Catcher, 1B, 3B: Biggest surprises (ottoneu lwts)

We’re a third of the through the season, so I’d like to take a look at which players have provided the best production relative to their auction cost in ottoneu leagues…and by extension, probably much of fantasy baseball.  While this is specific to ottoneu, I think this will apply to most of fantasy baseball.

We’ll start today with catchers and corner infielders.  I’ll give you the top three values at that position, plus my pick for the player of those three that is least likely to regress (or, perhaps, the player likely to regress the least).   Avg. Cost is just their current average cost in ottoneu, whereas the “Performed As” number is essentially an measure of what the dollar value would be for performance at this level for an entire season (using the lwts-based FanGraphs Points system; methods description at the bottom).  Expect this to change by season’s end: these guys are all overperforming, and are all good bets to regress to some degree.


Alex Avila, DET
Avg. Cost: $1.33
Performed As: $26
Value: +$25
Ramon Hernandez, CIN
Avg. Cost: $1.24
Performed As: $22
Value: +$21
Russell Martin, NYY
Avg. Cost: $3.86
Performed As: $22
Value: +$18

My Pick: Ramon Hernandez

Granted, I’m a Reds fan, so maybe I’m just a homer.  In truth, I’m not wild about any of these three: I’m not as high on Avila as most seem to be, and I expect Martin to break down any second now.  But Ramon Hernandez is having a second consecutive excellent season with the bat.  His timeshare with the Reds seems to have helped him stay healthy and rested over that span.  His .231 ISO to date would be the best of his career.  While that 23% HR/FB ratio is bound to be much closer to his career rate of 11.6% the rest of the way, it’s still true that he’s shown opposite field power this year that he hasn’t demonstrated since he was in his prime.

He’s not walking much (BB% might actually go up?), but his strikeouts are down.  I think he’s a good bet to finish with a bit better than league average numbers in ~350 PA’s.  Some fantasy owners might not be excited about Hernandez given that he’s a close to a 50% timeshare, but remember that ottoneu caps your two catchers at 162 games total.  If Hernandez can produce at anything even close to what he’s done over the remainder of the season, that’s a heck of an asset to have.

Others: Jonathan Lucroy (+$17), Chris Iannetta (+$15), Mike Napoli (+$14), Yadier Molina (+$12), Miguel Montero (+$10), J.P. Arencibia (+$10)


First Base

Gaby Sanchez, FLA
Avg. Cost $4.86
Performed As: $41
Value: +$36
Todd Helton, COL
Avg. Cost: $1.79
Performed As: $23
Value: +$21
Brett Wallace, HOU
Avg. Cost: $2.22
Performed As: $23
Value: +$21

My Pick: Gaby Sanchez

I don’t think Sanchez will post a .401 wOBA by the season’s end.  But I think, he’s a solid hitter who has shown some meaningful improvements.  Most appealing to me is the uptick in his walk rate, and simultaneous down-tick in his strikeout rate.  This is driven by what looks like a change in approach, as he’s swinging at almost 5% fewer pitches than last year.

Also encouraging is the uptick in his HR/FB%, which has jumped from 9% last year to 12% this year.  It’s entirely possible that the power boost is sustainable–it’s pretty modest, and perhaps he’s swinging only at pitches he can handle, resulting in better contact and better power (even if not more contact overall).  I’m not sure if he’ll cross the 30 home run mark this year, but he’s got a shot at it.

Others: Ike Davis (+$20), Paul Konerko (+$17), Justin Smoak (+$14), Juan Miranda (+$12), Joey Votto (+$6)

Third Base

Jose Bautista, TOR
Avg. Cost $29.72
Performed As: $109
Value: +$79
Alex Gordon, KCR
Avg. Cost: $5.07
Performed As: $37
Value: +$32
Greg Dobbs, FLA
Avg. Cost: $1.00
Performed As: $22
Value: +$21

My Pick: Jose Bautista

You’d never pay $109 for a player.  But that’s because you could never reasonably project that a player would perform at the Bondsian levels that Bautista has over a full season.  That said, of the three, he’s the guy I’d most expect to exceed his salary (which is already moderately high) the rest of the way.  I still expect him to regress, and perhaps regress considerably.  But he’s been so ridiculous that a bit of regression would still have him one of the best fantasy players in baseball.  I mean, if nothing else, pitchers are likely to keep pitching around him (38% balls in zone this year vs. 46% league average) for the foreseeable future, so the walks should be there.

THT’s Oliver projects him to hit .274/.382/.579 the rest of the way, while ZiPS has him at .272/.395/.571.  Those seem reasonable, and that level of performance is easily worth $30 over a full season.

Others: Casey Blake (+$19), Chipper Jones (+$15), David Freese (+$14), Chase Headley (+$14)

Methods (you can skip this if you don’t care!): here is what I did to generate the performance $’s.  First, I used average auction values to identify replacement level players at each position, which I defined as players averaging less than $2.  Based on their average production this year, I calculated a replacement level baseline average points per PA or IP at each position (note: there is selection bias here, but this approach worked better than others I tried).  Next, I calculated the extra points per game that players have produced for their owners above those replacement levels.  And then, I used model II regression to calculate an equation comparing average extra points per game to player cost, which allowed me to find an average $ per extra point above replacement players conversion.  Finally, knowing a) replacement production, b) an average marginal points per marginal $ conversion, and c) actual production for each player, I could calculate a “production salary” and compare that to their actual salary.  If you’d like to know more, feel free to ask.

Print This Post

Justin is a lifelong Reds fan, and first played fantasy baseball on Prodigy with a 2400 baud modem. His favorite Excel function is the vlookup(). You can find him on twitter @jinazreds, even though he no longer lives in AZ.

12 Responses to “Catcher, 1B, 3B: Biggest surprises (ottoneu lwts)”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. prankmunky says:

    I’m glad I drafted Gaby Sanchez. I thought he might have a little bit of room for improvement after last year. Unfortunately his walk rate seems to be fighting itself: “Most appealing to me is the uptick in his walk rate, and simultaneous down-tick in his walk rate.”

    Vote -1 Vote +1

  2. LuckyStrikes says:

    Great article. We looked at this in our league recently but looked at the worst “busts” so far instead. A simpler approach, we just evaluated “points per $1”. The five worst:
    Hanley Ramirez: 3.19 points per $1
    Jason Heyward: 4.02 points per $1
    Albert Pujols: 5.08 points per $1
    Carl Crawford: 5.09 points per $
    Dan Uggla: 5.77 points per $1

    This makes guys like Smoak, G. Sanchez, and Gordon seem all the more valuable.

    Vote -1 Vote +1

    • Justin Merry says:

      Yeah, I’m planning to do a bust series also. These guys all figure prominently. I’m lucky enough to have Uggla, Pujols, and Mauer on my team!

      Vote -1 Vote +1

  3. Justin Merry says:

    In case you’d like to play along at home, here’s what I did.

    First, these are my replacement levels, calculated as points per PA or points per IP:

    C 1.009
    1B 1.040
    2B 0.905
    3B 0.909
    SS 0.905
    OF 1.024
    SP 3.794
    RP 5.108

    SS was weird because replacement SS’s were actually better than replacement 2B’s or 3B’s. This occurred every way I tried to do it. I think that must be a quirk of the dataset, so I instead set replacement level at SS = 2B. C also seems high, but there’s a lot of good, $1 catchers out there. Problem is that catchers don’t play a whole lot.

    To calculate $ values:
    1. Figure your focal player’s points per PA (or IP) and then subtract the above replacement level from that number.
    2. Multiply by PA (or IP). This is “points above replacement.”
    2. Plug into this equation: $ = 1 + 0.3138 * Points

    It’s not perfect, but it works pretty well. The .3138 coefficient will change (decrease) every day, as it’s fit to the current point totals.

    Vote -1 Vote +1

    • mikedouglas says:

      Wouldn’t the appropriate measure of position player performance be on a points per game basis (as that’s the limiting factor in ottoneu)? Your current system penalizes players on less out-prone offences and those who hit higher in the batting order.

      Maybe the difference is small enough that it doesn’t matter, but it might be something to look into.

      Vote -1 Vote +1

      • Justin Merry says:

        In theory I completely agree, and it’s what I initially did. What happened, though, is that players with a lot of PH appearances or defensive substitutions saw a big hit on their points per game, especially early in the season. From a value perspective, given that it’s the games that matter in ottoneu, that’s fine. But I was worried it was overly deflating my replacement level benchmarks, because replacement players were more likely to have a higher % of PH appearance games than more expensive players. So, I opted to go with per PA.

        There’s a more time consuming way to get around this (maybe look only at GS by replacement players, for example), but I didn’t have time/inclination to go that route…

        Vote -1 Vote +1

  4. descender says:

    and I expect Martin to break down any second now.

    For a guy with 1, count em… 1 injury shortened season in his career, this reputation is getting out of hand.

    Vote -1 Vote +1

    • Justin Merry says:

      You could well be right about this. I have to keep reminding myself that he’s only 28 too. Somehow it feels like he should be a lot older than that.

      Vote -1 Vote +1

      • descender says:

        He’s 800 in catcher-years thanks to Joe Torre :)

        I just see this sentiment about him all the time and it blows my mind. The guy was overused by Torre, this is clear. No one should be playing 150+ games at catcher. He had one injury that led to another… and missed some time. That has turned into a “history of injury problems” around the media-fed world some how.

        He looks pretty healthy to me, and he started stealing bags as soon as the season kicked off. After watching Posada all of these years he looks like the Flash out there. If he was concerned and hesitant I might buy into the injury hype… but right now it looks like the “Steal of the off-season” to me.

        Vote -1 Vote +1

  5. Zeezil says:

    I don’t know what model II regression is, but I assume you’ve estimated a linear regression between average points per game and player cost. However, it seems to me that this relationship is nonlinear. Player cost per point is likely increasing as production rises from replacement level, as it does with actual MLB salaries.

    Also, do you estimate cost per extra point separately per position or did you pool positions?

    Vote -1 Vote +1

    • Justin Merry says:

      Model II regression is the same as reduced major axis regression if that helps. It is a different way of calculating a regression line than least squares, and is better when you’re interested in the slope specifically as opposed to prediction. Because of the scatter, an ordinary least squares approach results in far lower slope estimates.

      I’m actually calculating production above replacement this in a manner that very closely mirrors how we calculate WAR. And it is not necessarily true that salaries increase with production in a nonlinear way in reality–we do not assume it does in the FG salary calculations for that reason.. Tango’s done a lot on this. That said, I did try a couple of alternative fit models, and none seemed to clearly be a better fit for the data, graphically or in terms of explained variance To be sure, however, with so much scatter (it’s still early), it’s a but hard to tell.

      The $ per point conversion is done across all positions, because they are all competing for your same pool of $. And, once you take into account different baselines across positions, a point is a point is a point.

      Vote -1 Vote +1