Category Archives: Statistics

How we found the best players on our paintball team (Part 7): New vs Old Players

To recap what we’ve covered so far:

If you remember from the first post, one of the main reasons we decided to start keeping statistics is that I really wanted to know how much better our so called “tournament experienced” players were than our players who had little or no tournament experience.

That answer came slowly because over time we discovered that there are a lot of dimensions to what makes a player “good”.

“Pair up in groups of three then line up over there in a circle, alphabetically by height”
-Anonymous Football Coach

In the aforementioned first post, we broke our players down into three categories. The category we were most interested in were the “Category A” guys: players with local, regional and sometimes national level experience. Were they really better than the Cat C players who came diligently to practice?

Before I answer that question, let’s think about what it means to be in each category. If you are Cat A player your story probably sounds like the following:

I started playing on the kids team of a local D1/2 team when I was 14. They were kind of disorganized but overall they had a good direction for the team and practiced if not twice a weekend then at least every other weekend. I went to all the practices, most of the regional events and eventually the PSP events on the D4/D5 team. After a while, I moved up to the D3 team and by the time I was 18 I was playing on the D2 team. This year we’re bumping up to D1 and my goal is to eventually play pro.

Next, we will switch to Cat B players:

I first played at a birthday party when I was 13 and then started playing with some friends of mine pretty much every weekend. We started to play some local tournaments and we lost pretty much every game we played so a lot of my friends stopped playing. However, a couple of us kept going to events and pretty soon we started to doing better. After a while, I got better and started playing with other local teams. Most of the time, the team wasn’t very organized and the roster kept changing so I was usually playing with people I had never met before. Nobody was really ever in charge and it was a bunch of teenagers so whoever yelled the loudest decided what was happening.

Finally, here is Cat C:

I played high school football/basketball/soccer/hockey etc for four years and first played paintball when I was 16/17 at a birthday party. I played with my friends for a couple years and we entered a local tournament or two but that’s pretty much it.

Given the above, hundreds of hours of data collection and processing and thousands of hours thinking about this I can tell you this: every player is both really good at something and really bad at something else. I don’t care what player you choose and and what statistic you are measuring, everyone has a weak spot (at least in college).

If you go back and read the descriptions of the player in each category you can probably guess why. Playing time is the biggest limitation in paintball especially when considered to other sports. It’s pretty easy to set up a pick up basket ball game but you can usually only play paintball on weekends.

Inside of that limitation, you have to pick what you want to work on. If you are a Cat B player, that usually ends up being two things: how to maximize your individual skills and doing what looks good to the people that are selecting teams to go to events.

Why individual skills?

The secret of winning football games is working more as a team, less as individuals. I play not my 11 best, but my best 11.
- Knute Rockne

If you are playing with people you’ve never met before there is not going to be a lot of “teamwork” going on. You don’t have common codes and you certainly don’t know how the other players will react to situations that come up during games. Your best option then is to focus on things that you can control like gun battling and moving yourself up the field. This shows up in the stats as being fantastic in any scenario “less” than a 3v3 e.g. 3v2, 2v2 etc.

Basically anytime one person can win a game by themselves by making the right moves. The fact that these skills are also the “flashiest/look good” is even better because that’s the easiest thing for team managers to use when selecting a team. Given that these managers don’t have an objective system and are going solely on personal observation, that’s the best they can do. Unfortunately for them, this isn’t the best option.

Why? The most common scenario a paintball player will face is a 5v5. The two next common are a 5v4 and 4v5. All three of these require substantially more teamwork than a 3v3 or 2v2. Winning a 4v5 requires patience, discipline, field awareness and tons of communication. These skills don’t look very good and are not easy to see so they tend to be overlooked by most team managers. However, they are the most important purely because they happen the most often.

If you are player trying to maximize your tournament playing time, your obvious choice then is to focus on the skills that are the easiest to display and that team managers are looking for. Since nobody is tracking the statistics that actually have a big impact, why focus on developing those skills?

I’ll give you a specific example: one of our best Cat B players was fantastic at gun battling. He easily won over 60% of his gun battles. This was because he came from playing at an indoor field where quick reaction times pay off a lot. His problem? He was terrible at 5v4s. Like, out 25% of the time terrible. Again, since the team he was on before college played at an indoor field, they selected for and rewarded player who made quick aggressive bumps. That works great in a small, crowded indoor field but not on a regulation sized field.

Because of the above, we realized as time went on that a lot of what we thought of as Cat A guys were really Cat B and that there was quite a lot of variation within each category. What we really needed was some kind of number that could give us an overall view of a player’s effectiveness.

You’ll have to wait for an upcoming blog post to find out how we did that.

If you liked the above and you would like to receive updates about new posts and email only articles, you should join our mailing list.

How we found the best players on our paintball team (Part 6): Laning

As we mentioned in the last post, being in a 5v4 is a great place to be. Your odds of winning the point just went from an even 50/50 to almost 75%!

So how do you get to a 5v4? One way, in a word: laning.

As many of you already know, laning is one of the easiest ways to get an elimination. You don’t have to gun battle somebody out. You don’t have to run anywhere to get somebody. You figure out where they are going, put a stream of paint there and BOOM, they’re out.

“Want to win a race, just run really fast”

You are probably saying: “That’s great Alex but not everyone is good at laning!”

Ah ha! Very true.

Our next challenge in collecting the statistics was how to measure laning. We started out by creating a definition:

A player eliminated in the first X seconds of a game counts as being “laned out”

We experimented with several values for X ranging from five to thirty seconds and we finally settled on fifteen. That gave us the right balance between eliminations that came from early gun battles (which we don’t consider laning) and getting out after a slightly delayed break out.

One of our other biggest questions as we were tracking these numbers: how do college players compare to, say, professional players when it comes to laning? We assumed that pros are probably much better but how much?

Thanks to a rather interesting twist of fate, we found out! It turns out that four of the players on the pro team Vicious were currently enrolled students at the University of Nebraska-Omaha(UN-O). This made them eligible to enter the National Collegiate Paintball Association (NCPA) National Championships, which they did.

This stirred up some controversy as some of the teams didn’t think this was fair but I was excited to finally see how top college teams faired against pros. I remember watching UN-O’s first point of their first match. All of their players superman dove into their bunkers, even the bunkers right next to the back center. My assistant coach, Ryan, commented that laning must be pretty good in the pros if you had to always dive into a bunker.

As the points progressed and we gathered more data on UN-O, the number on pro laning vs NCPA laning become pretty obvious. On average, a college Class A team will get at least one elimination from laning about 30% of the time.

Percentage of points a regular team lanes someone out

pro_laners

UN-O on the other hand, was getting a lane elimination almost 85% of the time! That’s incredible. That would be the equivalent of two regular college teams playing each other but one team only starting with four players instead of five. From what we know about 5v4 winning odds, you can probably guess that UN-O proceeded to stomp all over the competition.

Percentage of points UN-O laned someone out

pro_laners

What’s even more amazing is that they did the above with only SIX players on the roster. That means their lines were running back to back points in Florida heat AND laning people out at almost TRIPLE the rate of regular college teams.

Moral of the story: work on your laning kids!

Up next in the series: who’s better: new players or old players?

If you liked the above and you would like to receive updates about new posts and email only articles, you should join our mailing list.

How we found the best players on our paintball team (Part 5): Creating the First Stats

To recap what we’ve covered so far:

Given the above, one of our first big questions was “How much does getting the first elimination in a game matter?”. We guessed that it was kind/sort of important but we had no idea how important and therefore couldn’t answer questions like: “Is it better to try and lane someone out or take the risk and have our snake guy make a big break?”

If you’re reading this I’m going to go out on a limb and assume that you know something about paintball. That’s great because to answer the above question we’re about to go all Socratic Method and get you in the frame of mind we were in as we started creating our statistics.

There are two kinds of statistics, the kind you look up and the kind you make up.
-Rex Stout

You’re at a paintball field watching two college teams about to start a game. Let’s call them Team A and Team B. You know that the two teams are evenly matched.

First question to you: “Given that the two teams are evenly matched, what are the odds that Team A will win?”

This is not a trick question even though it sometimes stumps people. If you think of it as flipping a coin the answer becomes clear:

Odds of Team A winning a 5v5

5v5

If the two teams’ skills levels are balanced then Team A should beat Team B 50% of the time. The more observant among you probably realized that we didn’t need a database, Excel spreadsheets and lots of Perl to figure that out which brings us to the next step in our line of questioning.

“If Teams A eliminates a player from Team B, what are Team A’s odds of winning now?”

Aha! Now we are getting into the mysteries of the paintball universe! Over the years I’ve asked a lot of people both in and out of paintball this question. Most people go through the following thought process:

  • it has to be more than 50%
  • it also has to be less than 100% because I’ve seen teams win 4v5s lots of times
  • I would guess somewhere around 60-80%

And you know what, they would be correct!

Odds of Team A winning a 5v4

5v4

As I mentioned back in the first post in this series, not only was this the first time we knew this, it was the first time anyone knew.

It also kicked off some pretty interesting observations. First of which was: if you are in a 5v5 and you lose a player, your odds of winning get cut almost in half. Turns out getting that first elimination was pretty important.

An interesting corollary to that observation was that if you were in 5v4 and your team loses a player, you just almost doubled the other team’s chances of winning.

Below is the table showing every combination of players left on either team:

vs_table

At this point, you might be wondering: “This is good to know but how did you use this to find your best players?”

Good question! The first revelation we had was “Getting the first elimination is huge! Who are our best laners?” followed shortly by “Who is getting out a lot in 5v4s??”

I’ll dedicate a whole post to laning so for now we’ll focus on the 5v4 scenarios.

Whoever said, ‘It’s not whether you win or lose that counts,’ probably lost.
-Martina Navratilova

Now, we knew that going from a 5v4 to a 4v4 was not a good thing. We guessed that given the pool of players we had, some of then tended to get out a lot in 5v4s and some didn’t. In other words, we wanted to know who was “good” in 5v4s and who was “bad”.

To do that, we first had to determine how often an average player on an average team gets eliminated. After tracking several matches we determined that number was 8%. So how did our players stack up to that number? Check out the histogram below.

Number of players at each percentage odds of being out in a 5v4

5v4_chart

This was pretty shocking! Since we knew the average odds of being eliminated in a 5v4 was 8%, anybody below that we could consider “good” and anyone above “bad” at handling 5v4s. If you add up the numbers you’ll see that out of 21 players, 10 were below average.

“But doesn’t that make sense, Alex? Shouldn’t half of your players be below average?”

Another excellent question!

The 8% number is across multiple college teams. To be more specific in plain language: half of our players were not even as good as an average college player. You’re not going to win any college championships with half of your team being below average in one of the most important statistics in the game.

The craziest part about all of this is several of the players on our A line were being eliminated in 5v4s almost 25% of the time!. In some cases we would have been better off if some of those players didn’t even walk on the field!

If you’ve followed along this far: congratulations! This was the longest post to date. Coming up: How does a college team compare to a pro team?

If you liked the above and you would like to receive updates about new posts and email only articles, you should join our mailing list.

How we found the best players on our paintball team (Part 4): Creating a Model

At this point in our story we’ve:

This, as we all know, was fantastic! To our knowledge, nobody had ever had this kind of data on a college team’s performance.

Our next challenge was to figure out which individual statistic was most important. We knew that a couple were pretty obvious, e.g. laning, but other than that it was anybody’s guess.

Today its now who can shoot the most paint behind a bunker. Back then it was chess, who can out smart the other. Its now a game of twelve year olds who are beating the hell out of me.
- Ron Kilbourne

Many years ago I read a page on ELO rankings in Foosball. It was a simplified version of the chess ELO ranking system. For those of you who have never played competitive chess (I never have), the idea behind it is that everyone gets a “number” that indicates how good you are and works under the following basic assumptions:

  • Given the rankings of two opponents, you can calculate the odds of each player winning easily based on a statistical curve.
  • If the higher ranked player wins, the rankings go up for the winner and down for the loser. The key point being only a little in both directions.
  • If the lower ranked players wins, his ranking goes up a lot and the higher ranked player goes down a lot.
  • a “little” and a “lot” are calculated statistically
  • This means that “upsets” and “sure things” are weighted accordingly.

If you have a statistics background, I highly recommend reading more about it.

For the rest of us, let’s get back to the dirt and paint of the real world.

I, very modestly, decided to call this new ranking system as applied to paintball the “Alexpotato Rating”. Everyone was assigned a “middle” value of 1500 Alexpotato Rating points (the low being zero and the high being 3000). Each practice point was considered an individual event that updated each player’s ranking based on the formula described above.

Some of you might be asking “That makes sense for single players but paintball is a team sport. How do you make calculations based on a group?”

Excellent question!

I took the idea from the Foosball folks and just averaged the the ranking for each player. After ten or so points, we looked at how the rankings played out and we were surprised to see that for most of the players, the ones we thought were better matched closely with the AP Rating.

This let us know that we were on the right track.

The big question was: what made them better?

We answer that question in the next post.

If you liked the above and you would like to receive updates about new posts and email only articles, you should join our mailing list.

How we found the best players on our paintball team (Part 3): Storing and Processing the Data

So if you’ve been following along in our last couple posts, you’ll remember that we had decided to start keeping stats and came up with a relatively easy way to record the data.

We still had a couple kinks to work out.

“The plural of anecdote is not data.”
― Marc Bekoff

Up till now we had been “keeping” our data in an old spiral bound notebook that I’m pretty sure had been floating around in my coaching clipboard since college. As anyone who has been near paintball in any form or fashion realizes, stuff gets damaged/dropped/kicked/blown away on a regular basis. Add on top of that, free form note taking is one of those things like “style” for which there is no accounting for.

As a quick side note: for those of you born in the 90s WDP (who made the Angel) and Smart Parts (who made the Shocker) sued each other over the invention of the first electronic paintball gun. The case was resolved in favor of WDP when one of the engineers presented an engineering notebook as evidence of prior art. Save those notebooks kids!

Back to our story: it was time for some standardization. Since one of my former nicknames was “Alexcel” I whipped up the following format for tracking scrimmages:

paper_form

You can find a full sheet PDF version with multiple games here:
paper_form

So now we had a way to consistently record the data. Question is, how were we going to store it? Enter: the computer!

A computer lets you make more mistakes faster than any invention in human history-with the possible exceptions of handguns and tequila
- Mitch Ratcliffe quotes

As you can guess from the fact that I have awesomely nerdy nicknames combining my name and mad Excel skills, I’m pretty handy with computers in general and databases and Perl in particular.

I’ll add a post in the future with the specifics for those of you who speak SQL as a second language but here is a quick overview:

  • We created a format to enter in the data
  • This script would then take the data and enter it into our database
  • The database had:
    • all of our players
    • teams we scrimmaged against
    • when each player in every scrimmaged was eliminated

This was a pretty big step forward. In plain English, it meant that we could could go to any practice point we played and answer questions like:

  • Who was out first?
  • Who was the last player alive in our first game?
  • Of our snake side players, who is usually the second one out?

I even remember two of our players asking me a question along the lines of: “Can you tell us how often we win if Brian is the first one out vs me being the first one out?”.

Before tracking the data, nobody could answer all of those questions especially a week or two after the fact. This reminded me of the first time someone brought a video camera to a practice. Someone would say “Well, I shot out three people after I bumped!”. You would check the video and turned out one of the guys was already out, one of them was a legitimate kill and the third was him shooting one of our players. In a word: awesome.

Having the stats though was, amazingly, even better! The stats saw everything since every elimination was recorded. On a camera, you only saw what it was pointed at. Plus, unlike a video, you had a permanent store of data that was easy to search and could store hundreds of eliminations with ease.

After I got over our my patently obvious excitement of having more data on college and local team practices at LLP than anyone in the history of paintball, I was faced with the next problem: how to slice the data.

If you liked the above and you would like to receive updates about new posts and email only articles, you should join our mailing list.

How we found the best players on our paintball team (Part 2): Collecting the Data

As I mentioned in my previous post, I suddenly found myself with a burning desire to figure out who my best players were.

Unfortunately for me, I had no idea on how to go about doing that.

So how do you measure chaos?

One of the best quotes I’ve ever read about paintball was that, unlike just about every other sport, there is no center of the action. With no “game ball” that allows you to easily focus on where the next big play will happen how do you even know that something is happening? Even if you did, how would you track it? I guessed that I might need ten people just to get all of the detail.

On top of that, I didn’t even have ten people. It was just me and occasionally an assistant coach and, if I was lucky, somebody’s girlfriend.

Thankfully, necessity is the mother of invention and all that jazz and I had an epiphany!

Remember kids, the only difference between science and screwing around is writing it down

I heard that quote from Adam Savage long after we started keeping stats but it certainly applies here.

I don’t remember exactly how the idea came to me but I thought to myself “What if we just write down when people are out? All we need is a pen and paper and we can track who gets out in what order.”

So projects that sound small turn into hundreds of man hours of science!

First Steps

The very first iteration of keeping paintball statistics started at practice. As we scrimmaged other teams, I made a list of the five players on our side and the five players on the opposing side. Initially, I just wrote down the order of each person getting out.

Something along the lines of this:

  • Tom 1st out
  • Mike 2nd
  • Other team Guy #1 3rd
  • etc

This was pretty cool! It felt like doing science. Real science about something I really cared about.

As I often said to people when describing this whole process, I really understood how researchers feel when they make a breakthrough: not only did I not know what I had just discovered, nobody knew. Not anywhere. This was groundbreaking!

Like all great scientific endeavours though, we faced a new challenge.

“The early bird gets the worm, but the second mouse gets the cheese.” ― Willie Nelson

While it was great to see who was getting out in what order, we were really missing the key points about those outs: when they occurred.

For example, if a player was the first person shot out and that happened in the first five seconds of a game, it was a lot different than being the first person shout out two minutes into a game. How were we going to track this?

Easy. We added in that great tool of coaches throughout history: the stopwatch.

Now, instead of just writing down the order, we also started writing down the time of each elimination. This opened up all kinds of interesting possibilities. We could track who got laned out. Who was able to hold out against a 1v2 for the longest.

How we did that is in part 3

If you liked the above and you would like to receive updates about new posts and email only articles, you should join our mailing list.

How we found the best players on our paintball team (Part 1): How it all started

It all started with me saying “You are not fucking prepared!”

Let’s rewind a little before we go there and start with a little background.

The Background

For over 15 years I was involved with paintball in one form or another.

I was:

  • a college paintball player
  • General Manager of Paintball Sports Promotions aka PSP
  • a conference director in the National Collegiate Paintball Association aka NCPA
  • a college paintball coach

For that entire time I was fascinated by how teams beat their competition. It was pretty obvious that some teams were better because they scored more points or received less penalties but other than the things that were easy to see, there wasn’t much to go on.

Which led me to the next question: How do teams pick their rosters? Was it random chance? Watching people play? Choosing the player that showed up to practice most often? I didn’t know. Mostly because I had never coached or managed a team and never had a pool of players to choose from. Even as a college team captain, we usually needed five players and we only had four people with any experience.

The Players

That all changed when I became a college paintball coach. I suddenly was presented with three basic categories of players:

  • Group C: People who had never played before but came to practice all the time
  • Group B: People who has played before but had never been to a tournament and came regularly to practice
  • Group A: People with real tournament experience (regional or above) and who never came to practice

Most of the time, we couldn’t even get the Group A guys to even come to tournaments and while the Group Bs were ok, we certainly weren’t winning lots of matches (we were playing the college version of X Ball).

The captains and coaches always wondered: “How would we do if we actually had Group A guys on our roster?”

Turns out, we were going to find out.

The Event

Magically, we ended up with a roster that looked like the following for one of our events:

  • 5 Group A guys
  • 6 Group Bs
  • 4 Group Cs

Naturally, we thought that starting the A players would be the best idea. As we progressed through our matches, a couple things became pretty clear:

  1. The Group A line was winning just about 50% of their points
  2. They were making some pretty bonehead decisions when it came to match strategy
  3. When we pointed out #2, they told us we didn’t know what we were talking about

“What kind of bonehead decisions?” you might be asking, here is an example:

We are down by one point with a 1:30 left on the game clock. The other team has four players left and we have three.

Our Group A guys just sat there.
No big moves.
No run throughs.
Nothing.

They let the clock run out and guaranteed us a loss.

I was so mad I couldn’t even contain myself. Here were, supposedly, our best players making decisions that went against basic paintball strategy and, on top of that, telling the coaches that we didn’t know what were doing.

The Decision

Which brings us back to the opening quote of this post. At an end of the day team/coaches talk, our Group A players added insult to injury by trying to claim that it “just wasn’t our day”. Since they were very obviously not prepared, I let them have it in no uncertain terms.

In my frustration, it suddenly became clear that tournament experience by itself was not enough to win us matches. Preparation via practice, drills, layout review or some combination of all of those items meant more. The question was how much more?

Right then and there I resolved to find out once and for all: “How do you objectively and empirically determine who the best players on your team are and what is it, specifically, that makes them the best?”

How that all started, will come in Part 2.

If you liked the above and you would like to receive updates about new posts and email only articles, you should join our mailing list.