Timing your wins and losses?

[thone]

Hello everybody,

i'm currious about something. Do you get advantage by postponing your wins and rush your losses?
I made a little calculation and this is what i get.

Situation:
- i'm playing vs 2 players with equal points.
- I'm having 1000 points.
- I'm winning one, and losing one.


Situation 1: First win, then lose
Start total: 1000 points
Winning of somebody with 1000 points: + 20
New total: 1020 points
Losing of somebody with 1000 points: - 22 (because i have more points, i lose more)
Final total: 998 points


Situation 2: First lose, then win
Start total: 1000 points
Losing of somebody with 1000 points: - 20
New total: 980 points
Winning of somebody with 1000 points: + 22 (because i have less points, i win more)
Final total: 1002 points





Do i see this correct?
Can you really gain extra points by using this?
Or did i miss something?

 

[ndrm31]

What about frienly games where no one win nor loses, points sometimes I play trivia games in fb and u have that option

 

[Geoffrey]

I remember someone actually told me that he'd just be delaying his own win since he was about to win a game and wanted the maximum number of points out of it, so I skipped my turn completely to not allow him to control the time, as I myself am a powerful dictator who controls all time. He ended up not coming back online and I ended up winning - it felt like karma to me

Him delaying it would have saved me a few points and he probably saw it that way, but instead I got a victory!!

 

[Dalinar]

What about frienly games where no one win nor loses, points sometimes I play trivia games in fb and u have that option

There is an option for that, you just have to both agree to AWOL in advance once you decide a winner. I'd like to use a less.. kludgy way, but it works. I think there was also a fear that given a formal option to take points out of things, people like me might just decide to sit on the sidelines and never bet any points.

I'll admit to having consciously calculated which order I should take my wins in to get the most points. I don't think I've ever delayed a turn when I otherwise would have won to do it though. I also remember taking another win first so I'd take less points from a friend..

On a related tangent, I think calculating this stuff and doing betting part of choosing one's games on here well is just part of getting up the scoreboard. I actually wish it wasn't, but at least it does fit nicely with all the probability involved in-game.

 

[riskyone]

rob6483 and Cards both made good points here.

 

[BeanZ]

If I was going to change anything I'd change the formula for points awarded. I'd chance the * factor from 30 to perhaps 20. Makes it so that the 100 point catastrophic losses aren't as likely. Maybe that way more high ranking players would be willing to play lower ranks.

I think it's a good idea to try to limit the maximum points lost but in that scenario it would only make losing to lower ranks relatively worse. You might not lose the 100 point cap but the cap is there to limit how many points you lose... you'd get less points for a win and still be able to (if you were high enough) get close to the 100 point cap.

I think a better idea would be to just lower the cap to maybe 80 or something like that (although I know 100 is such a nice number)

 

[Geoffrey]

Or what about a set amount of points for a win? I think I've seen this discussed on MC before, not sure if it has on this thread, but if people are worried about losing a high number of points to low ranked players, perhaps games could be run like a tournament and have a certain amount to buy into it? Not sure how easy that would be to add as an extra setting to those who want it!

 

[goweje]

Couldn't that be pretty badly abused though? Say my score is getting down there and I crack into the kitchen patrol. So... I join 100 games against high level opponents - now all 100 games I'll get huge piles of points for winning and any I lose would be hardly nothing lost if points are based off of score going in, so if I win a fraction of those games I'd still jump up through the ranks like crazy.

Not saying that I love the current system since it sucks when your opponents were majors at the start of the game and are busted down to lieutenant by the end dropping your point winnings. Thus the problem with "timing wins/losses".

I fail to see how this is more open to abuse than the current system. If, you've lost a bunch of games your score should go down, as it reflects your diminished performance. I doubt that many players would want to take the certain loss of points up front in order to reap an uncertain bounty of victories against higher rated players in the future. In fact, if you read through other threads on the forum, we have a problem getting higher rated players to play games against lower rated problems. For instance, I must win about 81% of games against 1200 rated players, just to break even and I'm not a highly rated player. So, employing a tanking strategy would likely severely reduce your access to higher rated players.

It's not a huge deal, but I think it makes sense to calculate points based on start of game scores. As a player chooses games he/she calculates expected returns based on opponent scores at that time. I'd love to know if some people are crazy enough to factor in the timing of opponent wins & losses when choosing games.

Round numbers here, but in a recent tournament game, another player's score had temporarily dipped and was in the neighborhood of mine ~2300. I won the game, but as it transpired his score jumped back to where it is today ~3400. I probably picked up 15 extra points because of that player's success. It doesn't make sense to me that I was entitled to the points he's won in the interim.

Some downsides I can think of are increased opacity. At the end of the game you can necessarily back into the points you received. But, I think it has a side benefit of minimizing some of the score management techniques people have mentioned in this thread. I think we're always going to have people timing things if they're close to a milestone rank and that's fine, but I think it would discourage people who may be looking to time games all the time.

The bigger problem though, is longer games. I'm think 12 person flat rate for instance, where games can last for months. In such cases, a players beginning score may not accurately reflect his skill level and performance over the duration of the game.

 

[LordAdef]

Rob makes a valid point. Basically I guess both could be abused. I just don't see a pressing need to change the way it's done.

If I was going to change anything I'd change the formula for points awarded. I'd chance the * factor from 30 to perhaps 20. Makes it so that the 100 point catastrophic losses aren't as likely. Maybe that way more high ranking players would be willing to play lower ranks.

Maybe there could be an option for that. Game formula: 10, 20, 30

10 would maybe be closer to a light game (no points at stake) as one suggested in another thread.

 

[Cardinalsrule]

I think it's a good idea to try to limit the maximum points lost but in that scenario it would only make losing to lower ranks relatively worse. You might not lose the 100 point cap but the cap is there to limit how many points you lose... you'd get less points for a win and still be able to (if you were high enough) get close to the 100 point cap.

I think a better idea would be to just lower the cap to maybe 80 or something like that (although I know 100 is such a nice number)

I wasn't suggesting removing the cap. Just lowering the multiplier from 30 to 20.

 

[BeanZ]

I wasn't suggesting removing the cap. Just lowering the multiplier from 30 to 20.

Yeah I know, but by lowering the multiplier and keeping the cap the same it's just effectively increasing the relative points a higher ranked person can lose so it's not going to be the a great way to get people who are holding onto their high ranks to play more games

 

[goweje]

The 30 multiplier was probably chosen somewhat arbitrarily. But, much like baseball, records matter. At this point, you dance with who ya brung. As such, I'm opposed to multiple multiplier options, a formal setting for no points games, 2nd accounts etc.

Also, I don't understand the obsession with the cap. Am I doing the math wrong, or does this apply to perhaps a dozen players at most? If a 4000 loses to a 1200, that's exactly a 100 point hit right?

 

[goweje]

We're getting somewhat off topic here. But, my two cents on the multiplier is it's fairly meaningless. It obviously impacts the absolute number of points won or lost, but decreasing it to 20 or any other number has zero impact on the underlying math on a relative basis. Your not going to find many higher rated guys playing classic evolved real time games whether the multiplier is 30, 20 or 1. As an example, a 4000 has to beat a 1200 92% of the time to break-even, regardless of multiplier. You drop the multiplier to 20, and losing 50 points becomes so "catastrophic." you'll have to spend the next month bludgeoning new joiners in 4 v 4 games with your high ranked buddies just to get the points back.

The problem is with the defeated/winner portion of the calculation. Ideally, you'd use a normalization process to reflect the players performance above/below site/map averages. I read elsewhere on this forum that the average player on this site's score is slightly above 1500. With the current calculation, a 4000 rated player would have to beat the average player 88% of the time just to tread water. Given the 1st turn advantage, the drop, double ice that's a win rate that's not possible and encourages higher ranked players to avoid games they'd otherwise consider joining.

 

[Dalinar]

I favor lowering the multiplier or increasing the average number of points, they'd have the same effects. In short, because math (detailed below). Increasing the average number of points by just 750 to everyone's account and making the base 2000 would probably also be more palatable to everyone than changing the multiplier and adjusting the scoreboard.

(you can skip this part Name )
-------------Technical Part (you can get the point without math)-------------------------------------------------
The scores right now have a a mean somewhere around 1500 depending on how many people come in, lose, and quit. People wager about 30/1500 = 2% on a game between even ranked players. However, they can gain 8% in a single match against someone at 4500 or above (it's a short list, I know, but it's still a lot for anyone above 3k).

The change from lowering the multiplier from 20 to 30 would come in because scores would then cluster quite a bit closer 1250 - the top ranked people (other than Jaejae who I believe more or less escaped the point system by playing only tournaments) would then be at something like 3200 rather than 4500.

This makes playing people who are closer to average a lot less painful. To make the math easier, let's say you're playing a 1600 player, under the old system I'm around 4000 and lose 75 points with the 30 multipier. If my score were 3200 under a new system (similar rank)and the multiplier were 20. I'd lose closer to around 40. Scaling to the smaller number of points, the 40 would be a bigger deal than it is now, but it still isn't as bad as the 75.

It has to do because the of the way the win rate grows as the score differences grow.
1500 v 1500 you need a 50% win rate to stay even
1500 v 2250 you need a 69% win rate to stay even
1500 v 3000 you need a 80% win rate to stay even

That part wouldn't actually change, but the narrowing the distribution of scores would make it a lot better.
It would also make it harder for people to fall as far, which is where you get the really crazy multipliers (a lot of people who hit something 900 are having a bad run and wouldn't fall as far if the multiplier to score ratio were lower).
-------------Technical Part over------------------------------------------------------------------------


Basically, how the score system works is heavily dependent on the relative sizes of the win multiplier (currently 30) and the average score (1505 for actives per Sheriff). I think one way of fixing the current system is decrease the win multiplier relative to the current score - the relationship is the key though, (win_multiplier / average_score).

It also means you fix it by giving everyone 750 points and making the starting score 2000. That might also go over better with everyone, since if you cut the win multiplier, you'd also have to cut people's score or just restart the board. A score addition just seems nice, even if everyone's getting it.

Lowering the cap wouldn't likely help very much because a major factor is how few points you win for the majority of the games. The cap imposes a maximum break even win rate - right now it's 91%. At 80 it would be 89%, not too helpful.

 

[Dalinar]

We're getting somewhat off topic here. But, my two cents on the multiplier is it's fairly meaningless. It obviously impacts the absolute number of points won or lost, but decreasing it to 20 or any other number has zero impact on the underlying math on a relative basis. Your not going to find many higher rated guys playing classic evolved real time games whether the multiplier is 30, 20 or 1. As an example, a 4000 has to beat a 1200 92% of the time to break-even, regardless of multiplier. You drop the multiplier to 20, and losing 50 points becomes so "catastrophic." you'll have to spend the next month bludgeoning new joiners in 4 v 4 games with your high ranked buddies just to get the points back.

The problem is with the defeated/winner portion of the calculation. Ideally, you'd use a normalization process to reflect the players performance above/below site/map averages. I read elsewhere on this forum that the average player on this site's score is slightly above 1500. With the current calculation, a 4000 rated player would have to beat the average player 88% of the time just to tread water. Given the 1st turn advantage, the drop, double ice that's a win rate that's not possible and encourages higher ranked players to avoid games they'd otherwise consider joining.

I agree that scrapping the current system and replacing it with something more sensible and appropriate to the actual realities of the game would be the best way to go. I'm not sure what system I'd suggest off the top of my head, but there are a lot of good ones out there to take ideas from. As gtivan has pointed out elsewhere, we have a lot of data now so we could determine how often players win going 1st and could adjust for that, etc.
...However, the current system seems pretty entrenched so I'd rather the tinkering possible be done.

Given constant scores, the win multiplier is meaningless - that you're right about. However, if lowering it condenses the score distribution, it'd help. But as I said in the way too long post above though, we'd probably just be better off giving everyone a lot of points for the same effect.

 

[ndrm31]

Lately Im finding players that delay in purpose they turns, with no aparently reasson (wining steaks, points or an argue) even when you see them online on the site, I know that there is no that much to do but asking them, any toughts or suggestion?

 

[Redstorm]

perhaps a good reason to incorporate a 12 hour game option? I'd like that.

 

[haWD96lz]

I sometimes play some easy turns where I know what I want to do when I have only a few minutes in the morning and save the thought provoking turns for lunch or after work. Or sometimes I have 5-10 lined up and only time to get through 3.

I've only delayed losing twice. Once when I was at a record score and once when I was on a record winning streak. By delay I mean one or two turns were played in the evening when normally I would have played them in the morning so maybe a 15 hour delay. Maybe this makes me a bad person, but there are incentives here.

I have thought about the points issue people have pointed out and I agree any small gains from losing fast and winning later are not worth the effort and the value will degrade over time and the sustained effort for a few points is not worth it. So this says something good about the scoring system.

 

[linkinpark]

On Dalinars math: I appreciate your effort but it's not true. You ignore the fact that, if you raise the base score, 6000 points won't have the same meaning compared to present 6000 points. How good you are has to be calculated as a ratio like (score/base score) or (score/avg. score). Because if the base/average points is doubled, it's also equally easy to achieve 8000 points compared to 4000 points previously. Scaling is just cosmetic change. So it doesn't matter what the base is. Lowering the multiplier also doesn't change anything about it, as goweje already pointed out. The only way to change it, is by dropping the idea of the multiplier being a constant number. For example 30 points if both players have equal points but only 45 if player 1 has double points compared to player 2. If that is applied, the avg. ratio of half point players to beat compared to avg. number of equal point players to beat comes closer together.
I hope it's not too confusing, sorry for my English

 

[Redstorm]

whew, the math wizards!!!! No wonder I cant beat u guys LOL

 

[riskyone]

I sometimes play some easy turns where I know what I want to do when I have only a few minutes in the morning and save the thought provoking turns for lunch or after work. Or sometimes I have 5-10 lined up and only time to get through 3.

I've only delayed losing twice. Once when I was at a record score and once when I was on a record winning streak. By delay I mean one or two turns were played in the evening when normally I would have played them in the morning so maybe a 15 hour delay. Maybe this makes me a bad person, but there are incentives here.

I have thought about the points issue people have pointed out and I agree any small gains from losing fast and winning later are not worth the effort and the value will degrade over time and the sustained effort for a few points is not worth it. So this says something good about the scoring system.

It doesn't make you a bad person. Anybody who can extend their longest streak by putting your turn off a few hours. That's not purposely skipping turns.