Don't be shellfish...Share on FacebookTweet about this on TwitterShare on Google+

Looking for a great fantasy baseball draft kit to help you draft your fantasy baseball team in 2013? Good, because otherwise you are the worst web surfer in the world.  (Did I search for fantasy baseball rankings?  Damn, I meant chicken cordon bleu recipe.)

The Razzball 2013 Fantasy Baseball Draft Kit is free for the first 999,999 visitors.  The millionth visitor will receive an oversized check for $100,000 with their name in big letters (it won’t have our names or signatures on it though).  Everyone after that will also get a free fantasy baseball draft kit.

Rather than make you download a PDF, we are just going to give you a bunch of links.  Where don’t you have wi-fi access these days?  Plus, by keeping these as stand-alone pages, we can constantly update our fantasy baseball ratings throughout the preseason.   Until then, have fun in your mock fantasy baseball drafts and keep posted for more updates.

Fantasy Baseball Projections

  • 2013 Hitter Projections (Steamer) - Will be updated at least two more times before the season starts.  The projections are by Steamer.  The playing time estimates (plate appearances) are by Razzball.  Includes every player we project with 1+ AB.
  • 2013 Pitcher Projections (Steamer) - Will be updated at least two more times before the season starts.  The pitcher playing time stats as well as Saves, Holds, and Quality Starts are courtesy of Razzball.  Includes every pitcher we project with 1+ IP.
  • 2013 Fantasy Baseball Projections (Grey) - May be updated before the season starts (particularly if there are notable injuries).  Includes Grey’s top 400 players.
  • What were the best fantasy baseball projections in 2012?  - A thorough review of 2012 Fantasy Baseball Projections focused on how much they helped predict fantasy baseball success.  Includes Baseball Prospectus (PECOTA), ZiPS, CAIRO, FantasyPros.com, Rotochamp, Steamer, Oliver, Marcel, FanGraphs Fans Projections, ESPN, Fantistics and our very own Grey Albright.

Fantasy Baseball Rankings

  • 2013 Fantasy Baseball Rankings - Written by our own Grey Albright, these fantasy baseball ratings bring together the exhausting word counts of early 20th century Russian novels with the humor of late 20th century Russian comedian Yakov Smirnov.  Includes rankings by position and an overall top 400.
  • What were the best fantasy baseball rankings in 2012? - A thorough review of 2012 Fantasy Baseball Rankings.  Includes Razzball’s Grey Albright, various ESPN Fantasy Baseball rankings (Matthew Berry, Tristan Cockcroft, the ESPN Top 300), SI.com’s Eric Mack, KFFL, CBS Fantasy Baseball rankings and more.

Fantasy Baseball Auction Values (Standard 5×5)

These auction values will be updated throughout the preseason in concert with MLB depth charts.  The 5×5 projections are based on a combination of Steamer and CAIRO rates (e.g., a HR per 20 plate appearances).  The fantasy baseball dollar values, Point Shares, and playing time estimates are Razzball (e.g., 600 plate appearances * HR/20 PA = 30 HR).

Fantasy Baseball Auction Values (6×6, 7×7, All ESPN 12 Team format)

These auction values will be updated throughout the preseason in concert with MLB depth charts.  The OBP and OPS projections are courtesy of Steamer and CAIRO.  The Holds and Quality Starts are by Razzball.

Fantasy Baseball Auction Values (AL/NL-only – 2 catcher roster format)

Bonus Features

  • Fantasy Baseball War Room – *COMING SOON*
  • Fantasy Baseball Names Generator - Looking to create a funny fantasy baseball team name but don’t feel like putting in any effort.  Look no further.  Just use our fantasy baseball team name generator instead!

 

74 Responses

  1. Randall says:
    (link)

    Grrrr, why nothing for H2H points league?

    • I researched it a bit and I don’t see how or why to change $ values or rankings to reflect H2H.

      The only distinctions I see is are qualitative. I’d focus on SP depth (to max 2-start pitchers and avoid depending on the waiver wire for that since there’s so much competition) and invest less in SBs and SVs (since they are flukier).

      But I don’t know a way to quantify those. I looked into a ‘streakiness’ factor for hitters to see if there’s a skill in being great/bad (think Hamilton 2012) vs. good/good – with the latter being more valuable in H2H. With 3 years’ worth of data, it came out that there’s about 0% predictability in month-to-month variations in hitter values.

    • I stand corrected. I can quantify one aspect of H2H vs. Roto. I think the average H2H team has a lot more GS than in a Roto league.

      If someone can give me some ‘average team IP’ numbers for a 12-team H2H league (specifying if it’s an 8 or 9 pitcher league), I can quantify the impact. (In simple terms, more IP = less pitcher value as their % of team Wins and Ks goes down and their impact on ERA/WHIP is decreased).

  2. sean says:
    (link)

    much appreciated, gentlemen. as always, look forward to another great season!

    • Thanks and good luck!

  3. shitbird says:
    (link)

    All these numbers and equations make me giddy.
    My favouite equation though is Rudy = Da Man.
    Thanks as always for providing us with some next level geek ish Rudy. It is truly appreciated.

    • thanks shitbird! (that’s really fun to type)

  4. ReleasetheMcKraken says:
    (link)

    Curious if other leagues besides ESPN and Yahoo are considered…CBS for example

    • I think CBSSports (and most other sites) tends to use the same defaults as ESPN (20 games, 5 OF/1 UTIL).

      Yahoo and ESPN (in that order) are the two most common sites for Razzballers so it’s easier to note them in that way. CBSSports is a close 3rd I believe.

  5. Giacomo says:
    (link)

    Any chance of a 5 x 5 yahoo with OBP? I’m probably just being greedy. Anyways thanks as alwyas.

    • Never hurts to ask, right? I’ve only coded it so far for ESPN. Should have this format ready for during the season for in-season player rater.

      Assuming you’re not completely Excel-phobic, here’s the easiest way to get this info for now:
      1) Cut/paste the 5×5 OBP for ESPN
      2) Cut/paste the 5×5 Yahoo league that’s closest to your leagues’ # of teams (10,12,15).
      3) There’s a difference column in the 5×5 OBP grid that shows the change in a player’s worth when you switch from AVG to OBP.
      4) Add that number to the player’s $ value in the 5×5 Yahoo league.

      Linking up the players is easy with a VLOOKUP function in Excel or Google Docs…if you know how to do it.

  6. Shags says:
    (link)

    Hi Rudy, I noticed you went from ranking 14 and 16 team leagues to doing just 15. Are there plans to run $ values for 16 team leagues? If not, whats the best way to (easily) take the 15 team leagues and apply them to 16?

    Thanks.

    • I’ll have 14/16 team for in-season. Just had to limit what I can do for pre-season since we’re in the midst of some transitioning.

      There is still 14/15/16 team data in the ‘Player Rater’ from end of year 2012 which might help to look at to see the differences (i think it’s pretty minor).

  7. Goose says:
    (link)

    Awesomeness. Thanks, Rudy!

      • Cwalker says:
        (link)

        @Rudy Gamble: @Rudy Gamble: our league has 13 teams, 1600 minimum innings pitched. Categories: Pitching – era, whip, saves, holds, net wins, strikeouts.
        Hitting – hr, RBI, rs, avg, net stolen bases, bbko%.
        Keeper league, auction $260.
        Wonder how I can get accurate dollar values?

        • Try Rotochamp. I think they have all those stats except bbko% (which I don’t even know what it is).

          The Steamer/Razzball stats are in there but those aren’t my Holds in there (project a little high imho).

  8. Rags says:
    (link)

    IT’S LIKE CHRISTMAS!

  9. Rags says:
    (link)

    When you say Last Player Picked was second highest in the rankings rankings, is that the Composite option? Don’t you just pick Steamer, Zips, Cairo, Marcel or Composite with LPP?

    • It was LPP’s composite which only had Steamer and CAIRO last year. I used Steamer and ZiPS and LPP outranked me b/c CAIRO outperformed ZiPS. (i switched to Steamer + CAIRO for that reason this year – as well as the fact that CAIRO’s v1 is done way sooner than ZiPS)

  10. Simply Fred

    simply fred says:
    (link)

    Rudy, thought you posted previously that Steamer was best at projecting AB. If that is correct, why would “The playing time estimates (plate appearances) are by Razzball.”?

    • Yup, Steamer did a good job with Playing Time last year but decided not to do them this year so we’re picking up the slack.

      • sean says:
        (link)

        @Rudy Gamble: Rudy, has any site compared playing time projections to find a winner for the last few years? Tango?

        • I did in October. There’s a link in the draft kit. It was part of the projections test.

  11. Tyler says:
    (link)

    This might have been covered elsewhere, but re: commenter leagues, it looks like Yahoo! is allowing leagues to be publicly viewable this year.

    • That’s good. Definitely needed for RCL. But our stats gathering system is already built for ESPN so it’s a high switching cost for us right now…

      • Tyler says:
        (link)

        @Rudy Gamble: Gotcha, makes perfect sense. Like Dave Cameron says about the Felix contract, “Frictional Costs” (though I don’t think transaction costs = $25 million…)

  12. Jesse Downing says:
    (link)

    Hi, I’ve been reading Razzball for 4 years now andI love the site! Yall rock!

    this question is for Rudy:

    First Rudy, as a finance and econ major, I have great respect and appreciation for all the quality stats you do on Razz. Thank you!

    I want to try something new for my draft rankings this year:

    1. Obtain player stat projections for 2013, probably Steamer

    2. Look back at last season to determine all players I expect to be drafted in 2013.

    3. Group players by position and calculate positional averages.

    4. Rank players based on the spread between a player’s projection and his positional average.

    Do you think this is a solid methodology?

    Finally, in regards to step 3, would it be wise to adjust a given positional average based on the standard deviation within the position? For example, to capture the fact that the variance within Shortstops may be much greater than that within OF. It would look like:

    Beta * Position Average = Expected Value of Position Average

    where Beta the expected risk, based on variance, within every position.

    Am I on the right track?

    Much love and respect buddy!

    Jesse

    • Hey Jesse -
      Thanks for the love. Appreciated.

      I would use the projections to determine your positional advantages instead of 2012 data as projections tend to be more conservative than actual stats. That’s what I do as part of the Point Shares.

      Most $ systems leverage standard deviations (I don’t). I know at least some leverage the standard deviation to create a z-score for each category and then sum up the z-scores.

      Good luck and let me know how the results end up comparing to my results…

      Rudy

      • Czernobog says:
        (link)

        @Rudy Gamble:

        I’m actually a fan of taking the z-score and turning it into a percentile using a normal distribution. Which is somewhat borked as there’s no reason to expect a normal distribution for a stat like steals. But it reflects my preference for balanced players over one stat players, my dislike of paying a premium for a single stat from a single player, makes someone projected for 30 SB get a near identical $ value as someone projected for the same stats but 50 SB.

        • Jesse Downing says:
          (link)

          @Czernobog:

          This is interesting. So is what you do something like this:

          For a given player, calculate 5 z-scores: HR, SB, R, RBI, AVG (within his position)
          and sum the z-scores. This number is used to determine the player rankings.

          Or, are you saying you convert the z-score into a percentile?

          • Jesse Downing says:
            (link)

            @Jesse Downing:

            Or, as an alternate, for a given position, you could use the z-score to compute the probability that a competing player will produce lower than the player you are targeting. The larger the probability, the better the player is expected to perform in a given category. Sum these probabilities for each category and the player with the higher number is ranked higher.

            • Jesse Downing says:
              (link)

              @Jesse Downing:
              And like you said, you assume a given category for a given position is normally distributed.

          • Czernobog says:
            (link)

            @Jesse Downing:

            “This is interesting. So is what you do something like this:

            For a given player, calculate 5 z-scores: HR, SB, R, RBI, AVG (within his position)
            and sum the z-scores. This number is used to determine the player rankings.

            Or, are you saying you convert the z-score into a percentile?”

            Basically, yeah. And I convert the z-score to a percentile. So someone with a z-score of +1 in all 5 categories would score 4.21, but a SAGNOF with +4 in SB, +2 in R, -1 in HR, 0 in AVG & RBI only scores 3.14 despite the same total z-score, and someone right on the average in all 5 scores 2.5

            What I did last year was actually to take the 2011 stats, and fiddle with the weights a bit until it corresponded better to my own preferences. So the overall value was one part percentiles, one part raw z-score within a position, and one part raw z-score across all batters. RP & SP I treated entirely separately. Then second part was to take the values, and convert them to $totals. Easy then to get new values for deciding to punt a stat, for splitting the budget differently, for deciding to get whichever $1 special you can in a particular position like catcher. Unfortunately, don’t have it anymore, so I’m going to be redoing it for this year.

            And since player value rating are subjective, unlike the stats themselves, eyeballing those weights isn’t a problem. Does mean that 2 groups of players could produce identical stats but have different total value, but I don’t see that as a problem, and I suspect it would be the case in most value systems that aren’t based on pure points anyway.

          • It’s not my methodology but, yeah, that’s my understanding of a z-score methodology. I think it’s just add up the z-scores.

  13. Czernobog says:
    (link)

    Looks excellent, as always.

    The only change I think would be really good would be that instead of multiple, pre-determined setting sets, would be add the option for a custom set. Haven’t played with your settings to try and reverse-engineer the value formula, and I don’t know if your average/$0 value/starting point/whatever is hard coded, or based on an average of the top x players’ actual projections.

    But if the projections cover a stat, then it’s easy/simplified enough to decide that the league average player in a league that requires 100 OF will produce the average of the top 100 OF in that stat, and value that stat accordingly. The only stuff to hardcode is AB or PA per batting roster spot, IP per pitching roster spot. And they could be left as custom entries, too, can test for yourself the difference between a weekly change, 180 IP per SP slot and a daily change, cap free, streamer’s paradise where you can get 400 IP out of each SP slot.

    Yes, this does mostly boil down to a request to save me from trying to c&p bucketloads of projections into my own spreadsheet. :D

    • I’ve been working this offseason to build code that’s more universal for all league types. It’s getting there. It is harder with my model vs. z-scores because I’m actually trying to measure each player in a category based on their impact in standing points and then i create a standing point value for hitters and pitchers separately. Z-scores just provide how well a player does vs. a standard deviation which does not directly tie with standings points. Perhaps next year…

      • Czernobog says:
        (link)

        @Rudy Gamble:

        Yeah, makes sense.

        To me, trying to accurately aim for standings points seems like more work for not much benefit. Though I could easily be wrong. Looking at how reliable it is that a certain stat total will give you a certain place in the standings isn’t something I’ve done, I tend to just be aiming for a rough benchmark in each category, and assume being 10% or 20% above the league average for team totals will be roughly equal value regardless of category.

        I don’t think z-score vs raw total should make much (if any) difference. The differences should only come if there are differences in the population of players used for each one, or because your standings numbers expect that the average total will be better than 12 average players produce, thanks to switching players due to form/injury/etc. But if the EV for your total HR across 12 teams is the same as the EV for 12 teams with a cumulative 0 z-score, then the $ values for just that stat should be identical, I think.

        • I think z-scores – if done right – ends up with results similar to what I end up with Point Shares. There are other decisions that end up going into $ estimation that can also factor into the accuracy. The only point I can say is that when I tested my end of season $ values against ESPN’s Player Rater (z-score) based, mine were more predictive of RCL team results. But I’d recommend people try z-scores before Point Shares b/c it’s a better bang for the buck (if your time is money).

          • Czernobog says:
            (link)

            @Rudy Gamble:

            That would be interesting actually, if it’s doable. Take a couple of less conventional but still systematic valuing systems, and plug them in against the RCL results, see how they compare to point shares and ESPN/Yahoo plaer raters.

              • Czernobog says:
                (link)

                @Rudy Gamble:

                I’ve just found my 2010 version, when I started trying to work out some player ratings for an 8×8, H2H league. It’s got point totals based on all 3 of z-scores, percentiles, and a customisable combination of the two. Going to fiddle with that, and put in 2012 final stats, see what I come up with for 2012 values.

                Is there any easily accessible source of all the initial RCL teams, and how they finished? I think that would be a good resource for razzballers to test value systems, even though turning projections into $/rankings is less important than getting decent projections in the first place.

                • It’s more than just how they finished. Also involves every draft pick. I made 2011 data public in March 2012 but there weren’t really takers.

                  Will consider it w/ 2013 data.

                  • Czernobog says:
                    (link)

                    @Rudy Gamble:

                    Bummer, I missed that one.

                    Yeah, I assume it would just be a case of taking the rosters from draft results, taking post-season $values (and capping the lower end for injured/terrible guys that wouldn’t actually pollute the roster all year) and then correlating post-draft team $value with total points scored.

                    Is there still al ink to the 2011 data? I can’t see it in the post you linked earlier, it’s only got the average RCL finish for teams that drafted each player.

  14. Iron Mike says:
    (link)

    I hope there will be APPLES again this year!

    • Iron Mike says:
      (link)

      @Iron Mike: And thanks for all the hard work. Love the site. Second year visitor here. Keep up the great work team.

  15. NEW EMAIL ADDRESS
    Mr. Rudy: I am a bit intimidated writing to you here and like this… I am approaching you with much trepidation..

    Wait!!! Let me backup here for a sec… And give you a smidgen of backdrop here to assist you in understanding my statement above..

    I have been dealing with your Mr Gray for most of the spring with my FBB questions on your website ( That should tell you a lot)…

    But first, a bit about myself that might shed some light on all this too… I am a shy, sober and serious fellow … Many might describe me as almost taciturn in my postings – ((not liking to let my feelings show or wax on without need, I keep my postings on point and succinct))…. I would never see myself as an apathetic or a sluggish phlegmatic though – at least, I would like to see it that way…… But, I am certainly reserved by any measure..

    Sorry, I know, you are probably very busy now…. Moving on… How do I put this???..

    OK, Get it out!!! Wobbly!!! That’s it!!… I often get this wobbly sensation when dealing with the “mouthpiece” of your website, Mr Gray…

    Well, as they say, “In for penny – In for a pound”… Here is the question… Is he out of balance??? Don’t be taken aback here, I’m just a little concerned… You will see why – if you would just keep reading…

    He has requested, I join one of your “Commenter” leagues.. Well, needless to say, my personal alarm bells and warning klaxons have sounded upon reading this.. Below, are found a few of my feelings and thoughts on this… Along with the reason, I have come to you now…

    First: I am more than a little reticent to hand over any of my personal information to this man, given the questions I have regarding his – stability, mental makeup, moral character, and his bizarre way of imparting information – (His tendency towards vituperation, when discussing innocent BB players – who, after all, are just doing their jobs and trying to make a living) and (his strange sense of what a metaphor should and can be – not a usage issue here – more of a question of where these metaphors come from (I fear, they may be from some dark dark place that I don’t want to go to – ever and how, they are applied to what is being discussed at any given time)… Is this for league real??? Is it safe??? Is it a place for someone sensitive – like me???

    Second: It has become clear to me, that you, Rudy, must be the brains and grounded one in this outfit… You have to be the rock this website is built on… So, I felt, it was important for me; to touch base with you to introduce myself; add my concerns regarding your partner (your sorcerer’s apprentice) and perhaps, get some information regarding this league – rules, time, the like, before deciding on joining a league on this website… *****Have you ever considered, that you maybe are being retarded from achieving your full bloom in the world of fantasy BB by virtue of your connection with Mr. G??? *****.

    Third: A couple more questions…

    1 Please, tell you are really 2 different people???

    2 Can you send me proof of that???.. I have a lingering feeling, your website images may have been photo-shopped..

    3 Do you post in the nude???

    4 Have you toyed with or have you incorporated Improbability Theory into your FBB statistical analysis .. I hope so… You are missing out Big Time if you have not.. It is the new wave…

    Again sorry, for taking up your valuable time and I am certain, you get postings or Emails all the time of this nature and on this topic – too often..

    Thank you for your attention to this matter…

    Signed: Yours in FBB… A confused and somewhat leery, T Moore…

    • That’s a lot of stuff to answer. Here’s the quick version – grey and I are two different people (our pics are completely different!), grey isn’t imbalanced any more than the rest of us – it just shows more because he writes a zillion words a week between posts and comments, I’m the stat/ spreadsheet brains of the operation but grey is the brains behind everything else, we don’t collect any personal info for the commenter leagues – you can sign up directly on ESpn after clicking the link of an open draft, I do not write in the nude except on Howard Hughes birthday

      • joet says:
        (link)

        Rudy,

        Could you once again relate the lineup requirements for Yahoo and ESPN?

        • ESPN = C/1B/2B/SS/3B/5 OF/CI/MI/UTIL/9 P

          Yahoo = C/1B/2B/SS/3B/3 OF/2 UTIL/2SP/2RP/4P

      • @Rudy Gamble:

        Mr Rudy:

        1 OK… I am taking a leap of faith and am believing you that you and Mr. G are not the same person…

        2 I am assuming you are not using Inprobability Theory in your Player Performance Matrices… Too bad… You don’t know what you’re missing…

        3 I forgot to mention it…. But, whatever you do, please, please, don’t let Mr G read my first comment… I get the impression, he good get real vicious over something like that…

        Later.. And thanks for the good hotstove work… You guys are top drawer!!!!

        Thanks again…

        • Nope on improbability theory.

          Grey reads the blog and your post is on the blog. So he might read it.

  16. Steve says:
    (link)

    Amazing to see how far Razzball has come since I first came across it in ’08.

    Bravo to all concerned, but especially Grey and Rudy.

  17. Q says:
    (link)

    I love reading the comments on your posts, Rudy. It makes me feel like I’m good at math even though I have no idea what you are talking about.
    I do have a question about how best to take your brilliance and exploit it for selfish gain.
    The point shares are projections of what dollar values players will earn. But does that mean that those dollar values are what you would actually pay for them? E.G. you would pay $50 for Trout at auction? Seems like an awful lot to spend on a single player. Thanks!

    • Glad to be bringing the I to your Q. The Point Shares represent the standing points contribution of the player vs the average player. A guy with a 5.0 for SBs would raise a team from, say, 5 points in SB to 10. The $ are based on a $260 budget. I price what a guy is worth – the goal at the auction is to get the biggest bang for the buck. If you pay my rates for every player, you’d be projected as just an average team. Hopefully, you get Trout for $40 and you benefit from $10 of surplus value.

  18. Rags says:
    (link)

    So I’m in an 11-team NL-only OBP 2 catcher league. Should I just average the prices between the 10 and 12 person league for an approximation? I was thinking of avg’ing the point shares for each category (adjusting with your OBP pointshares), summing them up and converting them to dollars, but there doesn’t seem to be a consistent price for point shares.

    Also, is there an error with Daniel Murphy? Right now in NL-only you have him listed as the 4th best 2B worth $20+. Grey’s projections seem to strenuously disagree…

    • I would average the 10 team and 12 team $ and then adjust the $ based on the $ difference column in the 5×5 Obp sheet. make sense?

      As for Daniel Murphy, Grey’s projected stats aren’t that far off than those used in my calcs. He has a blind spot/bias against hitters with mediocre power and speed but are good in AVG and okay in R/HR/RBI/SB. NL 2B is weak this year and I like Murphy to be above average for his position in R/RBI/AVG. NL 2Nd basemen have no speed this year so he won’t hurt you there. The biggest if for Murphy is health – I have his PA based on 154 games while several other NL 2B i have closer to 550 PAs due to injury risk or internal competition (eg, Espinosa from Lombardozzi). So his NL $ is probably on the optimistic side and one where I’d want to get him a couple $ cheaper.

  19. donkeycorner says:
    (link)

    Thanks for the great advice Rudy. Finished second with your help last season (behind the guy who got two others owners to trade their first round picks to him include M. Cabrera….) We bought Grey a daquiri. Is there a “buy Rudy a margarita” link somewhere?

    • You are welcome, donkeycorner. No need to buy me a margarita. Just Facebook or tweet a couple of my posts/$ value charts during the year and we’re good.

  20. AJA says:
    (link)

    @Rudy Gamble:
    In H2H (though not for points leagues), streakiness helps or hurts depending on the team construct. For example, a team that doesn’t have many HR hitters and in expectation will lose the HR category each week, would benefit from a streaky guy like Hamilton to win that category when he gets hot. On the other hand, if a team is strong in the HR category then all those weeks when Hamilton explodes doesn’t help as much since the team probably would have won anyway, but the weeks when he hits 0 hrs may allow a weaker team to pull out the category.

    • I totally agree that if streakiness existed as a forecastable skill that it would be important for H2H value. I’ll do a post this pre-season on my attempt to quantify the skill.

      • AJA says:
        (link)

        Ah, I see. You are a step ahead of me. I look forward to your write up.

      • AJA says:
        (link)

        @Rudy Gamble:
        Rudy,

        Not sure if you check old posts but I had an idea for a statistical model for measuring streakiness, that would be pretty easy to estimate if you have the data and could be a pretty neat tool if it works. I don’t know if you appreciate these kinds of suggestions from the “math” guys in the audience so feel free to disregard if you aren’t interested. Anyway, here goes.

        Basically, the idea is to figure out how much weight to give a hot streak in predicting how well a player will perform over a short time horizon (say 1-game).

        I would start by using OPS as a proxy for player performance. I would specify a regression:

        OPS(t+1) = X * OPS3 + (1 – X) * OPSseason

        In the above equation:
        OPS(t+1) is tomorrow’s player OPS (i.e. the value you are trying to predict).
        OPS3 is the player’s observed OPS over the last 3 games.
        OPSseason is the player’s season to date OPS.
        X is the weight you would be estimating.

        The neat part about the model is that the weight X, would have a very intuitive interpretation as it is tells you how much weight to give a hot or cold streak relative to how well a player has performed over the long term. A weight close to 0% means that the streak isn’t predictive and season to date performance is the key. A higher weight would mean a guy is expected to outperform his season to date performance over the short term.

        I would first find X over the full population of players (conditioned on having a sufficient number of PAs, maybe restrict the model to post ASB?). Let’s say X turns out to be something like 5%. You can then run the same regression for individual players. Players with a higher X (say 10%) are more likely to be streaky. Players with a low X (say 1%) are not.

        Hope this makes sense and isn’t too much gibberish. Given the time and the data, this is how I would approach the issue of streakiness. As I said above, feel free to disregard this post if you aren’t interested.

        • @AJA: Thanks for the suggestion. I don’t think it’s gibberish – I get what you’re saying for the most part – but my larger point is that I don’t think there’s enough statistical confidence in ‘streakiness’ to use it as part of player forecasting once you separate out factors like opposing pitcher quality and ballpark. I think there are players who are more prone to hot/cold streaks but this would be, at best, a secondary factor for forecasting tomorrow’s game (I think player skill, opposing pitcher, and opposing pitcher’s handedness would be the most important factors). Given the amount of work + this being a best case factor for evaluating players, I think I’m shelving this for 2013. Maybe in the offseason I’ll review it again….

  21. Drew Chainz says:
    (link)

    I want that check

  22. Justin Cook says:
    (link)

    This is only my 2nd year with razzball, but last year I fell in love with the WAR ROOM excel sheet. Any idea when that will become available?

    • A war room release is coming soon!

  23. Cheez Whit says:
    (link)

    Maybe I am really high, but somehow glancing over the list for EPSN 12 team 5×5, I somehow got Kelly Shoppach ranked at 119 overall, ahead of Matt Wieters, Wilin Rosario among other Catchers? This has to be an anomaly no? Otherwise, great work as usuall!!!

    • Ha – yeah, there’s a slight bug every once in a while when a player’s stats don’t calculate right and they get 0.0 Point Shares which gives them a rank in that range.

      Fixing it now and should be fixed by the time you check again. there’s also a couple of playing time estimate changes in there that’ll probably be imperceptible (I had Madson missing the first month anyway for IP but forget to dock him a couple saves – now done)

Comments are closed.