Jump to content
You need to play a total of 20 battles to post in this section.
Horsie_ships

Statistically, Average Battles Needed to Rank Out by WR

35 comments in this topic

Recommended Posts

23
[WHS]
Members
37 posts
6,215 battles

With a python script, I can simulate a singular ranked sprint, taking into account irrevocables, star saving, and win rate. This can help predict an approximate number of battles to either rank out or get N wins. Of course, due to randomness, different simulations will often have different results. To compensate for these differences, I ran each simulation 100,000 times and took the average of all these simulations.

Here's the data, visualized:

WoWsRanked.thumb.png.09a7defef684f440dff8912cd876e875.png

The average number of battles it will take you to rank out, by WR

Note that these statistics are averages of the simulations, meaning that although it's an approximation, you can still play more or less games by chance. The chart also assumes a fixed Star Saving chance of 1/7, because there are 7 players per team. In actuality this chart would be steeper, because players that win more games will also be more likely to save their star, and vise versa, but I've kept it constant since I'm not sure how I would accurately scale it with win rate.

Here's the same data, visualized on a battles per day basis:

WoWsRanked2.thumb.png.ac05f1e6e13e5380dec26c053fa05d05.png

Same data as before, but divided by 14 because there are 14 days per sprint

What about qualifiers? Here's the data for qualifiers:

WoWsRanked3.thumb.png.699723942955476e7c02aff7e2b74bbc.png 

Five stars are needed to qualify into silver, and seven stars to qualify into gold

What about the entire season? Here's some more data:

WoWsRanked4.thumb.png.645c67dcbff059a336a4ce74144bf16a.png

The average total number of battles played the entire season to reach those goals

While some players may choose to stay in bronze and some in silver, I've decided to take a look at "Doubloon Farming" and "Maximum Rewards." Doubloon farming means to qualify into gold as soon as possible, then getting the 29 wins per sprint until the end of the season, and maximum rewards means to qualify into gold as soon as possible and rank out every gold sprint. As a clarification, while ranking out of silver does yield the same amount of doubloons as 29 wins in gold, it is statistically harder to rank out of silver than get 29 gold wins.

Keep in mind that while this graph suggests that a 40% WR player can still doubloon farm (although extremely time consuming), they will still need to push pass the near impossible (for them) silver rank out + gold qualification.

Again, all the data is an approximation, and does not mean you will be guaranteed to accomplish your goal if you play at a pace of your approximated number of battles. Take what you will from the numbers to help you decide what is and what isn't feasible.

Edit: Here's some more graphs, with "bounds." The bounds represent the position of 5th percentile and 95th percentile simulations. Basically, you can reasonably expect to be within the bounds, but you could still be very unlucky or very lucky.

WoWsRanked5.thumb.png.673290e38e0205a5b9e15c805fe34e51.png

WoWsRanked6.thumb.png.ddd12c5de78314b3c18ed4fb1d21dc10.png

WoWsRanked7.thumb.png.2bb27a80242924b0dd4967189ce2467e.png

Same data again but in battles per day:

WoWsRanked8.thumb.png.1f9c3fa32e2625e13fe02bac9518fed4.png

WoWsRanked9.thumb.png.e8644966fefc0a94814771320407b72e.png

WoWsRanked10.thumb.png.2ac321840ef8dbe0d9cd6df0b1ec271b.png

By suggestion, here's another graph:

500991913_BattlestoRankOutbyStarSavingRate50WR(1).thumb.png.487d25a3b2008337d5dc0b53440b7935.png

Edited by Horsie_ships
Added new charts, fixed old charts
  • Cool 15
  • Thanks 7

Share this post


Link to post
Share on other sites
Members
4,098 posts
44,150 battles

You said near impossible 40%. Hoo, hooo. Whenever a statistician says that, they get in a whole lot of trouble when the numbers overwhelmingly show otherwise. You raised your own flag there bud.

  • Sad 1
  • Meh 2

Share this post


Link to post
Share on other sites
1,822
[1984]
Members
4,482 posts
21,506 battles
4 hours ago, Horsie_ships said:

With a python script, I can simulate a singular ranked sprint, taking into account irrevocables, star saving, and win rate. This can help predict an approximate number of battles to either rank out or get N wins. Of course, due to randomness, different simulations will often have different results. To compensate for these differences, I ran each simulation 100,000 times and took the average of all these simulations.

Here's the data, visualized:

WoWsRanked.thumb.png.09a7defef684f440dff8912cd876e875.png

The average number of battles it will take you to rank out, by WR

Note that these statistics are averages of the simulations, meaning that although it's an approximation, you can still play more or less games by chance. The chart also assumes a fixed Star Saving chance of 1/7, because there are 7 players per team. In actuality this chart would be steeper, because players that win more games will also be more likely to save their star, and vise versa, but I've kept it constant since I'm not sure how I would accurately scale it with win rate.

Here's the same data, visualized on a battles per day basis:

WoWsRanked2.thumb.png.ac05f1e6e13e5380dec26c053fa05d05.png

Same data as before, but divided by 14 because there are 14 days per sprint

What about qualifiers? Here's the data for qualifiers:

WoWsRanked3.thumb.png.0622dc8ef1881d05ececddd099d42cfb.png

Five stars are needed to qualify into silver, and seven stars to qualify into gold

What about the entire season? Here's some more data:

WoWsRanked4.thumb.png.8fb61b6ec36d0d60e6861c90a7615283.png

The average total number of battles played the entire season to reach those goals

While some players may choose to stay in bronze and some in silver, I've decided to take a look at "Doubloon Farming" and "Maximum Rewards." Doubloon farming means to qualify into gold as soon as possible, then getting the 29 wins per sprint until the end of the season, and maximum rewards means to qualify into gold as soon as possible and rank out every gold sprint. As a clarification, while ranking out of silver does yield the same amount of doubloons as 29 wins in gold, it is statistically harder to rank out of silver than get 29 gold wins.

Keep in mind that while this graph suggests that a 40% WR player can still doubloon farm (although extremely time consuming), they will still need to push pass the near impossible (for them) silver rank out + gold qualification.

Again, all the data is an approximation, and does not mean you will be guaranteed to accomplish your goal if you play at a pace of your approximated number of battles. Take what you will from the numbers to help you decide what is and what isn't feasible.

Cant read you black text. But the graphs look cool and seem to be accurate based on my experience.

Share this post


Link to post
Share on other sites
122
[V1-PI]
Members
435 posts
10,136 battles

so with my 48% WR and i can only play 4-5 games a day it seems you are telling me to play a different game.  lol stompy mechs and borderlands here i came. 

i do a bit better in the more competitive game modes but it looks like no matter what my low games per day will keep me in the bronze age. 

 

thank you 

  • Cool 2

Share this post


Link to post
Share on other sites
1,091
[WOLFC]
Members
2,343 posts
1 hour ago, rutilius83 said:

so with my 48% WR and i can only play 4-5 games a day it seems you are telling me to play a different game.  lol stompy mechs and borderlands here i came. 

i do a bit better in the more competitive game modes but it looks like no matter what my low games per day will keep me in the bronze age. 

 

thank you 

Yup, I realised that after a couple of days the first time around on this mode & these stats just prove my point.  Like, I'm OK playing a few games a day to get a bit of steel but 7 a day on this mode for days on end  when I can maybe do 4 a night most nights?

No thanks.

 

 

  • Cool 1

Share this post


Link to post
Share on other sites
5
[HAIFU]
Members
12 posts
10,416 battles
4 hours ago, SteelRain_Rifleman said:

You said near impossible 40%. Hoo, hooo. Whenever a statistician says that, they get in a whole lot of trouble when the numbers overwhelmingly show otherwise. You raised your own flag there bud.

i mean, he's not wrong. In order to have a 40% winrate you'd have to do really poorly a majority of the time and have one really really lucky streak to just barely make up for it, which, mathematically, is a small chance

  • Cool 1

Share this post


Link to post
Share on other sites
23
[WHS]
Members
37 posts
6,215 battles
4 hours ago, SteelRain_Rifleman said:

You said near impossible 40%. Hoo, hooo. Whenever a statistician says that, they get in a whole lot of trouble when the numbers overwhelmingly show otherwise. You raised your own flag there bud.

To clarify, the average 40% WR player will find themselves playing at a rate of 81 battles per day during the silver rank out + gold qualification sprint, if their intention is to doubloon farm. This number far exceeds even the most active player of last season. Of course, there's always luck, hence why I said near impossible.

Share this post


Link to post
Share on other sites
756
[-TRM-]
[-TRM-]
Members
2,533 posts

80 battles per day times 20 minutes per battle comes out to 26.4 hours per day.

 

YA right.

 

The joy in ranked is not attempting the [edited] excessive gaming to qual or rank out, you just get bitter and burn out then quit. It takes me dozens of games to find the right string of victories to rank out or qual.

I refuse to play to that puppet string.

Share this post


Link to post
Share on other sites
SuperTest Coordinator, Beta Testers
6,806 posts
12,620 battles

At least one player last season managed almost 50 battles a day, every day, for the entire 6 weeks. 1900 games in 6 weeks.

Share this post


Link to post
Share on other sites
1,150
[DRFTR]
Beta Testers
3,920 posts
30 minutes ago, xHeavy said:

80 battles per day times 20 minutes per battle comes out to 26.4 hours per day.

 

YA right.

 

The joy in ranked is not attempting the [edited] excessive gaming to qual or rank out, you just get bitter and burn out then quit. It takes me dozens of games to find the right string of victories to rank out or qual.

I refuse to play to that puppet string.

if you are a 40%'er your matches will be closer to 5 mins than 20...

  • Cool 1

Share this post


Link to post
Share on other sites
7,068
[WORX]
Members
12,638 posts
19,907 battles
10 hours ago, Horsie_ships said:

With a python script, I can simulate a singular ranked sprint, taking into account irrevocables, star saving, and win rate. This can help predict an approximate number of battles to either rank out or get N wins. Of course, due to randomness, different simulations will often have different results. To compensate for these differences, I ran each simulation 100,000 times and took the average of all these simulations.

Here's the data, visualized:

WoWsRanked.thumb.png.09a7defef684f440dff8912cd876e875.png

The average number of battles it will take you to rank out, by WR

Note that these statistics are averages of the simulations, meaning that although it's an approximation, you can still play more or less games by chance. The chart also assumes a fixed Star Saving chance of 1/7, because there are 7 players per team. In actuality this chart would be steeper, because players that win more games will also be more likely to save their star, and vise versa, but I've kept it constant since I'm not sure how I would accurately scale it with win rate.

Here's the same data, visualized on a battles per day basis:

WoWsRanked2.thumb.png.ac05f1e6e13e5380dec26c053fa05d05.png

Same data as before, but divided by 14 because there are 14 days per sprint

What about qualifiers? Here's the data for qualifiers:

WoWsRanked3.thumb.png.0622dc8ef1881d05ececddd099d42cfb.png

Five stars are needed to qualify into silver, and seven stars to qualify into gold

What about the entire season? Here's some more data:

WoWsRanked4.thumb.png.8fb61b6ec36d0d60e6861c90a7615283.png

The average total number of battles played the entire season to reach those goals

While some players may choose to stay in bronze and some in silver, I've decided to take a look at "Doubloon Farming" and "Maximum Rewards." Doubloon farming means to qualify into gold as soon as possible, then getting the 29 wins per sprint until the end of the season, and maximum rewards means to qualify into gold as soon as possible and rank out every gold sprint. As a clarification, while ranking out of silver does yield the same amount of doubloons as 29 wins in gold, it is statistically harder to rank out of silver than get 29 gold wins.

Keep in mind that while this graph suggests that a 40% WR player can still doubloon farm (although extremely time consuming), they will still need to push pass the near impossible (for them) silver rank out + gold qualification.

Again, all the data is an approximation, and does not mean you will be guaranteed to accomplish your goal if you play at a pace of your approximated number of battles. Take what you will from the numbers to help you decide what is and what isn't feasible.

This is by WR based projection...I think other factors carry more weight then WR alone...  WR is dependent on the capabilities to win as/with a random fleet...

Now I have a question, have you done this with ship class specific projection ???

WR is not a good marker to base your projection on... I think observing what ships do Rank out and how many games they took to Rank out, tell the complete picture.


In other words, is there a way to lower the deviation to were its less speculative and more accurate ?

Share this post


Link to post
Share on other sites
Members
46 posts
199 battles
40 minutes ago, xHeavy said:

80 battles per day times 20 minutes per battle comes out to 26.4 hours per day.

 

YA right.

 

 

How many of your own battles actually hit the 20 min mark before you are either dead, or it ends early? I wager ALOT.

  • Funny 1

Share this post


Link to post
Share on other sites
566
[-TKS-]
[-TKS-]
Members
1,321 posts
13,178 battles
10 hours ago, Horsie_ships said:

simulate a singular ranked sprint, taking into account irrevocables, star saving, and win rate.

Well done. I like making simulations as well. I ran one for the prior ranked. if you want to shock everyone... Remove Save a star and post those graphs.  save a star dramatically reduces the number of games for players that can feasibly rank out.   

Share this post


Link to post
Share on other sites
23
[WHS]
Members
37 posts
6,215 battles
43 minutes ago, skillztowin said:

Well done. I like making simulations as well. I ran one for the prior ranked. if you want to shock everyone... Remove Save a star and post those graphs.  save a star dramatically reduces the number of games for players that can feasibly rank out.   

I just added a chart I initially didn't feel like adding that varies star saving percentage and keeps win rate constant. Probably not exactly what you meant, but in this case a 0 percent chance to save a star is essentially no star save system.

Share this post


Link to post
Share on other sites
5
[HAIFU]
Members
12 posts
10,416 battles
1 hour ago, Navalpride33 said:

This is by WR based projection...I think other factors carry more weight then WR alone...  WR is dependent on the capabilities to win as/with a random fleet...

Now I have a question, have you done this with ship class specific projection ???

WR is not a good marker to base your projection on... I think observing what ships do Rank out and how many games they took to Rank out, tell the complete picture.


In other words, is there a way to lower the deviation to were its less speculative and more accurate ?

tell that to wargaming then.

Literally the only thing that determines whether you gain or lose a star is whether you win or lose. Horsie isn't saying that's how it should be, he's saying that's how it is. This graph is accurate to what your ranked winrate is, not your overall winrate. 

 

Edited by Refudgerator
grammar

Share this post


Link to post
Share on other sites
7,068
[WORX]
Members
12,638 posts
19,907 battles
1 minute ago, Refudgerator said:

tell that to wargaming then.

Literally the only thing that determines whether you gain or lose a star is whether you win or lose. Horsie isn't saying that's how it should be, that's how he's saying that it is. This graph is accurate to what your ranked winrate is, not your overall winrate. 

No no no, I saying...  If we use this example (WR) because no other example can be used (suggestion from my last post)...

How can in your opinion (since its your work)... Lower the deviation to where its not an approximation???

I trying to think of a possible solution... Problem is the deviation is too high just an WR alone... 

 

Share this post


Link to post
Share on other sites
23
[WHS]
Members
37 posts
6,215 battles
6 minutes ago, Navalpride33 said:

No no no, I saying...  If we use this example (WR) because no other example can be used (suggestion from my last post)...

How can in your opinion (since its your work)... Lower the deviation to where its not an approximation???

I trying to think of a possible solution... Problem is the deviation is too high just an WR alone... 

 

The deviations are unavoidable because this is the nature of randomness. Let's say a simulation of 4 games are run, using a win rate of 50% (meaning each game has a 1/2 chance of being a win). In most simulations, it will be 2W and 2L. But there will also be 1W 3L, 3W 1L, and even 4W 0L, 0W 4L. I can't confidently say that every simulation will be 2W 2L, because there will be deviation, but it will be the closest I can get. While it's true that many factors can influence the outcome of a single game, when we are trying to figure out how many total battles will need to be played, overall WR offers the best prediction.

Share this post


Link to post
Share on other sites
7,068
[WORX]
Members
12,638 posts
19,907 battles
16 minutes ago, Horsie_ships said:

The deviations are unavoidable because this is the nature of randomness

Here is what I am doing (I should've made this public before the patch drop)...

I am doing a ship class statistics, Just on a given ship of choice.... I am formatting a theory stating, ships are more the driving force on a WR then the skill of the ships operator..


The study is in its infancy. Since we just started Ranked battles and I, just picked my ship of choice for the cause... My limitations

  • Limited data that fits the requirements for entry, analysis or consideration into the determination.
  • Not to mentioned, I am the data of one... (major flaw).

By doing this... I am trying to limit the influence

  • Theory of Random
  • Theory of chance
    • Also known as the probability theory

On the  final results.


Because its high tier... I also have to isolate the Snowball affect from the final outcome... HOWEVER< this is nearly impossible depending on the OP/over capable WG product used.
The work is only a few days old...

Edited by Navalpride33

Share this post


Link to post
Share on other sites
566
[-TKS-]
[-TKS-]
Members
1,321 posts
13,178 battles
1 hour ago, Horsie_ships said:

exactly what you meant, but in this case a 0 percent chance to save a star is essentially no star save system.

 Yes Exactly!!.

now looking at your very graphs. 

Gold - 50% w/r with 1/7 save a star  take ~300 battles

Gold -50% w/r without save a star  OFF THE CHART!!! > 600 Battles. 

The goal isn't to save the star its to win, but without save a star, even good players cant play enough games to finish.  

Share this post


Link to post
Share on other sites
173
[F4E]
Members
336 posts
6,327 battles
7 hours ago, rutilius83 said:

so with my 48% WR and i can only play 4-5 games a day it seems you are telling me to play a different game.  lol stompy mechs and borderlands here i came. 

i do a bit better in the more competitive game modes but it looks like no matter what my low games per day will keep me in the bronze age. 

 

thank you 

I'm an above average player, but I realized years ago that I would have to play ALL of my (limited) games in ranked to grind out the top.

Rank is toxic, terrible, and my least favorite mode to play.  So the choice became easy for me.

What would be interesting, is if the not so good players (45-50% WR ish) decide not to play ranked because it's not worth it, think of just how hard it would be for the top players to rank out.

--

+1 to OP, thanks for doing the math!

Edited by Glamorboy
+1 to OP
  • Cool 1

Share this post


Link to post
Share on other sites
566
[-TKS-]
[-TKS-]
Members
1,321 posts
13,178 battles
4 minutes ago, Glamorboy said:

if the not so good players (45-50% WR ish) decide not to play ranked because it's not worth it, think of just how hard it would be for the top players to rank out.

Yes!!!. exactly!

that's what happens first week in gold. and why so many are salty.  Happens as you go up in leagues, The winners are now playing against there own kind and a new bell curve is drawn, which means some winners begin to lose. 

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×