Share on FacebookTweet about this on TwitterShare on Google+

One of the bigger challenges for fantasy baseball drafts and auctions is how to value players who are not projected for a full-season’s worth of at-bats or innings. If you run their partial-season projections through any standard player rater, they are unfairly penalized as it assumes zero replacement stats when we know we can get someone on waivers to fill in for them while they are not on the major league roster.

These types of players fall into four buckets:

Type 2017 Example
Injured Wilson Ramos
Injury-Prone Lance McCullers
Prospect Austin Meadows
Unclear Playing Time Mallex Smith

One approach to handling this type of player is to neutralize the playing time component. FanGraphs has a Steamer600 projection which projects all players to 600 PAs. I do something similar to show a full season worth of PAs against RHP vs LHP on the player pages. You could also accomplish this by just multiplying all player projections by PA * 600/PA (or something equivalent for ABs and IPs).

But there are two key flaws with these approaches:

  1. A players PA/IP is not based on just the number of games they play. It is also based on their role in the batting order (hitters) or their average IP/Game. It is insane to look at Jose Altuve (guaranteed 1-3 hitter) and Alcides Escobar (possible top of lineup (Yost!!!!), possible #9) as having the same number of PAs when Altuve should have a higher PA/Game. The same would go for Clayton Kershaw projects as a 7 IP/GS pitcher where Lance McCullers is probably lucky to get to 6 IP/GS.
  2. At least for my projections, I adjust the RHP/LHP for part-time hitters to account for the likelihood that managers will platoon them smartly. So a left-handed hitter I project as starting 75% of the time will have a higher percentage of PAs against RHP than Anthony Rizzo. Thus, multiplying that player’s stats up will exaggerate their success.

I think we have come up with the most straightforward way to handle the above with the $/Game metric that is on the player pages (both for LHP/RHP on hitters) and in the Player Raters.

The $/Game metric represents the estimated Roto value per game for a player. For hitters at the top of the player rater, this will be slightly lower than their standard $ estimate since part of their value is based on their playing time advantage versus other players. Conversely, as you go down the player rater, the $/Game metric becomes greater than the standard $ estimate for hitters. For pitchers, the $/Game is higher than the standard $ value for some of the top pitchers and then foll0ws a similar pattern for hitters.

Here is how I use it for the above cases:

  1. Is the $/Game metric positive? If I am going to roster a player who cannot play starting game 1, he better be projected as a greater than replacement player ($1=replacement). At my 256 PA projection, Mallex Smith looks awful at -$9 but his $/Game shows he is a $9.50 player in 15-team leagues. Clint Frazier is at -$0.2 – no reason to look at him except if you are in one of those 50-round drafts.
  2. How do his RHP/LHP splits look (for hitters)? I can click on his name in the Player Rater to look at his platoon splits on the player page. For Mallex Smith, his value is tied up in SBs and, like most marginal LH hitters, he is really only fantasy useful when he faces RHP. His $/Game already bakes in the likelihood he would be platooned. If you think a player will be everyday, it’s best to just multiply the $/Game for RHP at .75 and the $/Game for LHP at .25.
  3. Create a revised $ metric to account for replacement stats. Let’s say you think Wilson Ramos will come back at the ASB. His $/Game is $4.1 for 15-team mixed. If you think you can backfill with replacement stats ($1), then that frankencatcher is $4.1*.5 + 1*.5 = $2.05. Lance McCullers is a $17/GS pitcher according to the 15-team projections but injury concerns bring him down to $10.8. But say you think he misses 5 starts and you can stream a $2 pitcher, then his true value is $17*.85 + $2*.15 = $14.75.

This is the same logic I used for selecting Kris Bryant in early 2015 drafts – almost winning ToutWars thanks to that pick and winning an NFBC league (intel that he was getting called up after the first 15 days helped). It is the same logic that makes me more comfortable gambling on Stephen Strasburg because his $/Game is elite (6th best for SP at ~$30 for 15-team, $5-$6 better than similarly drafted pitchers in Price, Cueto, and Verlander) and I am willing to take the over that he could make 5/6th the starts of those other pitchers.

Hope this helps and please ask questions in the comments.

 

   
  1. Jbona3 says:
    (link)

    So is it fair to cross-compare positional players and pitchers? I’m trying to evaluate a trade in a keeper league that centers on getting Moncada ($10 this year) for Carlos Martinez ($13) and based on the Per Game $ C-Mart is about double the value ($5.9 vs. 11.8). Is this a fair interpretation of the data and how to use it?

    • @Jbona3: For hitters vs pitchers, I’d compare on $ vs $/Game. For pitchers who aren’t projected at full-time, I’d use the $/Game and then discount based on durability concerns.

      • Snuka says:
        (link)

        I know you will eventually post the years roto category targets next month cause you are awesome, but was wondering where you get the final results from so many leagues? Does Yahoo make that information public?

        • I model the results for the rest of the formats. I have tested it through the years and it does quite well.

          • Snuka says:
            (link)

            @Rudy Gamble: Agreed, your targets are the best in the business. But where do you get the initial data? If for some reason you stopped publishing I would like to know where you get any quantity of league results if you are willing to share.

            • Thanks. The only stash of league results are the Razzball Commenter Leagues (12 team ESPN) and NFBC (15 team mixed). The results I share are based on a model that looks at the projections, determines rosters universe, and distributes the stats across the teams. I prefer this approach to using actual results because it adjusts for the hitting/pitching environment in the projections (in my case Steamer with some adjustments) which is always a little different than the previous year.

  2. Drunk Mets PR Team says:
    (link)

    This is incredibly helpful, thank you.

    Would you say that $/GM should be weighed more heavily when planning for H2H leagues? I’ve been wrestling with how to account (if at all) for the stats resetting every week.

    • @Drunk Mets PR Team: i wouldn’t view H2H differently. the biggest factor is whether your league allows Daily roster changes (even if it’s just swapping guys from bench to active) vs weekly changes. If Daily, $/Game becomes more important b/c you have the option of benching platoon players. If Weekly, the delta b/w $ and $/Game is relevant for platoon players because it includes the penalty for a guy sitting during the week.

      But when deciding on a late round pick, I’d veer towards $/Game – particularly for prospects or guys who could become full-time players.

  3. Milarky says:
    (link)

    Hey Rudy, curious as to your thoughts on coming up with category targets. Our roto league is a bit unique for roster sizes, so I can only benchmark against previous years for our own league. If I’m going to come up with targets to keep in mind during our draft, would you recommend taking an average of the winning number of the past x number of seasons for each category? I was originally thinking that until I saw the huge spike that was the winning HR total last year, and it made me think that I should be keeping it either compared to last season or the last two seasons to account for the way the game is trending more recently.

    • @Milarky: Good question. I personally don’t use category targets from previous years because projections assume a ‘hitting/pitching environment’ which should regress from any 1-year extremes (like last year’s HR barrage). I just focus on my category $ and aim to get over the 50th percentile in all categories ($260 * 67% – Categories) * 1/Categories for hitting, Same * 33% for pitching) – while doing my best to maximize value. I do that to make sure my team isn’t getting too unbalanced. If your league has categories I don’t project, I guess best case is find about the average from last year in your league.

      • Milarky says:
        (link)

        @Rudy Gamble: Sorry to be a simpleton on this, but can you give me an example? I generally don’t pay a ton of attention to the dollar values other than to give me an idea of the relative value Grey or you place on each player, so I’m not clear on where $260 comes from. I think I picked up from the writings on this site that the 67/33 split you’ve referenced depends on the preference of the drafter.

          • Milarky says:
            (link)

            @Rudy Gamble: Warning, it’s a little whacky. Roto 6×5 (Avg/Slg%/HR/R/RBI/SB x W/K/ERA/WHIP/S), pitching cats worth 90% of hitting stats. We keep 4 hitters, 2 pitchers, and 3 prospects. Rosters are 2 at each infield position (including DH for any hitters), 5 OF, 5 SP, 3 RP, 1 Util Pitcher, 1 rookie hitter, 1 rookie pitcher, 9 reserves.

            • That is a little wacky. I would probably use last year’s totals.

              • Milarky says:
                (link)

                @Rudy Gamble: LOL, thanks Rudy. Not even sure it’s worth it for comparison’s sake, but it may help a little. There’s just no ‘standard’ i can use.

  4. Duda Want to Build a Snowman? says:
    (link)

    Alternate Title: Why I Will Own Rich Hill Everywhere.

  5. Simply Fred

    simply fred says:
    (link)

    Rudy, I probably missed it, but did you do a comparison of how your auction values stood up against end of season razz rater last year, compared to Grey, or others? If so, where to find results?

    • Nope. Haven’t done those in a year or two. Got to be too much work.

      • Simply Fred

        simply fred says:
        (link)

        @Rudy Gamble: So, you have seen a lot. Do you put your faith in auction values or Grey’s?

        • Mine but I feel a little better about a player if Grey is also hot on him.

          • Simply Fred

            simply fred says:
            (link)

            @Rudy Gamble: Well stated. I like mine but feel better about a player when both of you are hot on him!

  6. Billy Beeeen says:
    (link)

    Rudy,
    in your auction values engine, if I want to use 7×7 QS+H, but I want to look at theranking without considering the ERA category, what can I do?

    • Easy. Download it. Subtract the ERA $ from the pitcher $ and then multiply by 7/6 (to weight everyone up to the right total for pitching).

      • Billy Beeeen says:
        (link)

        thanks Rudy. best to you @Rudy Gamble:

  7. RicoSuave says:
    (link)

    Great stuff Rudy! Thanks man…

    Do you have rankings for points leagues?

    thanks again

    • Thx. For points leagues, just download the steamer hit/pitch projections and then total the points in excel and sort.

      • RicoSuave says:
        (link)

        @Rudy Gamble: sweet, thanks!

  8. Devon says:
    (link)

    Hey Rudy,
    I understand that the steamer/razzball projections are yours. But do you have a list of positional rankings? I’m curious to where you rank certain players. For example, McCullers is #99 in steamer/razzball. Obviously that’s not his ranking. Where do you have him in your rankings and what do you expect of him this year if fully healthy? Top 20? Top 30? Ahead of Paxton and Duffy? Thanks in advance! Always appreciate what you do

    • I was just talking about this with Grey – that I have rankings based on Steamer + my playing time, a few stats, and valuation formula but they aren’t pure rankings in that I don’t always draft the first guy in the queue.

      For every draft, I grab the most relevant ADP (usually NFBC) and use that as a guide to wait as long as possible on players.

      I like McCullers but doubt I take him in any drafts because I just don’t believe he pitches more than 150 IP and wary he regresses if/when he throws a sustainable % of breaking balls.

Comments are closed.