|
Hm, i have a little noob question again, this time about the stuff spoken about in the above posts.
If i understand correctly, shanten is the number of tiles you need before you are in tenpai and uke-ire gives the amount of tiles that would reduce shanten by one.
Am i right then that there is a connection (though not necessarily one on one) between tile efficiency and uke-ire? I mean, if i discard correctly following tile efficiency rules, i should theoretically end up with a higher uke-ire than when i don't, right?
P.s. what is betaori? from blogs it looked like the way to play defensive against an opponent who're in riichi. Is this correct?
|
Hehe, I sure like using the Japanese words, don't I? The english equivalents are usually so... long and clunky.
Yes, your understanding of Shanten and Uke-Ire seems to be correct.
Tile Efficiency is pretty much all about Uke-Ire, at least if you're only considering closed hands. A ryanmen shape like 23 can be upgraded to a shuntsu (sequence, 123 or 234) with up to 8 tiles (minus the ones already visible somewhere). A kanchan like 24 only has up to 4 tiles to wait on. That's why you read in those tile efficiency tutorials that ryanmen > kanchan.
Of course pure Uke-Ire ignores all win conditions, dangerous tiles, and the value of your hand. But since finishing first is the only way to win, and finishing first is also the best defense, it's very important to be able to play efficiently.
Betaori means to give up your hand. You only try to prevent dealing into someone else's hand.
If you are very unlikely to finish a hand (say you got dealt something like 147m258p369sESWN, can hardly get any worse than that - a whopping 6 shanten towards Chiitoitsu) but still stack up all the good tiles (34567), maybe call a bunch of times hoping you'd somehow make a hand, then you will end up in a very bad spot once someone calls Riichi and you are still 2 or 3 Shanten. Not only will your hand likely not have any value, your tiles might all be very dangerous.
In such a case you might want to hold back honors to prevent people from getting a triplet, and also stack up on safe tiles against all enemies early on.
|
New versions:
LogGrabber: Configuration now only necessary once.
http://www.mediafire.com/?ef9av0m4d17nyh2
LogGrabber now supports the file format of the file the Tenhou client uses to store replay information. If you set the input file to the actual file used by Tenhou, you can simply start up LogGrabber every 20-40 games and grab the logs without any extra effort. Tenhou updates the file, LogGrabber remembers the location of the file and only downloads the new replays.
For me the file is located at:
C:\Users\myname\AppData\Roaming\Macromedia\Flash Player\#SharedObjects\C8WD8NJX\mjv.jp\mjinfo.sol
Analyzer: Now with custom hand analyzer.
http://www.mediafire.com/?eequtr1kv22s4y6
![[image loading]](http://i.imgur.com/1KrWU.jpg)
Click a tile in the bottom left to add it to the hand, click a tile in the hand to remove it.
Displays Shanten and Uke-Ire for the hand in the bottom right.
If you use less than 13 or 14 tiles, an unspecified called meld is assumed.
A hand with a multiple of 3 tiles won't display any information because such hands are not possible in Mahjong.
Now also with customizable Background for the Replay viewer:
+ Show Spoiler +![[image loading]](http://i.imgur.com/uA0YC.jpg) I included this image as images/mami.jpg, but the default is a boring monocolored background.
To change the background, simply go to /data/settings.xml and change the name of the image (which you should place in the images folder of course). In case your system opens .xml files in Internet Explorer or something like that, you can just open it in any simple text editor as well.
Improved Shanten counting algorithm. Still not as fast as I want it to be, but with less bugs now.
Shanten = 8 - 2 * melds - 1 * taatsu is still about right, but I also had to take care of some odd cases like this:
1111m123456789p
This hand has 4 melds and a single tile: 1m. Now obviously you can't make the 1m into a pair, so the hand is not Tenpai but 1 Shanten. Sure, if you call a Kan, you get a new tile, but that's still a draw and you might even have an enemy win on that call. And if your hand looks like this:
1m111p called: 123m 123m 123m
then you can't even save your tenpai by calling a Kan anyways.
As always: Tell me about any bugs you find and any suggestions you have.
Also if you abuse the hand analyzer while playing on Tenhou, don't come bragging about ranking up while using it!
I don't think it's too bad if you use it to practice, especially at the lower ranks. But if you play in the upper dan room, I suppose it becomes kind of morally questionable...
|
damnit, quote instead of edit.
|
I wouldn't be too worried about upperdan players. ukeire is not as important at this level since we (at least I) play around other peoples discards. Always discarding tiles that give the best ukeire may be efficient in getting to tenpai, but it's not always the best when it comes down to your EV. I don't particularly want to delve too much into this but you'll definitely understand once you hit the upper dans.
BTW whats your rank/rating atm spines. It sound like you understand enough to be able to beat the lower dans soon.
|
On October 25 2012 15:21 Rhaegar99 wrote:
I wouldn't be too worried about upperdan players. ukeire is not as important at this level since we (at least I) play around other peoples discards. Always discarding tiles that give the best ukeire may be efficient in getting to tenpai, but it's not always the best when it comes down to your EV. I don't particularly want to delve too much into this but you'll definitely understand once you hit the upper dans.
BTW whats your rank/rating atm spines. It sound like you understand enough to be able to beat the lower dans soon.
Sure, I know there's way more to this game than than just something as simple as tile efficiency. But I also don't care a whole lot about people getting a bit of help at the lower ranks.
I'm currently 2nd Dan, fell down from 3rd. Rating was up to 1700 or so before, now below 1600. I have no idea what happened, but when I was 2nd Dan before, it was pretty easy to get points, but ever since I hit 3rd Dan and won like 2 games, it's been going hilariously bad for me. Even back at 2nd Dan it feels like either I got way worse or my enemies got way better :p
Hell, I even fell below an average placing of 2.5, when I was better than 2.4 when I hit 3rd Dan TT
By the way, I would LOVE you to go into detail on how you play around discards. Or point me to some resources. I'm out of English resources to read...
|
I had that happen too haha
I got 3 dan, then I dropped down to 2 Dan, then dropped down to 1 dan, then dropped down to 1Kyu...
I was probably playing on tilt, this happened all in one day lol
I ended up making a new account, and it's at 2 Dan now
|
On October 25 2012 21:07 JSH wrote:
I had that happen too haha
I got 3 dan, then I dropped down to 2 Dan, then dropped down to 1 dan, then dropped down to 1Kyu...
I was probably playing on tilt, this happened all in one day lol
I ended up making a new account, and it's at 2 Dan now
In a single day? That's fast. Let's say you only had to lose 1 more game to drop to 2 Dan. Then you needed like 9 straight 4th places to drop to 1 Dan? And then another 10 or so for 1 Kyu?
|
;_;
I was playing bad, and like, I wanted to get more wins and it kind of spiraled down
I should've stopped :<
Hmm maybe I didn't drop all the way down to 1kyu then lol
Maybe I stopped at 1 Dan
but nevertheless, it was sad experience :<
|
![[image loading]](http://i.imgur.com/GlW10.png)
lucky ippatsu + haitei raoyue to take first place on the last round.
I believe that I'm back in form now. Gained about 700 points over the past ~30 games so I'm back to even points for my rank. I'll probably give you guys a few tips about when to push and when not to when I feel really confident about my game. I feel that I've learnt a lot after experiencing that massive downswing.
For you guys on a downswing as well, try to play a bit less aggressive. I feel over-aggressiveness was probably the reason why I dropped x.x
|
Overaggression? I kind of doubt it. After all I never get to Tsumo these 12000 point hands before anyone else gets to Tenpai like my opponents like to do... So if I play even more defensively I'm just gonna lose because everyone Tsumos and I don't get to win anything.
Here's why I'm losing:
![[image loading]](http://i.imgur.com/PnBU1.jpg)
Well ok, I didn't lose that hand. But what the hell is that? It's scary. I feel like I'm being preyed upon.
Also next round I was playing around a Riichi just to deal into TWO 3 or 4 Han damaten hands just a few turns before the hand would end in a draw... Lately this stuff happens so much that I just burst out in laughter...
So I collected some statistics from my replays:
![[image loading]](http://i.imgur.com/zcyPX.png)
Rank and Rating show the highest values among all the replays I have of that player.
Ukeire draw ratio is the ratio between how often a player draws one of his Ukeire and the number of draws he made total.
Ukeire draw chance is the ratio between how many Ukeire tiles are still invisible to the player when he draws and the total number of invisible tiles at that point.
So the former is how often the player actually has a draw that advances his shanten, while the latter is the theoretical chance for such a draw based on how many tiles he waits on and how many tiles he doesn't see.
Obviously it's better to have high values for both.
Now I am very close to the average (1/4th of the average is from me though), many people I played more than once are close to it too.
But there is an outlier. いかさましです, 1 Kyu, best rating 622.83. Massively higher ratio and chance. I mean, seriously, 43% chance to draw an Ukeire seems ridiculous. There are other people I played against with similar ratios, but they have a much smaller sample size. 35 Games however is at least several hundreds of draws, it can't just be a streak of luck. Is this a pro on a new account? And what does she do that makes her draws that much better than the average?
|
Seems more like a bug maybe? Probably not. If youve got 35 games of him and hes drawing at such ridiculous rate, he would be in the dans with a much higher rating (how do you even get as low as 600 lol). Otherwise, seeing how rating is that low, maybe they are like kokushi bots where they hands are always crap and most draws would increase their shantan. Should maybe check how they play.
Correct me if I'm wrong but shouldnt draw/chance ratio be a lot closer to 1 when you already have so many games on yourself? Its a bit strange how everyone is out drawing their tiles more often then their actual chance.
Seems like tenhou has finally recognized the firefox bug
+ Show Spoiler +※Flash版では、FireFoxの一部のバージョンで意図しないツモ切りが発生する
という報告があるため他のブラウザの使用を推奨しています。
※ Flash version is cut Tsumo unintentional in some versions of FireFox occurs. We recommend the use of other browsers because it is reported that.
|
On November 04 2012 21:14 Rhaegar99 wrote:Seems more like a bug maybe? Probably not. If youve got 35 games of him and hes drawing at such ridiculous rate, he would be in the dans with a much higher rating (how do you even get as low as 600 lol). Otherwise, seeing how rating is that low, maybe they are like kokushi bots where they hands are always crap and most draws would increase their shantan. Should maybe check how they play. Correct me if I'm wrong but shouldnt draw/chance ratio be a lot closer to 1 when you already have so many games on yourself? Its a bit strange how everyone is out drawing their tiles more often then their actual chance. Seems like tenhou has finally recognized the firefox bug + Show Spoiler +※Flash版では、FireFoxの一部のバージョンで意図しないツモ切りが発生する
という報告があるため他のブラウザの使用を推奨しています。
※ Flash version is cut Tsumo unintentional in some versions of FireFox occurs. We recommend the use of other browsers because it is reported that.
No, I definitely expect a draw/chance ratio higher than 1. For example you can get some read on the tiles your opponents have in hand (and try to move your Ukeire away from the tiles you expect them to have), while they are considered invisible tiles for the chance calculation. And I'm pretty sure that your actual chance increases if several players including you are stockpiling tiles of a certain suit, everyone a different suit. Triplets might also have some influence. I can't give you a solid mathematical model for this though, it's just my gut feeling and I know that gut feeling and probabilities usually don't go hand in hand...
I have a lot of people with 10+ games (usually 1 hanchan only though) with a ratio greater than 1.4, someone with 9 games even cracked 2.0 (1902.7 rating). She had an extremely low chance (less than 15%) but drew Ukeire at almost 30% ratio.
Only about 1/8th of the people had a ratio below 1.0. The lowest is a 5 Dan with slightly above 0.7. Of the highly rated people with high Ukeire draw ratios I have one each of 7, 6 and 4 Dan with draw ratios above 30%.
Hell yeah, all that ratio, chance, ratio-chance-ratio crap... Can't find any better names though damnit.
Btw I didn't include it in the picture, but the average Shanten of a starting hand seems to be ~3.6, with a big focus on 3 and 4 Shanten hands.
Well, I'll see if I can dig up a replay of that 600 rating girl...
HAHAHA, ok, here's the deal with いかさましです: she disconnected. In all 3 hanchans. From the very start. What the hell is that even. The guy with the second highest ratio disconnected for the last 2 out 7 hands of a hanchan. Despite his godlike draws he fell down to 900 points.
The 4th highest, a 2nd Dan at 1575 rating played legit through 10 hands, but the average Shanten of his starting hands was 4.4. Similar with many other players at the top of the list.
So it turns out your Ukeire draw ratio increases if you either disconnect from the game or have horrible starting hands. I guess that makes sense. The draw/chance ratio however still leaves some questions open... I'm not sure why it is so high for people who disconnected (another 1 Kyu who disconnected, named lulu, hits 1.6). Maybe I have a bug here... But I'm just dividing one value by the other, so one of the vaules must be wrong if it's a bug... And of course there is a whole bunch of legit players with high ratios too.
Maybe I should also calculate the average shanten of a player's hand after each of his discards, that might help weed out the people who stay at high shanten hands.
|
On November 04 2012 22:50 spinesheath wrote:
No, I definitely expect a draw/chance ratio higher than 1. For example you can get some read on the tiles your opponents have in hand (and try to move your Ukeire away from the tiles you expect them to have), while they are considered invisible tiles for the chance calculation. And I'm pretty sure that your actual chance increases if several players including you are stockpiling tiles of a certain suit, everyone a different suit. Triplets might also have some influence. I can't give you a solid mathematical model for this though, it's just my gut feeling and I know that gut feeling and probabilities usually don't go hand in hand...
Hmm I dont quite understand that part of your explanation. I reread the whole thing a few times and think I understand a bit better now and my question before was a bit silly. But now I think there might be another problem (or something I dont understand lol). Shouldnt your theoretical draw chance be greater than your actual draw ratio? Since well... your opponents are going to hold some of your ukeire tiles. It dosnt seem right to be the other way round.
Or is it the fact that your opponents hold more non ukeire than ukeire tiles that it flips the chances the other way.
|
On November 05 2012 02:00 Rhaegar99 wrote:Show nested quote +On November 04 2012 22:50 spinesheath wrote:
No, I definitely expect a draw/chance ratio higher than 1. For example you can get some read on the tiles your opponents have in hand (and try to move your Ukeire away from the tiles you expect them to have), while they are considered invisible tiles for the chance calculation. And I'm pretty sure that your actual chance increases if several players including you are stockpiling tiles of a certain suit, everyone a different suit. Triplets might also have some influence. I can't give you a solid mathematical model for this though, it's just my gut feeling and I know that gut feeling and probabilities usually don't go hand in hand...
Hmm I dont quite understand that part of your explanation. I reread the whole thing a few times and think I understand a bit better now and my question before was a bit silly. But now I think there might be another problem (or something I dont understand lol). Shouldnt your theoretical draw chance be greater than your actual draw ratio? Since well... your opponents are going to hold some of your ukeire tiles. It dosnt seem right to be the other way round.
Here's how I calculate those values, correct me if I made any mistakes in my model.
The tiles your opponents hold in the concealed part of their hands are considered invisible tiles. The face down part of the wall is also considered invisible.
The value ukeireCount is the number of invisible tiles that could be used to improve the shanten of your hand.
Draws are only draws from the wall, tsumo or rinshan tsumo. No calls. Of course I am only looking at the draws of a single player.
The draw ratio is calculated like this:
ratio = ukeireDraws / totalDraws
The draw chance:
chance = ukeireCount / invisibleTiles
For each draw a player makes, the values (draws, totalDraws, ukeire, invis) are summed up and then the final chance/ratio is calculated.
For the chance, I can't reliably (without trying to read my opponent's hands) tell whether a tile is in the wall of an opponent's hand, so I have to assume that I draw one out of all the invisible tiles. I have ukeireCount favorable tiles. My opponents can't magically draw all my ukeire into their hands, so for calcualting the chance it doesn't matter at all, the tiles are distributed evenly.
The ratio calculation should be straightforward, it's just the results.
Now I as a player can try to read an opponent's hand and figure out what tiles he has. Those tiles are less likely to be in the wall, since I assume them to be in his hand. They also use up space in his hand, so the chance for another tile to be in his hand rather than the wall decreases. I can try to adjust my ukeire in a way that allows me to wait on the tiles in the wall rather than those in an opponent's hand. This improves my chance to draw one of them assuming my read was correct.
If my opponents try to stock up on my ukeire (assuming they read my hand and know what I am waiting on), then they still only draw tiles randomly. That shouldn't affect my chance nor my ratio. They can't increase their chance to draw my ukeire unless they manage to increase my own ukeireCount which is rather unlikely and hardly useful for my opponents.
|
|
|
I want to play greatly, but the only person in my area that I really knew that played has stopped >:
I'm not the type that learns well just from reading, easier for me when I'm learning firsthand with some explaining. Makes it pretty hard to get into this game. Any tips?
|
On November 13 2012 22:10 Cirn9 wrote:
I want to play greatly, but the only person in my area that I really knew that played has stopped >:
I'm not the type that learns well just from reading, easier for me when I'm learning firsthand with some explaining. Makes it pretty hard to get into this game. Any tips?
So, have you checked out tenhou.net? If not, I guess you should try. Here's some info in case you don't know your way around the site (since it's in Japanese):
http://arcturus.su/tenhou/
If you jump right into it, you might end up losing a lot, but it's not like it'll do you any harm.
In this post I listed a bunch of info for a new player:
http://www.teamliquid.net/forum/viewmessage.php?topic_id=111078¤tpage=16#309
|
On November 16 2012 06:22 spinesheath wrote:Show nested quote +On November 13 2012 22:10 Cirn9 wrote:
I want to play greatly, but the only person in my area that I really knew that played has stopped >:
I'm not the type that learns well just from reading, easier for me when I'm learning firsthand with some explaining. Makes it pretty hard to get into this game. Any tips? So, have you checked out tenhou.net? If not, I guess you should try. Here's some info in case you don't know your way around the site (since it's in Japanese): http://arcturus.su/tenhou/If you jump right into it, you might end up losing a lot, but it's not like it'll do you any harm. In this post I listed a bunch of info for a new player: http://www.teamliquid.net/forum/viewmessage.php?topic_id=111078¤tpage=16#309
I have, but its intimidating. Anyone want to hang out on skype and play/explain things to a complete nub? D:
|
Sadly the replay viewer doesn't work for me, I'm on winXP. LogGrabber seems to work fine. Do I need anything like the .NET frameworks or other stuff I might not have? Anyone else using it?
|
|
|
|
|
|
EPT
