Fantasy Cornerstone: Buy low, Sell high

I don’t intend for this to be part of a series of fantasy baseball advice posts.  It’s not like you should listen to me as opposed to people who write for ESPN, Sportsline, or whatever other site you use.  After all, I’m in eighth place right now in my 10-team league.  Still, that won’t stop me from trying.

An important concept to realize over the first month of the season is that the stats are virtually useless for telling you who’s going to be good the rest of the year.  The sample size is just too small to say anything meaningful.  Then again, teams will make decisions on players based on early-season performance, so you’ll see a long hook for guys like J.J. Hardy and Heath Bell, who are probably playing a bit over their heads right now.

Hardy is especially important to me because he’s on my team this year.  My league is one of many online leagues where it’s very hard to trade, but I swapped Henry Owens for him straight up a week or two ago, and I’m already looking to deal him again.  It’s possible that he’s breaking out right now, but I’d rather move up in the world of shortstops (I already did, since I had Omar Vizquel before) than hang on and find out.  I’m looking for guys like Jimmy Rollins and Jose Reyes, if the other owners will even listen to me.

It’s important to be able to identify players like Hardy, since there’s more to buying low and selling high than just looking at the players’ basic stats.  I’ve used a little Excel trick (web queries) to create a master stats file for 2007.  It grabs current stats from major sources like Baseball Prospectus and The Hardball Times and puts them in an easily sortable Excel file, which I can then manipulate in whatever way I want.  It makes sifting through the current stats easier than just looking at players’ profiles on my league homepage, and it tells me a lot more, too.

For instance, I can see that J.J. Hardy’s PrOPS (predicted OPS based on batted ball data) is .903, a full 100 points short of his current OPS.  PrOPS is a great tool for making decisions early in the season, since it helps sort out some of the inconsistencies of simply looking at a player’s actual production.  The more accurately I can figure out a player’s actual performance, the better the value I can get for him.  While I might be able to expect Hardy to settle in with a .900 OPS (still great for a shortstop, by the way), I might want to move him in case someone else sees his .339/.396/.606 line and thinks he’s a superstar.

At the same time, I can also figure out players on whom I should buy low.  If someone’s under-performing on his OPS (a negative OPS-PrOPS) by a significant amount, he’s likely to recover well (assuming his team hasn’t already benched him by then, a la Wilson Betemit).  Fortunately for me, Ryan Howard, my #4 overall pick, is on the short list of players who should return to their normal level of solid performance (PrOPS of .979 compared to an OPS of .781).  For the same reason, Chris Snelling may have been a great buy-low guy for Oakland.  His .875 PrOPS in the NL will make him a solid reserve for Billy Beane if the A’s ever get healthy.

For pitchers, I think the best way to evaluate over- or under-performance is through DIPS (defense independent pitching) theory.  A good portion of the variation in pitching stats is due to the performance of the defense.  While it’s hard to find a good version of DIPS online, THT does a pretty good job estimating it with Tangotiger’s similarly-named FIP (fielding independent pitching).

FIP is a very simple formula that uses only walks, strikeouts, and home runs allowed (the “three true outcomes” that a pitcher can control best) to predict future performance.  It does that well, just like PrOPS, actually predicting future ERA better than ERA itself.  THT even goes a step further with an “expected” FIP based on even more factors.  The bottom line is that you can use either in a FIP-ERA formula to figure out the good buy-low/sell-high players at this point in the season.  The more luck you can filter out of a player’s performance, the better off you are.

The top names on the buy-low list includes another one of my own guys, Milwaukee’s David Bush.  With a 6.03 ERA, you’d think he’s a prime candidate to lose his rotation spot once the Brewers call on Yovani Gallardo (a current stashee on my roster, and one of the consensus top pitching prospects in baseball).  However, Bush has an FIP of just 3.45 and an even lower xFIP of 3.28.  Needless to say, I won’t be falling for anyone trying to use this strategy against me with him.  Bush’s track record and solid command so far this year suggests he’ll turn things around, which is more than can be said for another prominent Bush right now (sorry, I had to).  Other guys on my list include Aaron Harang (FIP of 2.64), Matt Belisle (2.38), and Anthony Reyes (4.01, and another one of my guys).

On the other end, I’d sell high on Brad Penny (FIP of 3.63 compared to a 1.64 ERA), Tom Glavine (3.27/5.06), and Rich Hill (1.73/3.51), the last of whom I’ve been aggressively shopping from my own team.

This strategy is one of several key areas of in-season fantasy play, and it’s perhaps the least used, from what I’ve seen in fellow reactionary owners.  While it’s fun to ride out a player’s hot streak, the chances are that it will end (and for some, badly).  Some fantasy owners are good at scouring the waiver wire for prospects and hot players, while others make the most of playing the right matchups.  After the draft, there’s still plenty of work to do, and mastering this area will help you stay on top all season.  Or, if you’re in my boat, it can at least give you hope that you won’t be in 8th place all year.  At least that’s what I’m telling myself.

(NOTE: Stats are through Sunday, 5/6)

Analyzing the standings before the season

It would seem like an exercise in futility to try and predict your fantasy league’s standings before the season even starts, but to me it’s actually worthwhile.  Actually, I find it important to do so during the draft, because it can help you see which categories are worth chasing the lead in and which you can ignore for a while.  Last night, it helped me see that I was high in the rate stats (batting average, ERA, and WHIP), so I could focus in the later rounds on power categories, maybe steals, or strikeouts.  The result is a fairly balanced team with a chance to win pretty much any category except saves, which (as usual) were highly over-valued.

For all this talk about my detailed draft strategy, my team is perhaps not as good as I would like.  In years past, I’ve predicted that I would get 80 of a possible 100 points (scoring goes down from 10 to 1 in each category based on team standings), which would destroy the competition in most leagues.  This year, I’m predicting just 68 points, but enough for a win in a very close league.

The way I make these predictions is not just by taking the league’s standings based on my player projections, although that is the main idea.  I actually take each team’s z-score for a category, which tells me how many standard deviations above or below average each team is in each category.  While that means five teams could finish with six points in a category in my estimates (in other words, it’s not going to be accurate in that way), it seems like a better way to get a feel for where each team stands at the start of the season.

Below are my predicted standings using that method:

As you can see, I actually do predict myself to finish first, so the draft wasn’t a total disaster.  (Since I’m using my own projections for this analysis, it would be a disaster if I didn’t predict that.)  The column for “Order” is the team’s draft position.  The draft was “snake-style,” where the teams reverse the selection order for even-numbered rounds.

From these standings, it looks like I’m only slightly above average in power and steals, and actually slightly below average in runs.  Saves are a loss for me, unless both Wheeler and Linebrink become their teams’ closers mid-season.  Moving up from there, I’m moderately above average in strikeouts.  Even though I have seven starters, having a guy like Maddux reduces my potential there.  I’m hoping Rich Hill really turns it on, though.  Wins, ERA, and batting average show me as about a full standard deviation above average, which is pretty comfortable.  Then we have WHIP, where I expect to win the category unless Lieber’s replacement is a very low-end starter.  The total of all this comes out to a win for me, and a second place finish for my friend (and #5-picking draft neighbor Doug).

If I can work out the details on a better estimation method than this (I have an idea for one, but I’m not sure it will work), I’ll post those later today.

The 2007 jfwiii All-Stars

Or, in other words, my fantasy team.  I drafted my team last night, and as usual, there were plenty of surprises during the draft.  I’ll get to those in a minute, but first, I have a disclaimer/explanation:

I know no one really cares about anyone else’s fantasy baseball team, unless it’s another team in your own league.  Fantasy baseball is arguably the lowest-maintenance fantasy sport, especially when you play with rotisserie (category accumulation) style scoring.  That’s what I do, and I play in an NL-only league, which makes my team even less interesting to some baseball fans.  At any rate, this post is mainly here to serve as an introduction to the “meat” of what I want to write about, which is my fantasy strategy, rather than just my fantasy team.  I feel like I need to share the team details first, however.

So, here’s my lineup for a 10-team NL-only league with 5×5 scoring (which, if you’re not familiar, includes batting average, runs, homers, RBI, steals, ERA, WHIP, wins, saves, and strikeouts):

C – Ronny Paulino (PIT)
C – Miguel Montero (ARI)
1B – Ryan Howard (PHI)
2B – Ray Durham (SF)
3B – Scott Rolen (STL)
SS – Omar Vizquel (SF)
OF – Jason Bay (PIT)
OF – Jeff Francoeur (ATL)
OF – Willy Taveras (COL)
OF – Chris B. Young (ARI)
OF – Corey Hart (MIL)
U – Scott Thorman (ATL)
U – Wes Helms (PHI)

P – Ben Sheets (MIL)
P – Dave Bush (MIL)
P – Anthony Reyes (STL)
P – Rich Hill (CHC)
P – Greg Maddux (SD)
P – Dan Wheeler (HOU)
P – Orlando Hernandez (NYM)
P – Scott Linebrink (SD)
P – Jon Lieber (PHI)*

*Will be dropped this week for an injury replacement…he was my last pick anyway.

Overall, I’m mildly pleased by my team.  I thought the rest of the league did a fantastic job at the draft, with at least 4-5 other owners consistently taking picks off the top of my board.  I’m pretty solid in every category, perhaps with an emphasis in batting average and ERA/WHIP.  Obviously I’ve pretty much punted saves, but I do have two closers-in-waiting (Wheeler and Linebrink) who should help me out in the rate categories even if they never finish a game.

Where did I go wrong?

Well, the #4 pick may have been the first place, because I didn’t take the guy who was theoretically at the top of my draft board.  Every year, it seems like pitchers are undervalued, but when I follow my spreadsheets, I always end up way ahead in pitching and needing to make up ground in hitting.  I decided to adjust for this in my head and take an unconventional-for-me balanced hitting/pitching approach.

For a few years now, I have been using the method that noted sabermetrician Tangotiger seems to think is the most reasonable way to value players in fantasy.  Basically, the idea is to take the players you expect to be drafted, find the standard deviations of each of the stat categories, and use that to figure out how many SDs above average each player is in each category.  The mathematical way of saying that would be to find each player’s z-score for each category, and add them all up into one rating.  You also have to find a player’s real contribution to the team in batting average, ERA, and WHIP, rather than just using the rate stats themselves.  Tango explains all that and more in the post linked above, and you end up using what he calls xH (and I guess x.  Then you adjust for positional strength (I use replacement level at each position) and you’re done.  It’s actually simpler than it sounds, although I make a few additional adjustments during the draft to reflect my own team’s changing composition.

As it turns out, I had a bug in my spreadsheets that may have caused me not to draft ideally from the middle rounds on.  I pretty much had to wing it at the end of the draft because of this bug, but I don’t think it affected my team substantially.  It caused me to weight several of the categories and one of the positions improperly, but since I was mainly looking for players at specific positions at that point, it probably didn’t affect me too much.

Next, I’ll look at the standings I would predict for my league using my own player projections.  It would be a shame if I didn’t come out on top using my own system, but we’ll see how I did.