Sunday, February 28, 2010

My strategy: Point accumulation (Part 2 of 7)

This is the second in a seven-part series documenting my fantasy baseball strategy. Part one discussed my keeper theory. Now we'll explore how I utilize roster spots to accumulate points. But be forewarned: You're about to enter the Matrix.


The Urbana Premier League (UPL) where I play fantasy baseball is a 6x6 roto format. Since each team has 24 roster spots, you can think of each team as having 24 active players at any given time, and each player has 6 statistical categories associated with him. So in a way, each manager is managing his own matrix; and as the team matrices accumluate stats and are merged daily, they form a bigger and all-important matrix: the scoreboard.

When I look at my fantasy baseball team, I don't start by looking at the players. I start by looking at an empty matrix of roster spots that needs to be filled by players. Each roster spot is a valuable piece of real estate. Every player on my team needs to have a clear purpose for being there. Even players on the bench. Players who start for my team are expected to produce now. Players on my bench are expected to produce when a starter can't go. Period.

When looking at each starting slot where stats can be accumulated (i.e., catcher, first base, starting pitcher, relief pitcher, etc.), I kinda think of these slots as guns on a warship. Each manager gets the same size warship with the same number of guns (starting slots). Some guns might be bigger and others might be faster. But regardless, I want to be sending out a steady flow of gunfire from every gun on my ship. In order to accomplish this, I have two players available at each position whenever possible. That way if one player has a day off or gets a nagging injury that lasts two or three days, I have a ready backup already on hand. Gotta keep the bullets coming.

In a roto league where each stat counts equally (my league is 6x6) and points are determined by where you rank in each category, my stated goal is to pursue first place in each category. In order to be very competitive in each category, my team needs to be balanced. After all, how can you lead the league in R, HR, RBI, SB, OBP, SLG, W, L, SV, K, ERA, and WHIP, unless your team is good at hitting for power, getting on base, stealing bases, scoring runs, striking out batters, saving games, and keeping opposing hitters off base?

Now some people might say, "Wait a second, if you're truly in a competitive league, it's impossible to lead the league in every category." And to this I'd say that's likely true. But there's a difference between an expectation and a goal. And besides, there's something else at play here. Almost as important as how many points my team accumulates is where the points come from. My main emphasis is to have a balanced team.

Okay, so maybe this all seems obvious. After all, everyone wants to have strong hitting and pitching. But what I'm referring to here goes beyond a balance between hitting and pitching. I want balance from roster spot to roster spot. How do I achieve this? Whenever possible, I want players who can rack up points in multiple categories, and it's a bonus if those players have eligibility at multiple roster spots (positions). That way, in a sense, the points that my team puts on the scoreboard can come from anywhere and everywhere.

The reason that I want the best set of bench players possible is fourfold:
  • First, I want to foster competition at each roster spot. If a player slumps, then I'll put a hotter player in and won't reinstate the former starter until he plays his way back into my lineup.
  • Related to my previous point, I also like setting up platoons at as many roster spots as possible. For instance, last year I noticed that Michael Bourn's OBP against lefties was terrible, but against righties it was okay. So whenever he started against a lefty, I tended to bench him in favor of a better option. If you're able to play the percentages to your advantage, then why not?
  • Third, in the UPL (and most leagues) there is a cap on the number of games that can be played at any position. Most fantasy managers don't even come close to maxing out all of their "innings pitched" or "games played" at each position. However, by having multiple players eligible for each position, you can come close to maxing out each position at season's end. By doing so, you'll squeeze out a few more points along the way. 
  • Fourth, and perhaps most importantly, my strategy assumes injuries. I don't know when or where, but injuries will happen. When they do, I want to have a solid short-term solution in place. Necessity is the mother of invention; and like the Internet, my team is built to survive a disaster.
As described above, my team mimics a greedy algorithm by going through every single day trying to maximize production from every single roster spot. My team also is like a distributed system in how I try to avoid becoming too dependent on one or two players for overall team success. So if my strategy really can be summed up in this way, how might we expect my team to perform in a given season? In theory, by deploying a greedy algorithm, my team should get off to a relatively quick start; and when it loses a key player to injury, while the loss hurts, the distributed system of the team should still be able to play in a competitive manner.

So what did my team do in 2009? Well, I blogged on May 6 that my team had gotten off to a quick start. At the time, little did I know that on the following day my #3 pick Manny Ramirez would be hit with a 50-day suspension, and then the day after that my #4 pick Aramis Ramirez would hurt his shoulder and land on the 60-day DL. Sure, that made me nervous. But I had already done the best realistic planning I could possibly do for such an unexpected situation. I shifted some of my team's parts around, and we stayed competitive. Later in the season my #2 pick Carlos Beltran also was lost to injury. However, my team was still able to hang on to win the championship.

Now don't get me wrong. The fact that my team finished in first proves nothing about my strategy. The key here (at least to me) is that my team was able to get off to a quick start, and then despite significant personnel losses, stayed competitive. Those characterics seem line up with a strategy that claims to emulate a greedy algorithm running on a distributed system.

Earlier I mentioned that when I look at my fantasy baseball team, I start by looking at an empty matrix of roster spots that needs to be filled by players. But of course, things get more complicated when you start looking beyond an empty matrix. With a 24-man roster in our 6x6 UPL league, every team's matrix is 24x6 if you're looking at each player's counted stats. Although, there are really more than 6 numbers per player. We can add age as a number, plus we could try to rate how players are trending at any given time or in which matchups they excel. That 24x6 matrix can start getting much bigger when you consider all the ways a manager can rate the players.

But let's keep this simple. Fantasy baseball always comes down to filling the empty matrix of roster spots with players. In order to do this well, we need to do a good job of rating the players. In the next part of this series we'll discuss hitter evaluation; and then in part four we'll look at pitcher evaluation.

4 comments:

Greg said...

Post script:

One of the reasons that I enjoy roto baseball is that it requires more than just baseball knowledge. It's also a math challenge.

The Chairman, a.k.a. O.N. Thugs, has actually written about this very subject. He begins the topic by saying:

One of the most important things that you can do in a Roto league is to figure what you will need to win. Those requirements then help you narrow down what you can reasonably try to do, as you shape your overall strategy.

Here's how I'd summarize the Chairman's strategy in that post:

1. Study scoreboards from previous seasons of your league
2. Make judgements as to what will likely be needed to win this season
3. From the limited resources available, prioritize your efforts to gather only what's necessary to beat all of your opponents into submission

The Chairman concluded with:

So why bother with this post? I think that it's useful to frame how you plan out a season, so that you have a better idea as to what moves you forward, and what does not.

What's my take on the Chairman's strategy? Well, it's a fact that his O.N. Thugs are the most successful team in UPL history. Not only is his strategy sound, it reminds me of something that I learned in college.

In one of my software engineering classes, we studied a software development methodology called the Rational Unified Process (RUP). The idea behind RUP is to carefully study what the entire project will require, model all of the concepts visually, test and revise the model for accuracy and efficiency, then finally start coding the software. In short, the meticulous planning at the beginning of the project should result in increased efficiencies once it's time to actually write the program.

The Chairman's general approach to a roto league reminds me of RUP. He studies the big picture and what needs to be accomplished, searches for efficiencies, and then uses what he's learned to frame his more specific strategies that will be implemented during the season.

In that same aforementioned college class there was another software methodology taught, called Extreme Programming (XP). As I recall, the professor explained XP in one sentence: "Do the easiest thing first."

Compared to RUP, XP is almost comical. Whereas RUP preaches extensive planning, modeling, and testing before any software is even written, XP would have you glance at the project's big picture, figure out the easiest thing to accomplish, get it done, and repeat. What's funnier is that my professor said that both methods were equally valid in the sense that both have their strengths and weaknesses. Much of it really just depends on your own personal preferences.

My general approach to a roto league season is probably closer to XP. Rather than studying how many total points will likely win the league or estimating what my team's stats should be at season's end, my gaze is more short-term in nature. In fact, starting out, I'm not even looking at numbers. I'm viewing roster spots... empty roster spots.

Chairman said...

Intriguing post and PS, my friend.

I'd argue that a RUP is harder to do right up front, but once you develop the fundamentals, then expertise takes over, and then you're using expert heuristics.

Heuristic thinking is good because it's quick and easy. The problem is that novices employ heuristics that aren't great for problem solving, whereas experts employ effective heuristics.

And let's be honest. Given my recent drafting history, I'm not doing the legwork like I used to, though my success rate is still high. Could it be higher? Yeah - I haven't dominated quite as much as I have in the past. But the success:effort ratio is much higher, I think.

That said, I like my team this year, though I may have to post an updated Math for Winners posting.

-RG

Chairman said...

By the way. I had to go look up how to put hyperlinks into these comments, which I had never bothered with until now... be proud that this sort of effort came in this forum.

Greg said...

Admittedly, my success:effort ratio wouldn't have led the league last year. That honor probably belongs to Westy, who scored a hundred just doin' nuthin' in the draft. ;-)