Have you ever been in the middle of a draft and said to yourself, “How could this player still be on the board? I would TOTALLY draft him if I hadn’t already drafted a guy with a similar skill set!”?

If so, that is crazy I was able to nail every word of your thought. But even if the above is just a paraphrasing of your in-draft frustrations, this post may be helpful.

It is common practice to compare one’s fantasy baseball rankings with the best ADP proxy for your draft to ensure that you aren’t reaching for your draft picks. Through this type of analysis, you quickly get a feel for positions that you are valuing more or less than the market so you can determine whom your values rate as the best bargains at each position.

While prepping for my first draft 0f 2016 (see my LABR draft recap), I leveraged ADP in a more powerful way for hitters. I did an analysis to identify how NFBC ADP values correlate with each of the category dollar values in my Preseason Player Raters. This allowed me to see how the market is weighting each category and adjust my rankings accordingly so I wasn’t overpaying for one category (say AVG) vs another (like HRs).

To do this, I ran a regression test using 136 hitters where my playing time estimates seemed about in line with consensus. The reason category dollars work so well for this analysis is because it puts each of the five categories on the same scale and, thus, you can quickly identify that any category with a weight above 1 is valued higher and under 1 valued lower. The same principle is in place if you have category SGP, Z-scores, the FanGraphs calculator, etc. You might be able to do the same with just projected stats but the weights will look crazier since the scales are so different.

When I tested with all five hitting categories, my $RBI came in with a weight of -0.004 which means that it is not a relevant variable for predicting the market’s valuation of a player. Recalculating with the remaining four categories resulted in the following:

Intercept 1.43
$R 1.01
$HR 1.83
$SB 1.07
$AVG 0.96

The equation for converting my category dollar values to these weights is: 1.43 + $R * 1.01 + $HR * 1.83 + $SB * 1.07 + $AVG * 0.96. The ‘default’ equation is: 1 + $R + $HR + $RBI + $SB + $AVG.

While $HR and $RBI are certainly correlated (meaning players with higher $HR tend of have higher $RBI and vice versa), I was still surprised to see that $RBI was useless in explaining a player’s NFBC ADP. It could be that $R are still effective because they reflect expected playing time and batting order spot without as high a correlation to $HR.

Here are the five players that were most helped/hurt by the adjustments. Note that the value of $1 changes throughout the draft. Miggy was only knocked down 2 spots, Bogaerts went down 26 spots, while Granderson went up 50 spots.

Top 5 Value Increases Top 5 Value Decreases
Byung-ho Park (+$3.6) Miguel Cabrera (-$3.1)
Curtis Granderson (+$3.1) Prince Fielder (-$2.7)
Pedro Alvarez (+$3.7) Xander Bogaerts (-$3.3)
Colby Rasmus (+$2.9) Eric Hosmer (-$2.7)
Derek Norris (+$2.9) Robinson Cano (-$2.7)

The common denominator here is that the five players with the highest increase are power guys (average $HR of $3.7) with low average (average $AVG of -$2.2). The players who decreased the most are solid AVG/RBI guys (average $7 per category) with so-so power ($HR of $3.7).

Below is a side-by-side comparison I ran against the FanGraphs Auction Calculator using the same roster parameters (15 team mixed, NFBC roster (2 catchers), 64/36 split) with Steamer Projections and including all hitters that the FanGraphs calculator viewed as $1 or more. I am just using their category $ values which include no positional weights as they bundle positional weighting under a variable called ‘aPos’ that ranges, in this run, from $6.7 to $9.1 for non-Catcher positions and $21.9 for catchers. (This is why their ‘intercept’ is so much higher than mine as i’m just adding $1 to the sum of the category values since that’s the minimum bid).

Variable FG Auction Calc Razzball
Intercept 11.67 1.82
$R 0.30 0.62
$HR 2.27 1.99
$SB 1.20 1.07
$AVG 1.19 0.98

I see the same result. Power is going at a premium in NFBC drafts. This probably holds true for just about every mixed draft (at least based on my category values). It is not that HRs are more valuable or SBs are less valuable. It’s just cheaper to acquire the other categories.

My kids love bananas. Bananas are stupid cheap at the supermarket. We’d pay three times the market rate for them. But why would I? That’d be you know what. I can spend that surplus on milk. Blech. Those dairy farmers owe big cereal some commissions.

Mock drafting with my adjusted dollar values convinced me that this weighting makes for more effective drafting. Instead of chasing power late in the draft (when it becomes scarce) while wondering why my team skewed too heavy on average, I front-loaded power (Bryant/Upton/Fielder) and targeted a few hitters like Duda and Ozuna to maintain. Everything equal – I chose the power guy. I liked the value on a few power guys in the last half of my LABR draft (Schoop and Rasmus come to mind) but it is not a good idea to be in a position where you are dependent on a couple players to go at the right price for your draft strategy to work.

So why is this so important? Because you typically only have 60-90 seconds to decide on the right player. Any differences in category weight from ADP affect the top of your queue. In my case, it would be high AVG / low HR guys. Proactively adjusting for this before the draft allows you to avoid wasting time shuffling through your queue to find players to counteract the market’s different weighting of the various categories.

One added bonus to all this is that if your category values are fakakta, this could help fix them (or at least minimize the damage).

It’ll be interesting to see if the same holds true in ESPN and Yahoo drafts. I’ll write another post in early March once their ADPs stabilize. I imagine it will hold true.

 

 

  1. TheTinDoor says:
    (link)

    THIS is the reason I keep coming back to Razzball… you’re thinking through how to beat the market on so many different levels.

    Can you clarify something? The market is overvaluing HR, right? They’re not worth 1.8x, but that’s what people are paying for them.

    Your response is to take your dollar values (which are a more fair valuation than the market)… and adjust those values to the (overpriced) market numbers?

    On the one hand, I get that you’re trying to describe what it will take to purchase certain stats. On the other hand, the valuation gap caused by overpaying for HR could also be described as a market inefficiency to exploit, no?

    • Thanks! Maybe if HR was an isolated stat like SB or SV, it’d be possible to exploit. I see it as just a minor tweak to my values that better ensures my hitters don’t become too imbalanced. My avg still was my category and rbis were fine :)

      • TheTinDoor says:
        (link)

        @Rudy Gamble: But aren’t you tweaking more accurate valuations to line up with less accurate ones? Why not let the market overprice certain players? What am I missing?

        Maybe it’s a difference without any real impact… everyone knows power goes at a premium. So instead of looking for bargains, be prepared to spend the full price to get a big bat, etc.

        • We aren’t talking about a massive overpricing based on my values. While the difference is small on the rankings, I’ve found it’s enough to enable me to build a balanced squad without overcompensating throughout the draft. Otherwise, I’ve got a team that’s likely bonkers in AVG (vs just very good), strong in R/RBI, and has a hole in HR. Perhaps I’m just more willing than others to sacrifice a couple dollars in my projected values to achieve some semblance of category balance so I can make moves from a position of strength vs necessity.

          • TheTinDoor says:
            (link)

            @Rudy Gamble: So it’s basically a difference between making mental adjustments throughout the draft (“I need to be willing to pay a little extra for power guys”) and baking into the valuation. I think I prefer the former, as it helps keep me aware of the “true” value… I don’t see any real upside in knowingly making the values less accurate.

            Either way, great thought exercise, haven’t seen this discussed EVER on any other fantasy site.

            • Whatever works for you. Just saying that I ran a number of mock drafts and this method proved better.

  2. Mike says:
    (link)

    Just drafted a CBS 10 team league HtoH weekly league. I like the league because they let you know in advance who is pitching each day for the week you are in.

    I could not believe who I drafted. Rizzo, Machado, Correa, Betts, Marte, JD Martinez, Franco, Odor and Hundley as my catcher. With Polanco on my bench.

    I didn’t draft a pitcher until the 5th round, My pitchers were Carrasco, Igleslas, Matz, Carlos Rodon, T. Rosss, Nola and Walker. I waited on relieve pitchers and drafted Capps and New Yorks
    Miller. Didn’t realize I had two more picks, and ended up with 3 catchers. Trying to get Schoop and another pitcher. I wish my yahoo drafts will be as good.

    • What? Who were these other drafters because I have a bridge to sell them?

      • Mike says:
        (link)

        @Rudy Gamble: 6 guys showed up out of 10. Most of them went with CBS rankings. The most one sided draft I have ever been in. I didn’t draft JD Martinez until the 6th round. I had JD last year on all but one of my 10 teams. This year I never thought I would get him because I am in 14 team leagues in yahoo. I want a 1st baseman my first pick and a Outfielder in the second round.

        • That is why we have our commenter leagues!

  3. Scooter G says:
    (link)

    I’ve been drafting bogaertes early (about pick 60)/in a lot of my mock drafts because alot of SS seem to lose value in the 5×5 obp format I play in. Any SS that gain value in obp off the top of your head?

    Thanks

    • @Scooter G: Here is the net $ difference for 12-team ESPN leagues between 5×5 and 5×5 OBP (so positive means gains value). First number is 5×5 OBP $. Second is difference. :
      Name 5×5 OBP $ Diff vs 5×5
      Carlos Correa | 28.7 | -2.3
      Xander Bogaerts | 15 | -4.8
      Francisco Lindor | 13 | -3.4
      Troy Tulowitzki | 10.9 | 1.6
      Jose Reyes | 8.2 | -4.6
      Elvis Andrus | 6.9 | -1.2
      Corey Seager | 4.3 | -2.1
      Marcus Semien | 3.2 | 1.5
      Ian Desmond | 2.9 | 1
      Jhonny Peralta | 2.1 | 0.8
      Ketel Marte | 1.9 | -3.1
      Addison Russell | 1.6 | 0.8
      Brad Miller | 1.5 | 1.9
      Starlin Castro | 0 | -3.2
      Jean Segura | -0.4 | -4.5
      Alcides Escobar | -1 | -5
      Didi Gregorius | -2.9 | 0.3
      Jose Iglesias | -2.9 | -2.2
      Brandon Crawford | -3.6 | 3
      Erick Aybar | -4.7 | -3.5
      Asdrubal Cabrera | -4.9 | 2.1
      Andrelton Simmons | -4.9 | -1.2

  4. Zeus says:
    (link)

    Great article Rudy. Been thinking the same way.

  5. SefSef says:
    (link)

    Rudy, I don’t usually comment but I couldn’t help myself.

    That is one sexy brain you’ve got there, and you truly are psychic for writing this piece. I was contemplating this idea all day and I tune in to my favorite fantasy site and what do I find? It’s the gift that keeps on giving.

    Now for the brain buster. I’ve created my own dollar value for my second year in the big leagues(NFBC) via a projection of what the main event overall winner needs to score in each category. My numbers taken from the 75th percentile in 2014 and the 80th percentile(4000 points) in 2015 that should win the main event overall standings are:
    1018 runs = 72 per hitter
    259 hr =18.5
    984 = 70
    158 steals =11 per hitter
    .270 avg

    I plan on taking your beautiful weights and adding them to my projections but I’d like to see another weight that I am unable to figure out how to implement since I have the applied mathmatical skills of a monkey with a calculator.

    How would you weight stats on the max amount of replacement stats that can be obtained from a single player in each category?

    Steals are argued about a lot. Should Dee Gordon be going in the top 10? Hamilton top 20?
    There were nearly twice as many total homers last year as steals in the mlb. We need 18.5 dingers and 11 steals per hitter, so steals are harder to come by and thus more valuable. 55 steals > 40 homers right, if Hamilton gets you 5 batters worth of steals and Chris Davis only gets you 2.6 batters worth of homers?

    Not so fast, there guys going at pick 300 in the NFBC that can get you 30+ steals (Dyson) This point was highlighted by your article, homers are the focus of the NFBC draft market. No one out of the top 100 is projected over 30 dingers. We all know this right, where am I going you wonder?

    Well, if your guy gets you 55 steals like Hamilton(per your projections), he’s getting 5.5 batters worth of steals, but he’s only gonna net you 40 rbi’s. Those 30 rbi’s he lost you under the 70 per-player average needed are a lot more difficult to replace since the most rbi’s you can gain from one hitter(projected) is:
    106(Trout)-70 runs per batter = 36 ~.5 batters gained at most

    In summation, the most difficult stats to fix deficiencies for in order are: RBI, Runs, Avg, HR, then SB.
    You can get players that get you 6x what you need in steals, but only 1.5x what you need in RBI’s.

    How would you weight your dollar values for this, the fact that some stats are easily replaced and others aren’t?

    Thanks for all the insight, keep ’em coming.
    -Sef

    • Thanks Sef. I know some people like to attack player valuation by having to reach certain stat targets. I much prefer my $. The only things that need to be added/considered IMO are 1) determining if you want to re-weight categories (this post) and 2) comparing player $ vs ADP to determine best bargains (no need to jump for a guy if he’s going to be around in a couple rounds).

      One thing I dislike about the ‘i need these stats’ is that projections have some regression to them. The $ base the values on the projection universe (and adjust the point shares proportionately for determining the value that each stat has on the standings).

      So I’d consider any weights of HR vs SB already accounted for in my $.

      Make sense?

      • SefSef says:
        (link)

        @Rudy Gamble:

        Thanks for the response Rudy,

        *Goes to your projections, rereads how point shares work, snaps out of daze hours later*

        I see that the points are all relative to each other in that they are on the same scale(12 teams points), but I’m not certain it solves my points(Ha!).

        I don’t think I conveyed my thoughts well, perhaps I’m typing it faster than I can verbalize. What I was going for is a metric that weights the swings in range from one stat to next as compared to the median, a sort of “Replacement $ difficulty metric”… In my head it’s starting to seem more important to not lose more stats from one player than can be replaced by one.

        I guess this could be outlined by sorting your stats by $ and looking at the min and max $ value for each stat.
        -Max- -Min-
        Runs $9.2 $ -9
        HR $13.7 $ -5.5
        RBI $10.7 $ -10.2
        SB $21.9 $ -1.6
        AVG $12.8 $ -5.8

        So you can gain at most $9 in Runs and lose at worst $-9, but in SB you can gain $21.9 and lose at the worst $-1.6. By looking at these numbers, we can see (we already know) that it is very easy to replace stolen bases lost from any player deficient in them, merely one player can replace the dollar value lost of the worst 13 players drafted . What is much more difficult is replacing Runs and RBI, it takes the best player to replace the 1 worst player in those $ values. Worse yet is our drafts are weighted in Runs, RBI and HR as we saw in this post, so you won’t be finding those high $ values out of the top 100 players.

        This means (to me) that RBI and Runs are most important to have non-negative $ values, followed by HR, AVG, and then SB . It’s not so much the points gained, it’s the assumption that you want all 12 points in a category, and if you draft 5 guys with zero SB it doesn’t hurt that bad, you could still take all 12 with some elite speed. But if you get 5 guys with $-9 RBI, you will need 5 of the best players in the league to replace the RBI lost, forget competing for the top spot.

        The first few players you draft should have as many RBI and Runs as possible, they are the most difficult $ to replace if your team is deficient.
        I can see that if you keep a running $ total during the draft this will be solved, what is more difficult is weighting the pre-draft rankings based on this. How to explain this with algorithms I have no clue.

        I’m not certain that the point shares take this into account, but correct me if I’m wrong. I hope this makes some sense. Oh and after reading the calculations done to get the $ value, I’ve been converted to $ value as well!

        Let me know if I should crack open my head and feast on the goo inside.

        -Sef

        • You probably have a point where SBs are the easiest to ‘make up’ with a replacement player and R/RBI are the toughest since those stats are tied to better players. But those SB guys are so bad at R/HR/RBI that it might be counterproductive.

          But R/RBI can be found for cheap – usually #2 hitters with little power. Like Melky Cabrera. Gerardo Parra falls into this category this year too. So don’t really buy into what you’re selling or see how I’d adjust rankings to account for it.

Comments are closed.