Find fields & stores near you!
Find fields and stores
Zipcode
PbNation News
PbNation News
Community Focus
Community Focus

Reply
 
Thread Tools
Old 04-21-2012, 09:51 AM #1
Alex Elliott
 
 
Alex Elliott's Avatar
 
Join Date: Feb 2004
Location: New Jersey
 has been a member for 10 years
Alexpotato College Paintball Stats (Part 1)

In the tradition of Alexpotato bringing out the kind of things smart paintball people enjoy(AP Scoreboard, AP Exchange etc) and this being a college paintball forum after all, I've started keeping track of statistics on matches and Rutgers practices and figured I would share what I've found.

Note: Appendix of data and method at the bottom.

Question #1: How important is getting that first elimination?

To answer this question we first need to know the following:

How do teams historically do when playing a point where they are either:
a. up one body (5v4) or
b. down one body (4v5)?

Now, obviously, on average, the odds of winning a 5v5 is 50%.

However, the odds of winning a 5v4 is ~75% and the odds of winning a 4v5 is ~25% which makes sense given that the live bodies are reversed.

"But Alex", you say, "these are just averages! What is the range?".

Good question. Overall, the team with the best record in 5v4's is Nebraska-Omaha with an average of 90%. The team with the lowest record was Northeastern at 55%.

To put it another way, if your team can win a 5v4 more than 75% of the time, you are doing an above average job of closing out games.

Question #2: So given the above, how important is laning?

This question logically flows out of Q#1. One way to answer this is to create a Monte Carlo model of an imaginary team using the odds found from Q#1. See Appendix #2 for the details.

The idea behind this is to say: "For a 10% increase in laning efficiency, how is your match win percentage affected all other things being equal?"

Below is the result:
Code:
Lane Efficiency - Win Percentage 0% - 50% 10% - 56% 20% - 61% 30% - 66% 40% - 72% 50% - 77% 60% - 81% 70% - 85% 80% - 88% 90% - 91% 100% - 94%
Or for those of you who prefer charts with a logarithmic trend line and equation:


In other words, if in 60% of the points your team lanes someone out and the other team is terrible at laning, you should be winning about 80% of your matches.

"That's great Alex" you say "but laning is really hard".

Another good question. Turns out, on average, teams lane someone out 37% of the time in points that they play. Nebraska-Omaha holds the record here with 90% and Ohio the lowest record with 12%.

QUESTION #3: So why doesn't everyone win 72% of their matches?

Excellent question.

Couple reasons:
a. Some teams have high lane efficiency ratings but low 5v4 odds of winning e.g. Rutgers
b. Some teams have terrible lane efficiency ratings but do well in 5v5s e.g. FAU
d. Some teams are great at laning but also tend to get laned out a lot which puts the game back to a 4v4. e.g. UConn
c. Q#2 only takes into account one team laning. In the future, I may put together a more complicated model that tracks both teams laning and/or odds of getting laned out.

Kind of long but don't worry, more coming in the future.

APPENDIX #1: Match Data

Data was taken from:

1. All of Rutgers games at NEIC #3 vs:
-UConn
-West Point
-Northeastern

2. Other non-Rutgers game at NEIC #3:
-Northeastern vs UConn
-West Point vs Northeastern

3. Rutgers games at Nationals 2012 vs:
-Kennesaw Sate
-Ohio
-FAU

4. non-Rutgers game at Nationals 2012:
-Omaha vs Kennesaw

APPENDIX #2: Monte Carlo Model

To do this, we create two teams (A and B) with the the following characteristics:
-Team A has a variable rate of laning people out (more on this later)
-Team B never lanes out anyone.
-Both teams, however, have the exact same odds of winning 5v5s (50%), 5v4s (75%) and 4v5s (25%).

Given the above, we start with 10 possible levels of laning efficiency starting with 0% and then going up to Team A laning someone out 100% of the points they play.

Next, for each level of laning efficiency, we create 10,000 sample matches of A versus B. Each sample match has 9 points which the data shows is the average number of points in a match.

For each point in a match, if A lanes someone out, 75% of the time they will win a point, if not it goes to 50% and so on.

In each sample match if you win more points you win the match and so on. Ties are possible and are counted as half a win for win percentage calculations.
Alex Elliott is offline   Reply With Quote
Old Sponsored Links Remove Advertisement
Advertisement
Old 04-21-2012, 04:14 PM #2
gbk2
ubpaintball.com
 
gbk2's Avatar
 
Join Date: Jan 2009
Location: Upstate NY
gbk2 plays in the PSP
gbk2 is an NCPA National Champion
Good post. As an engineering student I love this stuff, been thinking about this sort of thing a lot lately. Thank you!
__________________
University at Buffalo Paintball Club
WELLSVILLE FSU CREW#5
gbk2 is offline   Reply With Quote
Old 04-22-2012, 10:32 AM #3
HoktFonix
TTU & VT PB Ex-President
 
HoktFonix's Avatar
 
Join Date: Apr 2004
Location: McKinney, TX
Annual Supporting Member
HoktFonix plays in the PSP
HoktFonix donated to help Peyton Trent
HoktFonix supports our troops
HoktFonix is an NCPA player
HoktFonix is reppin' sidebar 4 life
HoktFonix is a top reviewer
HoktFonix is Boss
Seems like you should get over with paintball access before they take over paintball stats without you
Really cool stuff though, like hearing it.
__________________


VIRGINIA TECH X-BALL ALUMNI
TEXAS TECH PAINTBALL ALUMNI
HoktFonix is offline   Reply With Quote
Old 04-23-2012, 01:53 PM #4
UMDpaintball
TMNT4life
 
UMDpaintball's Avatar
 
Join Date: Feb 2004
Location: DC
 has been a member for 10 years
UMDpaintball owns a Planet Eclipse Etek
UMDpaintball is an NCPA player
UMDpaintball plays in the APPA D5 division
This is great Alex. Keep it up!
__________________
___________________________
University of Maryland Terrapins Coach
Mid-Atlantic Coast Conference Champions 2010-11, 2011-12

Paintball Adventure Park
University of Maryland
UMDpaintball is offline   Reply With Quote
Reply


Thread Tools

Posting Rules
Forum Jump