A nerd, a computer, and the Playoffs

Going into any new season, it’s comforting to know that your team has a shot at the playoffs. Bryant described my interpretation of the Mariners’ Marcel projections as “sobering,” and rightly so. I think a reasonable expectation is 75 wins, and 75 wins is not fun to think about. However, that doesn’t mean there is no chance at the playoffs. 75 wins is only a (semi) educated guess, and there’s a little thing that trickles into sports from every which way known as randomness.

The term randomness gets a bad rap from a lot of sports people. Some like to believe that 100% of what happens on the field is the product of player and managerial skills. I just don’t buy that. I think the game-theory nature of baseball—obviously a large chunk of the game itself is batter-pitcher battles—lends itself to a lot of randomness. I’m not necessarily calling it luck, just consistent unpredictability. Is that an oxymoron? Whatever.

One of the unique aspects of baseball is the constant flux of pitcher matchups. Very few team sports are as dominated by one player as baseball is—though that player can only start once every five days. It means that a team’s chances of winning aren’t even close to consistent. Felix versus Colon? I’d give the M’s a 70% chance to win. Millwood versus Jered Weaver? 30%? Maybe…20%? This variation means that occasionally, for stretches of time, teams are able to both over and under-perform expectations. The M’s caught the randomness bug in 2010 when some thought they could win the division, yet they lost more than I’d like to remember. Just the year before, I would argue that variability helped them stay in the playoff race for a while. My point? Mediocre teams can be good enough for a while to sneak into the playoffs because of the variable nature of this sport.

That's me!

I think the Mariners are about a 46%-winning team this season—75 wins. I also think they are a variable team, meaning that they are subject to hot and cold streaks more so than an established team like, say, the Red Sox or Yankees. I think the Mariners will see their theoretical winning percentages vary from as low as 25% to as high as 65% on a regular basis. So here’s what I did. I ran a computer simulation for this type of team—a team with subpar expectations and high variability. This team played 100,000 seasons, I recorded its wins, and then all the hypothetical players promptly took a long vacation.

So what did my simulation tell me? The pretend Mariner team I created won about 75 games on average, as you might expect. However, due to high variability, there were a number of seasons they made the playoffs. History shows us that in the last 10 years, a team would need to win between 86 and 93 games to capture that new playoff spot. Following that range as a guideline, this computer Mariners team made the playoffs nearly 17% of the time. Yeah, that’s not great if you’re a Yankees fan. But coming off back-to-back 60-win seasons I’m giddy about 17%.

The program was simply designed to mimic the random fluctuations in the game of baseball. I’m pretty sure baseball players are not guided by random number generators (p-value = .05), but simulations can at least help to model potential outcomes. Maybe Ichiro’s BABIP shoots back up to what we have come to expect…or maybe it doesn’t. The simulation effectively allows for things like that to happen every so often within those 100,000 seasons.

For those of you who don’t like my assertion that the M’s are a 75-win team with high variability, then take a look at the chart below. I ran simulations for five other teams with different expectations and variability, and then recorded the proportion of times they made the playoffs.

Expected Variance Bad Good Playoffs
40% Low 30% 50% 1.3%
40% High 20% 60% 12.3%
45% Low 35% 55% 6.9%
45% High 25% 65% 16.8%
50% Low 40% 60% 22.9%
50% High 30% 70% 31.1%


I hope the chart isn’t too cryptic. The teams “true ability” is the expected column. Then I show the variance as a range of expectations game-to-game. Row one, for example, suggests the team will average a 40% chance to win, but will often vary from as low as 30% to as high as 50% depending on the matchup. That type of team made the playoffs 1.3% of the time.

What the chart really articulates, in my mind, is how the variability of baseball can make all the difference. Playoffs are not likely, but possible, and that’s what I like to hear.

Tags: Computer Simulation Featured Playoffs Popular Seattle Mariners