Jump to content
Sign in to follow this  
You need to play a total of 20 battles to post in this section.
Merc85

Impact of Survivability on Win Rate

14 comments in this topic

Recommended Posts

571
[PEED2]
Beta Testers
2,013 posts
25,484 battles

I've been keeping data for a while on battles to look the impact of what occurs when your team loses the first ship in a match (i.e., when the other team gets first blood).  I did this for my last 200 matches, all of which were random battles tier VI and higher, most of which were tiers VIII - tier X.  The data collected showed that the team that loses the first ship also lost the match 65% of the time (for the first 100 matches it was 63% of the time and for the second 100 matches it was 67% of the time so fairly consistent over 200 matches). 

I wanted to see how surviving longer impacted players so I looked at the data for three groups of players in my clan (who had more than 3000 battles) to see what the relationship between surviving and win rate was.  I picked those with over a 60% win rate, those with under a 50% win rate, and the group of captains who had a 55% win rate (54.5% to 55.4%).  Their data were as follows:  those over the 60% win rate threshold averaged 62.3% win rate and survived 48.8% of their matches; those under the 50% win rate threshold averaged a 48.8% win rate and survived 27.4% of their matches, and those who were in the 55% win rate group averaged a 54.9% win rate and survived 43.5% of their matches.  I put these data points into a regression analysis and the formula for "predicting" %win rate based upon your %matches survived came out to be %win rate = 32.6 + 0.57 x %battles survived.  This is not based upon large amount of data but the trend is obvious.

So if you want to increase your win rate by 1 percentage point this indicates you need to increase the number of battles survived by 1.75 percentage points.

Another thing that comes to mind with these data is how a slight tweak to the MM could make for a little more balance to the teams (I am not for a skill based MM, but this could be a bit of improvement).  Maybe the MM could look at something like the four poorest captains (lowest PR or lowest survival rate, or?), and allocate two to each team.  It seems that the first to die is almost always a poor player who doesn't understand how to survive (DD rushes in or cruiser just needs to get close to torp and turns broadside, etc.).  And since when a team loses the first ship it gives the other team a huge advantage in terms of winning (65% vs. 35%), this allocation of a small number of poor players could help balance the team just a bit.  Just food for thought.

 

P.S. Interestingly, the regression equation also says that if you don't survive any matches at all you'll still win 33% of your matches.

 

  • Cool 13

Share this post


Link to post
Share on other sites
1,145
[ARGSY]
Members
10,326 posts
16,093 battles

its a fine line between not being aggressive enough and too aggressive.    my goal  every match is to survive with minimal health and have survival rate around WR .    if you div up more,  your survival rate should go up a bit as you will win more.   

Share this post


Link to post
Share on other sites
11,940
[ARGSY]
Members
19,977 posts
14,258 battles

The other thing is that your individual survival does not always equate to victory and your individual destruction does not always equate to defeat. 

The extreme  is a 2 or more v 1 in which your team is clearly losing on points, but you win the battle by sacrificing your ship to deny someone's Solo Warrior, or to take out the last ship the enemy possesses that stands a chance of turning the tide (e.g. by being at full strength or being close to capping your standard-mode base), leaving your colleagues to mop up the sorely wounded.

Share this post


Link to post
Share on other sites
Members
235 posts

I was pondering something like this myself the other day, I would love to see what the win rate is of the team with the last DD alive in a game.

 

We have all seen it a normal game say 1 CV 4 BBs 4 CAs and 3 DDs the first 5mins normally both teams are lucky to have any DDs left, yet later in the game they become huge, but most DDs are played hyper aggressively 

 

 

Share this post


Link to post
Share on other sites
1,629
[SALVO]
Members
4,335 posts
3,760 battles

+1 for your time and effort. The First Blood relation seems a bit "charged", maybe  is just the sample is too small.

Share this post


Link to post
Share on other sites
658
[CAST]
[CAST]
Members
2,598 posts
10,459 battles

I find that the team with all ships alive at the end of the game wins almost 100% of the time.  

In reality, any player surviving increases the chances of winning as long as they are actively participating in the battle.  Just surviving and hiding does not in itself add to any advantage your team has over the enemy team.  The better a player is, the more their survival impacts the game.

 

Share this post


Link to post
Share on other sites
571
[PEED2]
Beta Testers
2,013 posts
25,484 battles
11 minutes ago, ArIskandir said:

+1 for your time and effort. The First Blood relation seems a bit "charged", maybe  is just the sample is too small.

Maybe....but it was odd that the data for each group of 100 matches was very very similar.

Share this post


Link to post
Share on other sites
Members
165 posts
6,168 battles

Thanks for sharing!

Here is a number you might find useful. If you go to a player profile on worldofwarships.com you can find win rate, total number of battles, number of battles survived, and the number of victories in battles survived. From those you can calculate the winrate in battles destroyed which you will find to be much lower than the average win rate. In my case it's 36% compared to 56%. This suggests that my team's chance of winning (on average) is 36% in battles where I die.

35 minutes ago, Merc85 said:

could help balance the team just a bit.

I think this would make the games last longer on average but would also skew all win rates towards 50%. So if you are a 40% player your win rate will go up slightly, a 60% player it will go down. It depends what you're looking for (fair games or fair stats.)

38 minutes ago, Merc85 said:

Interestingly, the regression equation also says that if you don't survive any matches at all you'll still win 33% of your matches.

Interesting! This is somewhat consistent with the observation that the lowest win rates on the server are about 30-ish% once you cut out players with very few battles. If you'd like some insight into win rates, play around with a 2v2 thought experiment. Suppose you have 4 players in a 2v2 game. They each get a a skill rating of either 0 (poor),1 (average) or 2 (pro). The team with the higher skill always wins. If the skill is even, it's 50-50.

Case 1: All players have 1 point
In this case everyone has a 50% win rate.

Case 2: One player has 2 points, everyone else 1 point
Now the pro player wins 100%, everyone else wins 33.3%. Basically the average players are getting carried every match.

Case 3: One player has 0 points, everyone else 1 point
The poor player wins 0%, everyone else wins 66.7%. So having player crappier than you can boost your win rate. 

Case 4: One player has 2 points, one has 0 points, and two have 1 point.
The pro wins 83.3% (down from 100% if there was no poor players), the poor player wins 16.7% (up from 0%), and each average player wins 50%. 

There are so many variations and limiting cases that you can do with this experiment. Have fun.

  • Cool 1

Share this post


Link to post
Share on other sites
298
[PHD]
Members
1,505 posts
6,377 battles

I suppose you could (probably incorrectly) assume that hiding at the back of the map or behind an island is conducive to winning. 

Not that is a thing for my PvE DDs, depending on spawn location, go forward, spot then survive usually by strategic withdrawal. 

Me, I  like surviving. It can get boring waiting for the match to end. 

Share this post


Link to post
Share on other sites
571
[PEED2]
Beta Testers
2,013 posts
25,484 battles
34 minutes ago, Baritone4 said:

.........................................................

I think this would make the games last longer on average but would also skew all win rates towards 50%. So if you are a 40% player your win rate will go up slightly, a 60% player it will go down. It depends what you're looking for (fair games or fair stats.)   Yep, and that is why I am against a skill based MM.   But if this makes the battles last a littlelonger and gets them a little bit more balance I would be for it.  It might skew things a little, but not much imo.  Or just allocate the two "worst" captains to one to each team.

Interesting! This is somewhat consistent with the observation that the lowest win rates on the server are about 30-ish% once you cut out players with very few battles. .....................................   I didn't say it, but this is exactly why I said it was interesting...it verifies what the lowest win rates are which imo gives a bit of support that the regression equation is in the ball park.

 

  • Cool 1

Share this post


Link to post
Share on other sites
1,299
[VCRUZ]
Members
4,049 posts
9,180 battles

IMO surviving is one of the most important thing in the game. Not becuase of the stats per se, but because if you want to win you need to carry, and you cant carry if you are dead. The longer i stay in the battle, more chances i have of winning, or at least save some XP/credits from a defat. 

 

Of corse, suriving doesnt mean much if you are sitting in the back being useless. You need to balance your agression, so that you are doing something to your team but you are not dying. 

  • Cool 2

Share this post


Link to post
Share on other sites
571
[PEED2]
Beta Testers
2,013 posts
25,484 battles

Just out of curiosity I also looked at the data for these same captains in my clan to include not just survival rates but also average damage.  Using a multiple regression analysis the equation came out to be:

% win rate = 36 + 0.20 x % battles survived + 0.20 x (k damage)

In other words a 1 percentage point increase in battles survived is equivalent to doing an extra 1000 damage in terms of impacting your win rate.

P.S. This does ignore the pretty obvious auto correlation between surviving longer and doing more damage because of that.

Share this post


Link to post
Share on other sites
Members
667 posts
9,178 battles

Seems legit, both by the numbers and in regards to strategy and common sense. You don't make winning plays while dead.

As others have mentioned though surviving by hiding and not engaging the enemy effectively is not productive. The key is surviving while contesting key areas and caps and inflicting meaningful damage.

Edited by Dr_Powderfinger

Share this post


Link to post
Share on other sites
512
[-TKS-]
[-TKS-]
Members
1,152 posts
10,956 battles
4 hours ago, Merc85 said:

This is not based upon large amount of data but the trend is obvious

If you are not doing these analysis for work, you have a new product to sell in your toolbox. well done. 

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Recently Browsing   0 members

    No registered users viewing this page.

×