Mark: Thank you Jacob. I work at the Center for Disease Control (CDC) as a Statistician, in the Division of Violence Prevention. I am compulsive with numbers. My coworkers say I really put the anal in analysis. I’ve been a math geek all my life, learning everything I know from my father. I use these skills in many of my lifetime hobbies, which I tend to get obsessive over for a few years then cast them out like a used up pair of boxer shorts: chess, poker, imaginary worlds of warlocks and goblins, and now I’ve turned my attention to picking scores in the English Premier League. And of all the hobbies I’ve remained faithful to only one since I was 6 years old. That’s soccer. Some people go to church every Sunday. I meditate for 90 minutes between 2 goals, sometimes 2 or 3 times a week, for over 36 years. Through good times and bad times, soccer has always been my balancing force.
My Loyal Readers,
Mark Steven was Crowned Champion of Atlanta Soccer News predictthescore.net league in high fashion.
“An Interview with a Robot, by Jacob Hoffmeister
I recently sat down with Mark Stevens, the 2012 upstart champion of the AtlantaSoccerNews marathon EPL betting challenge PredictTheScore to talk about his amazing accomplishment, how he did it, and a little more about the life and times of Mark Stevens. Mark was recently coronated as the 2012 champion at an AtlantaSoccerNews ceremony held on May 20, 2012 at Los Arcos, passing the illustrious PredictTheScore Prognosticator Cup from previous 2011 champion Quique Lopez, with pictures of the event now going viral on FaceBook. Here is what Mark had to say:
Jacob {resting his chin on his interlocked fingers}: First off, tell me about yourself, what you do for a living, your hobbies, and what makes you tick.
Jacob: May the force be with you! Speaking of soccer, how you got involved in this motley crew of hardcore soccer players called the Rovers?
Mark: Well, playing all through childhood and college with my Irish twin, Sean, we needed to keep playing after graduating from Georgia Tech, so took some of our college buddies to join the Cobb league for a year, and then moved to Dekalb league, running an amateur men’s soccer team called The Wave for maybe 13 years (about 1994ish to around 2006 or so). It wasn’t long before a pair of Colombians in the form of Weo and Quique Lopez joined our brotherhood, going all the way back to maybe 1995 or so. We quickly took the team through the lower divisions to the very top in a few short seasons, and maintained a winning record for most seasons even after quickly reaching Division 1. As the years wore on, the demands of managing a team took its toll, and I started passing around the duties like a redneck passes out off-color analogies. Quique and Weo at times took over for very successful seasons, and as late as 2007 Weo took us to the Division 1 Dekalb finals (and that trophy I got last Sunday looked awful familiar).
Jacob: So you knew Weo and Quique for how long?
Mark: If you do the math, it may be over 16 or 17 years. And when The Wave finally came crashing down after the 2007 season, we still played together with a core group of Wave faithful (Akil, Raoul, Scott, Richard, just to name a few) on other teams. I understand Quique had gotten involved with this motley Rovers team you refer to way early, and Weo also joined a few years back. They kept mentioning this over-30 team the Rovers which was winning championships left and right, and winning all their games along the way. So I went out to practice one day and with the good graces of Quique’s recommendation, coach and manager Mike Kinion signed me on.
Jacob: And how have you liked your time with the Rovers?
Mark: It has been fantastic. When I joined, we were already on a winning streak, maybe Franklin can tell you how long, but we won all 8 games in the first half of Fall 2010 in the over 30 division, having like 48 goals for, 8 against. The team voted and Mike arranged to finish the season (Spring 2011) in the 3rd division, but starting over on points didn’t matter, we finished the regular season with 7 straight wins, and won out the tournament to become Division 3 champions. Then the next season we were set to be promoted to Division 2, but they ended up collapsing Division 2 and Division 1, so we had to face the big boys. Keep in mind that our team is made up of seasoned veterans, meaning most of the players were in their 40s and 50s, while we were set to play against the youngster many fresh out of college coming from very competitive teams. Being on a 19 game winning streak, I challenged the team that if we remained undefeated in Division 1, I would shave my head.
Jacob: Hmmm, you seem to have a full head of hair, what went wrong?
Mark: Well, surprisingly, we played all the best teams to start the season, and still just kept winning, including beating #3 Jamhuri 3-1, #1 Los Gringos 3-2 and a difficult Westside C 3-1. Finally, our streak finished at over 25 straight wins (practically unheard of in soccer) as we had a very difficult first 15 minutes against #4 Hooligans who scored 3 goals quickly, and neither team scored the rest of the game. And while the season was somewhat anticlimactic after that, we still finished with 7 wins, 4 draws and 4 loses and a positive goal difference of +8, proving we can play with the big boys. And we almost exacted our revenge in the playoffs, playing against #4 Hooligans after a well deserved first round bye, coming from down 0-2 to tie it up in the last minute of the 90 minutes regulation. But the soccer gods didn’t smile on us in overtime as they scored the golden goal 9 minutes later, in spite of a well fought game. Yet even though we fell short in playoffs, we also had 3 tournament championships (Father Christmas, Frosty Balls, and Mardi Gras) won over the winter, so we can still hold our heads up high and look forward to next season.
Jacob: Oh yes, I remember that Frosty Balls tournament with you making that fine save in the penalty kick shootout, and Quique closing the deal for the win. Did you read my article on that?
Mark: You mean the one posted at http://www.atlantasoccernews.
Jacob: Speaking of AtlantaSoccerNews, tell me more about this PredictTheScore trophy that has gone viral on facebook. How did you rise from almost a nobody, to nipping at the heals of previous champions and genius prognosticators midseason, to almost taking #1 in the overall leaderboard of over 600?
Mark: Well there’s a saying, “Ties are like kissing your sister.”
Jacob: You kissed your sister?!
Mark: No of course not, let me finish. If ties are like kissing your sister when you’re actually playing soccer, then picking ties as the most probable outcome is like placing a wager that you can kiss your sister. It might happen, but probably won’t, and even if it does, you never know when. If you calculate the odds of one team scoring X goals and another team scoring Y goals, and match up all the combinations, ties are *never* the most probable event, even if the teams are otherwise a toss-up on who wins. At best it’s a 30% possibility with a 35% chance one or the other team will win. Even history shows in the EPL that ties only happen 25-30% of the time. And if you look at the betting odds, there is always one team, possibly both, whose chances of winning are better than the chances of a tie. Always! So I never kiss my sister, and I never bet on a tie. Well maybe that once early in the season…
Jacob: You kissed your sister?!
Mark: No, but I might have bet on a tie the first week or two before figuring out the math.
Jacob: So you did figure this all out using your math skills as a statistician. They say you programmed in all these binary blips, a bunch of 0s and 1s for each goal scorer and every score of the EPL season. Some even say you are a robot. Are you like C3PO? Do you speak bocce?
Mark: Of course I can, sir. It’s like a second language to me…
Jacob: Yeah, alright. Shut up and tell me how you predicted these scores.
Mark: Oh, pardon me sir! At the beginning of the season my strategy started fairly simple, only involving predicting the actual scores, while I didn’t do anything more than guess for the goal scorers. To pick the scores in the beginning, I looked at the home and away records of goals scored for and against for any two teams playing, and came up with an average number of goals I thought the home team would score, and the average for the away team. For example, Man United started the season playing away to West Brom. Man United’s away record over 19 games from last season had them scoring 29 and being scored on 25, while West Brom’s home record had them scoring 30 goals and being scored on 30. So on average I figured West Brom would split the difference between 30 and 25 over 19 games to average about 1.45 goals a game while Man United would split the difference between 29 and 30 goals over 19 games to average about 1.55 goals. So I would pick Man United to win that game.
Jacob: Yes, picking the winning team was for 3 points a game and a big part of winning in the overall points. But how did you pick the actual score? I noticed you sometimes picked 2-1 and sometimes 1-0, only rarely picking 2-0 and never picking any team to score 3 or more goals. Why is that?
Mark: To answer your first question, I used the above average goals each team was expected to score and simulated 100,000 games just randomly generating if a team were to score 0, 1, 2, 3, 4, 5 or even 6 or more goals. Crossing home and away goals and adding up the times the home team won, drew or lost, I could get a good guess which team was going to win most, and again draws were never the most likely event. But I could also see what actual scores were most likely according to the simulation. When both teams were likely to score a lot, usually that meant 2-1 was most likely, but if both teams were not likely to score a lot, usually that meant 1-0, and only when one team was supposed to score over 2 goals on average while the other team was supposed to score less than 1 goal did I pick a 2-0 score. As it turns out, often the most probable score line was 1-1, but because I didn’t want to pick a draw, I chose the most probable non-draw scoreline according to the simulation.
Jacob: But I saw tons of people picking sometimes 3 or 4 goals when Man City or Man United played some crappy team at home, and sometimes they did score 3 or more goals, why not pick it then.
Mark: Just because it’s possible doesn’t mean it’s the most probable event. In all of those games, the average goals they were expected to score were just over 2 goals, and usually that meant scoring 2 goals was more likely than 3 or 4. If you pick 3, it might be almost as probable as 2, but anything more is just a bad bet and therefore I never picked more than 2.
Jacob: But if all your fuzzy math was based on last season, what about all the off season signings and the newly promoted teams?
Mark: That was tricky about the newly promoted teams, so I just defaulted the home and away goal averages to be what other promoted teams had over the last 10 years. And sure last year’s results were only a best guess to begin with, but as the season played on and teams did better or worse than last year, I programmed in those scores to update a simulated season from last year with the actual score already played this year. By the time the season was wrapping up, my calculations were based almost entirely on this season’s results.
Jacob: Great Caesar’s ghost, that sounds like a lot of work!
Mark: Well it was to start with, but consider that once I got the program running, all I needed to do was add new scores each week, click the magic button and wahlah, there are all your results. Compare that to others who might have toiled each week, maybe even looking at statistics like I did automatically, and having to think about it for 38 weeks.
Jacob: So that takes care of the results and actual scores, but where you really shined this season was picking the goal scorers, finishing 2nd in that category among over 600 prognosticators, with 195 points. Divided by 38 weeks, that’s just over 5 points a week! Even if you could pick a Robin Van Persie clone 5 times, scoring on average 30 goals over 38 games, you would still end up with only about 150 points. How did you do it if you didn’t start the season using your program?
Mark: Good math Jacob, you are right, I didn’t start using my programs to figure out the best goal scorers until about the 6th week, when I started with the basic idea that I could sift through all the teams to pick out a handful of good goal scorers per team, look up how often they scored over the past few years or 100 games to get a crude percentage, and multiply by the number of points they get in given week, which would give you roughly how many points you could expect to earn from that scorer on an average week. But since that doesn’t account for whether a team is playing home or away and against a tough defense or not, I decided just multiply that average by the X number of goals the team was supposed to score in a given week. While this doesn’t give you the actual goals a scorer is expected to score, you can now compare all scorers in a given week, taking into account the difficulty of a match.
Jacob: So how did that work out for you?
Mark: It worked out almost instantly. I believe that week may have been the first good week I had of many to come, picking Demba Ba for 3 points a goal, and he scores a hat trick for 9! Then of course other people catch on, start picking him until he is only worth 1 point by week 10 when he scores another hat trick for 3, but I believe I stopped picking anyone for just a measly point by this time, instead moving on to the likes of Yakubu who still was worth 4 points when he scored 4 goals in week 15 for 16 points! I hit more jackpots again with Helguson, Vermaellen, Oxlade-Chamberlain and Sygurdsson while also hitting my fair share of 5 or 6 pointer single goals (McLean, Kightly, Benayoun, just to name a few). The basic idea was to catch the rare hot goal scorer while he had lots of points, and if you see a 6 pointer who is scoring every 4thgame, he might be worth about 1.5 goals a pick.
Jacob: Is this how come you were still picking your nose when the season started, but catching on fire to put the heat on Quique, Weo and Franklin midway through? And how is it that you fell short of first overall?
Mark: I credit some of this work that I put in up front, but don’t regret it, otherwise I might have obsessed each week on who I would pick. I did have to update some stats on a weekly basis, but it was fun to see it pay off. I do regret not starting this from week 1, when I was even picking ties if I remember correctly.
Jacob: And kissing your sister?
Mark {sigh}: No, but I also figured out other tricks that cost me points well into the second half of the season, such as wasting time trying to guess who was starting each week, even not picking hot scorers I had doubts would play, often settling for second rate scorers, or worse, picking scorers I thought would play but didn’t. I noticed the leader “Strev87” one week had picked Berbatov out of the blue for 3 points even though he never starts, and he scored a hat trick for 9 points. It was pretty late when I figured out you can see team lineups about a half an hour before kickoff for any match, and can change scorers right up to the last minute. I really regret not figuring this out sooner, because I now realize most of the leaders in the top 10 never pick players who don’t start, so they must have known about this sooner. I think it could have made a difference.
Jacob: Do you have any more advice?
Mark: My advice to my AtlantaSoccerNews team mates, being fair and sporting so the playing field is level, is to keep an eye out for starting lineups, and have a list of backup scorers ready in case your first picks don’t start. As for coming in 2nd, it was probably too late to catch the leader by the time I figured it all out, but I did notice him picking ties like he couldn’t run out of sisters to kiss, so I think I can best him next year.
Jacob: Aren’t you worried that after I print this story and expose all your secrets, everyone will copy your ideas and maybe beat you next season?
Mark: Sort of, kind of, but not really. First, I like to exchange my ideas as an open source, particularly since these guys are my team mates, and if AtlantaSoccerNews does that much better for the free exchange of ideas, then we will be a hard team to beat next year. As for someone using my own ideas to beat me, I am always finding ways to improve my calculations. At first, I based my goal scorers chances of scoring in a given game on how many goals they scored over the last few years or about 100 games, not even paying attention to those scoring less than 10% of the time. But then later in the season, I found a website that gave how well each player was doing in the EPL this season on a goals per minute basis. I not only considered the long term and current season statistics, but I even compared with some betting websites to see if for some reason some of my top picks were not favored by people putting their money down. As a rule of thumb, if people weren’t willing to bet on it, there may have been a good reason. I have more ideas for the future, so my methods are constantly evolving. I am not worried about others on my team catching up to me, I want a good fight, and I want us to do better as a whole.
Jacob: Yes, you beat second place Franklin by over 60 points and your friends and supposed “Colombian brothers” by about 100 points. Would you say that others in AtlantaSoccerNews couldn’t predict themselves out of a wet paper sack?
Mark: Oh sir, how rude! Might I point out that these three people you mentioned along with myself and any 5th person among 3 who came in close behind would have made the best team of 5 in the entire PredictTheScore?! Mind your manners, sir!
Jacob: Forgive me. One last question, do you think Man United will bounce back next year to reclaim the title from Man City, or maybe some other team?
Mark: Maybe yes, though I’m going to start out next season with Man City as the favorite, since they had a better goal difference, but remember they were tied in points at the end of the season, so it’s going to be close.
Jacob: Would you say then that Man City and Man United were like kissing sisters?
Mark: You could say that, but Man City was the more attractive sister.
Jacob: What about Liverpool? They had a tough season, sacking their manager Kenny Dalglish and having disappointing seasons from the likes of Andy Carroll.
Mark: Sir, the possibility of successfully navigating an asteroid field is approximately 3,720 to 1. That’s about the same as Liverpool winning the English Premier League.
Jacob: Never tell me the odds! So there you have it, the Life and Times of Mark Stevens. Thank you Mark, and may the force be with you {curtsey}!
Mark: And also with you!”
Posts
 as a Statistician, in the Division of Violence Prevention. I am...){kind=link}
Stevie G was the winner amongst the human group...where is my damn trophy!
ReplyDeleteWe had a surprise ceremony for you last night at the Alan Vaughn concert ....but you didnt show up. Lets see if Stevie G can make a run at the Euro 2012 this summer
ReplyDelete