6750   72900   486
16   850   7600   76650   511
17   900   8500   80400   536
18   950   9450   84150   561
19   1000   10450   87900   586
20   1050   11500   91650   611
21   1100   12600   95400   636
22   1150   13750   99150   661



A is the amount of combo
B is the points for that particular combo
C is the sum of all combos up until and including the current combo
D is the final sum - C * 150 / A + (54*100). The 5400 points come from perfect usage of T-spin zeroes.
Edit: added - E is points per line

Important all values are inflated compared to the other as combos are never affected by b2b while tspins are so since i did not include b2b in my claculations before they are much lower than they should be. Not including b2b is ok when comparing the other methods against eachother but when comparing with combo it should be included.

Below you have non important stuff really, did it for my own sake mostly
[spoiler]Keep in mind that some values are a bit above the perfect value as 150 / A (number of times a combo of 'A' magnitude can be done in 150 lines) in many cases gives a non integer number and so gives you a .xx value of the C value and as the value of C grows exponential the value is higher than it actually should be.

E.g a combo of 20 can be done 7.5 times. A full combo gives 11 500 points but a 0.5 combo - which is 10 gives only 3250 which is only 0.28 of the full combo. But the calculation gives you 11 500 * 0.5 which is 5755 and so the value is 2505points too high.[/spoiler]

So we can see that 4w is amazingly good, to good to be true? - did I do something wrong?

Of course I am assuming that this is the real values for the combo, havent checked it all yet. It is true for the first couple of values though.

[Kitaru]

Seems the optimal strategy is to string only triples until the 20G makes that impossible (because looking at some of the placements required, I think it is impossible) and to then go to ST stacking from then on.

TSS, TSD, and TST are all worth the same points per line, so it seriously doesn't matter. If EZ TSS or TS0 sets up b2b, then those are sensible openers, but all that matters is sustainability and a TST with 1 line left in the game.

So we can see that 4w is amazingly good, to good to be true? - did I do something wrong?

Well, you didn't show us the resulting value in points per line. A successful 20~22 combo would be 575~625 points per line for an exceedingly high amount of effort per execution compared to a sustainable t-spin setup like ST, which awards 600 points per line for each t-spin and 300 points per line for the occasional tetris to drop the stack height. The way things average out, yes, perhaps continuous 22 combos is truly optimal, but only if you have the TAS-like efficiency to execute it perfectly every time and not sacrifice any drop points. Remember that you get ~50 points for hard dropping a piece all the way from the top to the bottom of an empty well.