THT Essentials:

#### Get It Now!

The tenth Hardball Times Annual is now available. It's got 300 pages of articles, commentary and even a crossword puzzle. You can buy the Annual at Amazon, for your Kindle or on our own page (which helps us the most financially). However you buy it, enjoy!

And here's the full roster.

## Fantasy values by parallax

Posted by John Burnson at 4:00am

parallax n. : The apparent displacement of an object caused by a change in the position from which it is viewed.

Generating dollar values for fantasy players can be tedious. A common approach is to sum the stats above replacement level in a category and then divvy up those stats among a portion of the total budget and add up the contributions for each player. That’s doable, but there are challenges. For one thing, there are wrinkles to handling rate stats like BA and ERA and “clumpy” stats like saves and steals. Also, there is something unrealistic about treating categories as freely floating when there are obvious dependencies, such as between home runs and RBIs, or ERA and wins.

There is another approach. This one has its own challenges, including a longer time to derive the values, but it sidesteps the bumps with the usual method, and it’s easily tailored to many formats.

The key is to look at fantasy value from a different angle. Suppose that Roy Halladay is valued at \$30 in your league. It’s true this says that Halladay’s stats are “worth” \$30. But you could re-state this to say that paying \$30 for Halladay neither helps nor hurts your odds of winning. If you get Halladay for less than \$30, then your odds of winning go up, and if you pay more than \$30, then they fall. But paying \$30 neither raises nor reduces your odds; if it did, then \$30 would be the wrong price.

So we have turned a statement of value (“Halladay is worth \$X”) into a statement of probability (“Drafting Halladay at \$X neither raises nor lowers your odds of winning your league”). Why is this good? Because now, to find the value of a player, we need only to find the price at which ownership of the player doesn’t alter your odds of winning. There are no other calculations—no defining of the spread of player stats, no breakdowns of categorical value.

Note that this method works in fantasy because we have a fixed budget. In the real world, things are looser—there is no price at which owning C.C. Sabathia “hurts” your odds of winning. However, real businesses are in the business of maximizing profits, and C.C.’s salary can surely hurt those.

So we have the bare bones of an approach. Let’s create a two-team league. (In this exercise, we’ll stick with pitchers, so that we don’t have to worry about accommodating multiple positions.) On one roster, we’ll put our player of interest—in this case, Roy Halladay. Halladay always appears on this roster. The other eight slots on Roy’s roster, and all nine slots on the other one, are open:

```Roster #1              Roster #2
============           =========
Pitcher                Pitcher
Pitcher                Pitcher
Pitcher                Pitcher
Pitcher                Pitcher
Pitcher                Pitcher
Pitcher                Pitcher
Pitcher                Pitcher
Pitcher                Pitcher```
The open slots will be randomly filled with 17 distinct pitchers (no duplication within or across rosters.) After populating the rosters, we will determine the side that “won,” based on whatever categories we have in our league, and behaving as if these were the only two teams in our league. For example, in standard 5x5 roto league, there would be five categories—wins, saves, ERA, WHIP, and strikeouts. Finishing first in a category in our two-team league is worth two points, and finishing last is worth one. We’ll repeat this exercise 1,000 times for various roster configurations and track the winners.

(Why do we need to track only two rosters, even if our real league has more teams? Because each Halladay-less roster is identical. Suppose that there are 10 other rosters like Roster No. 2. Each is indistinguishable from Roster No. 2, because all rosters draw from the same pool. If we can balance Halladay’s roster with Roster No. 2, then we’ll also have balanced Halladay’s roster with the other rosters. A one-in-two chance of beating Roster No. 2 equates to a 1-in-12 chance of beating the league.)

Our ultimate aim is to make Halladay expensive enough that his team wins exactly half the time. “That’s swell, but you have no dollar figures. So you can’t turn your probabilities into prices.” And that’s true. We need points of reference.

How many points? Perhaps as few as two. If we have two points of reference, we might be able to adapt the method of parallax, which is used by astronomers to determine the distance to stars. But that’s getting ahead of ourselves, because we don’t have two points of reference.

But we do. For any fantasy league, there are two statements that we can say with certainty (both statements require us to identify the draft-worthy pool of pitchers—we’ll tackle that later):

1. The last drafted player is worth \$1.

2. The worth of a slot that freely floats among all draft-worthy players is the average price spent on that slot. If owners in a 12-team league historically spend \$99 on nine pitchers, then a pitching slot that freely floats among all 108 draft-worthy pitchers is worth \$11.

Now, in a real auction, you can’t draft a “freely floating” slot. However, in our simulation, we can—in fact, in our diagram, each slot labeled “Pitcher” is exactly that. In a particular run of the simulation, the slot could be worth \$1, or it could be worth \$50. But the expected value of the slot is \$11. (Actually, it is slightly less, since one pitcher—Halladay—is not available. But \$11 works for our purposes.)

Armed with our two points of reference, we can employ parallax. Here’s the approach: Roster No. 2 will never change—it will always contain nine freely floating pitching slots. For our first 1,000 runs, Roster No. 1 will also be the same. Over time, though, we’ll swap free-floating slots (worth \$11) for the last drafted player (worth \$1). Each switch means a drop in value of \$10 for Halladay’s team.

Eventually, we’ll reach a point at which Halladay’s roster wins only half the time. Since the odds are the same, the total value of each team must also be the same. We know the value of Roster No. 2 (\$99), and of the non-Halladay slots on Roster No. 1 (either \$1 or \$11), so it’s easy enough to solve for Roy’s value.

If we replace all eight floating pitchers, we could end up with a graph like this (not real numbers):

Here, when Halladay is paired with eight freely floating pitchers, his team wins more than 75 percent of the time. However, when he’s stuck with eight \$1 pitchers, he wins only about 15 percent of the time.

To find Halladay’s value, just read off the point at which the trend line crosses 50 percent. In this case, that’s around 3.5. So Roster No. 1 would be balanced with Roster No. 2 if 3-1/2 slots worth \$11 were replaced with the same number of slots worth \$1. Ergo, Halladay is worth \$35.

That’s the idea, anyway. Will it work?

NEXT WEEK: Will it work?

Compiled by THT Staff.

Dave Studeman said...

That’s a compliment!

Posted 09/18  at  10:39 AM
Ed D. said...

Hi John.  Interesting stuff, as always.  I’m curious, if and when you run this type of valuation for the entire player population, for your “certainty statement #2” wouldn’t you just use \$260/23=\$11.3 as your free-floating value for a standard league, or perhaps adopt a “normal” 65/35 hitting/pitching split and use \$260*.65/14=\$12.1 for free-floating hitters and \$260*.35/9=\$10.1 for free-floating pitchers?

Second question, remind me, how is this different than your WOW simulations from a few years ago?  (I get that the simulation mechanic is different, but I’m wondering if you’d expect the ranked order of players to be different between the two systems or if the ranked orders would be the same, just with WOW giving you a % likelihood of being on a winning roster and this giving you an actual dollar value).

Posted 09/18  at  02:10 PM
John Burnson said...

Brian, Thank you. I had to go deep into my inner nerd to pull this one out.

Ed:
(1) Yes, when I used “\$11”, I meant \$11.30. In any discussion of budgets, the hitting/pitching split is a gray area—I am not convinced that there is any more rigorous way to determine the proper split than to look at historical bidding records for one’s league. For now, I’ll probably use \$11.30 for every slot (hitters and pitchers).

(2) This builds on WOW, but I was never able to turn WOW into dollar values (or, rather, dollar values that I could accept). Whereas this approach, I think, is sound. I do expect the ranked order to be the same as with regular WOW.

Posted 09/18  at  03:02 PM
ourcellardoor said...

This is one of the lamest posts I’ve ever read…you must have been on drugs
guy and real real high…..I guess I’m pretty lame for taking the time to post a comment on this post…

Posted 09/18  at  04:19 PM
MrLarryDavid said...

How is this lame?  Because you’re not bright enough to understand what he’s trying to do?

Good work, John: I’ve been thinking about this question for a little while and am wondering what kind of results you’ll get.

Posted 09/18  at  06:33 PM
archilochusColubris said...

Yeah i’m quite impressed with the creative, ingenious method you came up with. Kudos for some great work John.

Quick Q: when you’re running this simulation pre-season, are you simply taking your 108th best pitcher projection to plug in for the 1\$ player, and filling the rest of the slots with a sample from the top 108 projections?

Posted 09/18  at  08:00 PM
John Burnson said...

I’ll cover this more next week, but the answer is essentially yes. The one tweak is that, instead of singling out one player as “last drafted,” I use a pool of the last 6 drafted plus the first 6 non-drafted. These guys are all bordeline, and in a given run, any of them could be “last drafted.”

Having a pool of last draftees keeps the program from becoming biased if, say, the original “last drafted” contributed only to Saves and nothing to the other categories. Also, when you go to stack Halladay’s roster with multiple \$1 players, they aren’t all the same guy.

Posted 09/18  at  08:38 PM
KY said...

I’m not convinced of some assertions here.
“Because now, to find the value of a player, we need only to find the price at which ownership of the player doesn’t alter your odds of winning.”
What the difference between this and saying, if Halladay is worth \$30 and I get him for less than that, I have upped my chances?  The price generated by the system you outlined in your fist paragraph must do that exact same thing or it is failing at properly pricing players.

I guess I’m lost on how this system will create a more accurate measurement for Halladay that the system in the first paragraph.  They both generate a dollar value, why is this way more accurate?

Posted 09/19  at  10:49 PM
John Burnson said...

The main attractions of this method are that it is interesting and more versatile. Is it “more accurate”? Remains to be seen, though I can’t believe that the traditional piecemeal method of apportioning value can’t be improved on. Still, any difference is probably not large, and a drawback of my approach is that you won’t get a definite result: In one cycle of simulations, Halladay might be worth \$31, and in another, \$29. There may be value is showing player values as a small spread rather than as a single number, but it’s bound to frustrate some people.

Posted 09/19  at  11:31 PM
KY said...

“but it’s bound to frustrate some people.”

...especially if it hasn’t been proven to return a more accurate dollar value yet.

I don’t think the normal method is piece mail at all.  Its 100% calculated in a simple way.  You take the best projections, assume those will be the stats for the end of the year, and divide them by the dollars available to spend.  Each player gets their portion.  If you used the actual end of year stats and a 50/50 split for dollars you would get exact dollar values for every player that perfectly reflect their value.

If someone spent a penny more then your dollar value using end of year stats and the other guy in your two team league spend a penny less, he would lose 100% of the time.

There’s nothing incorrect about the regular method, its the fact that the 65/35 split is up for debate, players enter the pool mid season,  that many league allow keepers and that nobody performs to projection that cause the dollar values to not be 100% accurate.  Now if you want to debate how to quantify those, that would add value.

I’d love to explore a better way but it just feels like here you are misrepresenting things to say the old way “seems wrong” and that this way, as of yet, has value.

Posted 09/20  at  12:36 AM
John Burnson said...

KY, So your first step is “You take the best projections”? And how do you determine the “best projections,” when you have yet to assign dollar values?

You’ve had two posts now in which you could have attacked the logic in my article. Instead, you’d like to turn this into a contest of “accuracy,” which, you’ll note, is not a word that I used in my article, not because I don’t think the method is accurate, but because I don’t think that’s the highlight of it. I didn’t set out to write 1,000 words on my problems with valuation methods, and I don’t plan to do so here. That’s not my target.

It’s possible there’s a flaw in my logic. But if the logic is sound, then the values are accurate. At that point, the challenge will be to explain differing prices between this method and others.

Posted 09/20  at  10:06 AM
KY said...

“The Best Projections”  Projections here refers to stat projections.  Pujols will hit .325 34HR 111RBI and 9SB next year for example.  Projections, as I was referring to them have nothing to do with dollar values.  They are the stats you use to create dollar values.  The original method is 100% accurate in creating dollar values if you were able to use end of year stats to generate dollar values.  If someone were to use a different system and apply it to the end of year stats, unless it too was a system that translated the stats into dollar values 100% accurately, it would lose to the original system.  The best it could do is tie it.

There’s nothing wrong with the logic of how your system works so why would I attack it?  My posts were not about how your system worked internally to itself, they refer to how it would match up against the old system.

I’m saying two things;

1)  I think you were not accurate when you said the original system had flaws.  The flaws in the system come from the data inputs, not the way it calculates dollar values.
2)  I don’t understand the point of your new system.  I don’t understand its value.  Or are you saying you don’t, as of yet, know if it has value and are just getting it out there?

Posted 09/20  at  12:16 PM
John Burnson said...

KY, I did not ask where your projections come from. I asked “How do you determine the ‘best projections’”?

You seem 100% devoted to the current system. Presumably you have had success with it. Me, I do not believe that our methods are as sound as they can be. What I am trying to do here is explore an alternate and untried means.

Posted 09/20  at  12:27 PM
KY said...

The best projections are the ones that come the closest to the eventual end of year numbers.  The only thing you can do is look at historically who has produced accurate projections in the past and hope they do so again.

I’m not devoted to anything. What I’m saying is, if you give the original method the end of the year’s results, it will beat or tie any system, because it is simply doing a calculation from what we already know has happened.  There is no way to improve upon a calculation that is saying 1 = 1.  At the end of a given season .287 in 50AB, 4HR, 8RBIs and 1SB, given a certain player pool, correspond exactly to some dollar value.  1 to 1.  If you used Marty McFly’s sportsbook and were able to know the end of the year’s results at the time of an auction, an auction that allow no keepers or pickups, you would walk into that action knowing exactly what dollar value each player available will end up being worth.  All that would remain would be for you to bid less than those dollar values while still winning the players and accumulating value.  If you know a guy is going to end up being worth \$30 (because you have McFly’s book) and you buy them for \$15, your team just moved up from the middle of the standings.  Towards the top.  Some other team will have moved down as well because you just took more value out of the pool then you paid in in dollars.  As long as you walk out of the auction with the most savings of any owner, you will win the league.

What I’m saying is, there is no belief in the above.  Everything above can be mathematically proven correct.  I didn’t understand that you were claiming the above was up for debate.  Now I do, but I do not agree that it should be.

Also, if I am wrong and there is something that is incorrect about the above, I’d love to learn about it.

Posted 09/20  at  12:44 PM
John Burnson said...

We’re starting to talk past each other. I’m concerned not with the best projections but with the truest dollar values for a given set of projections. How did McFly come up with \$30?

You said that the simple way is “You take the best projections, assume those will be the stats for the end of the year, and divide them by the dollars available to spend.” I don’t think that way is quite so simple. But this isn’t the forum for hashing that out.

Posted 09/20  at  02:00 PM
KY said...

Ignore projections.

What I’m saying is, if you take the stats at the end of the season and create dollar values from them using the old system you will get exactly how much each player was worth that season with 0 error margin.  At the end of the season each player is worth a particular amount based on what they produced.  Given a particular league, at the end of the season you can say “Albert Pujols was worth \$64 this season in this league.”  There will be no debate as to whether that is true, it will be a calculated value and be mathematically correct.  If you said, “no he was worth \$62” you would be lying.  To the best of my knowledge.

Posted 09/20  at  04:23 PM
John Burnson said...

Anyone who believes that the old system produces dollar values with “0 error margin” and “no debate as to whether that is true” will have no need of the ideas in my article.

Posted 09/20  at  04:46 PM
KY said...

If your inputs to it are the end of year numbers.  how can it not?  Its a calculation.  xHR = x\$

If you think the way many people and draft sites calculate xHR = x\$ is wrong you should publish that article because it would be very important information.

Posted 09/20  at  06:19 PM
thumble said...

John - Good start here, I like that you approached this from a business model POV rather than a projection POV.

KY -  “...if you take the stats at the end of the season and create dollar values from them using the old system you will get exactly how much each player was worth that season with 0 error margin.” - Sadly, we all draft at the BEGINNING of the season where our error margin can be quite large.

Posted 09/21  at  01:42 PM
KY said...

Right, but the point is the calculation of the dollar values by the old method is only wrong because you plug projections into it.  If you plugged the end of the season stats into it you would get accurate dollar values.

When John determined how many \$1 players he would have to add to balance Halladay he gave the Halladay some stats and he gave the \$1 players some stats.  That is how he could figure out how many \$1 players he needed to balance Halladay out.  Those stats are the projections.  Both systems must use some projection to calculate dollars from.

I don’t see how its mathematically possible to beat the old system, since you have to give every system some projection to start from.  And if you give the old system the end of year totals, it will produce equivalent dollar values to exactly how much each player was worth.  How can any newer system beat an exact match?

I know John wants to explore new systems, exploration is fun, I’m asking the question, why do we think it is possible to improve it?  It appears to me that it can not be done because the old system is a 1 to 1 calculation.  Its math.  xAVG + xHR + xRBI + xR + xSB = x\$.  I wish I understood why he thinks it is possible to improve on that formula because I’m not trying to be difficult, i just don’t get how it could be possible.

Posted 09/21  at  02:36 PM
John Burnson said...

KY, There is nothing simple about turning “xAVG + xHR + xRBI + xR + xSB” into “x\$.” That is as true of the old system as of this one.

There is a popular approach that gives (at a minimum) reasonably accurate prices. But it is not simple, and it is certainly not elegant.

I don’t have a ready link, but do some googling on methods for turning projections into prices.

Posted 09/21  at  02:52 PM
Derek Carty said...

Just gonna jump in here quickly.  Even if we know exactly what a player will do in all 5 fantasy categories with 100% accuracy, I’m not sure we are able to turn that into a dollar value with 100% efficiency, at least not yet.  I’ve yet to see a perfect system to do it.  KY, I’d be curious to know what method you use.  I know that I, personally, have advocated Standings Gain Points in the past, although I will be the first to say that the method has some flaws.

Posted 09/22  at  12:35 AM
KY said...

Hi Derek, always liked your blog.  I switched from standings gain to z-scores after reading the below discussion.  I won my league with Standing Gain two years but still thought Z-scores made more sense.  My main league has only 8 categories and used OBP so I can’t get that much history, 17th season but still not much.  Even if I could I’d contend, as Tango does below, that Z-scores are more accurate.  Perhaps we should have a Standings Gain vs. Z-score faceoff?  We could use this years projections and the end of year results.  I’m not sure how that would work but it would be very cool.  But that is a long discussion.  Anyhow, link;

http://www.insidethebook.com/ee/index.php/site/article/the_worth_of_sb_hr_and_all_other_categories_in_fantasy_baseball/

The method I refer to is the one John referred to at the beginning of the article “A common approach is to sum the stats above replacement level in a category and then divvy up those stats among a portion of the total budget and add up the contributions for each player.”  You use standings gain to determine how many points those stats are worth and then translate points into \$s.  I use a z-score, how much production above average they are, to determine the worth.  So for z-scores, this ratio is calculated, not dependent on league histories of how much the stats move up a category.  Using end of year stats with this method will turn the generation of dollar values into a calculated ratio.

Posted 09/22  at  10:41 AM
John Burnson said...

Here’s one problem with current valuation methods. The problem is that, on Draft Day, you are not drafting players, you are drafting SLOTS. Now, if you draft a player who actually plays all season (and whom you keep for all season), then the slot equals the player. But if you draft a player who won’t get called up until the second half, then the slot will be worth his stats plus the stats of whomever you use in the interim. Putting a price on a player who played only half the season (and as if the slot was filled only half the season) is silly; no owner would have held the slot vacant for three months. But one of the things you get with this player is “Three Months Free Space!!” and that’s not valueless.

I’ve previously created for myself a version of the method that I am discussing here that *does* handle slots. The idea was, for every calendar week during the season that a player did not play in the majors, you would add one week’s worth of replacement-level stats for the player’s position. The premise is that the player was not active and so his owner replaced him until he returned. I think this is a step in the right direction, but even though the “watering down” of players’ seasons was defensible, the final values were hard to love, because there was this element of alienness. But that doesn’t mean they were not truer.

Here is another problem: What’s the appropriate pool for comparison? Most z-scores are calculated against the league as a whole. But the league as a whole is not draft-worthy, or even plausibly draft-worthy. (I would guess that GM’s draft 95% of projected SB but only 75% of projected RBI.) In a league with 108 pitcher slots, the pool of plausibly draftable pitchers may be only 130 or so pitchers. For best results, we should probably pit these players against only one another. This is something that current methods could handle, but I have rarely seen it. (In my simulations, I ignore players with little playing time. They’re OWNABLE, to fill a slot for a bit, but they’re not DRAFTABLE, at least not in hindsight, and they shouldn’t clutter the pool.)

And I haven’t even mentioned multi-position eligibility, which is genuine player-added value but which is, at best, a post-valuation kluge in current methods.

Should I go on? I’m not saying that the method that I am introducing here solves all this; it doesn’t even attempt to. What I’m saying is that there is unused design space in player valuation.

Posted 09/22  at  12:42 PM
KY said...

Yes, when I spoke I spoke of a league in which you could not drop or add anyone and in which there is no bench slots, everyone starts.  As most auction roto-leagues are as far as I know.  Because all other situations make the example more complex and do not illustrate the point that xStats = y\$.

After that point, you begin doing adjustments such as these.  To quote the Art McGee book Derek referred to as the standings gain method, the adjustment of a minor league player who you can bench for a while is an “option” after the initial calculation.  Just like you mark up a young guy who may be a keeper a couple dollars.  That option of keeping him is worth some \$.

Any system you create will have to reward players who play the full year more then players who do not.  Unless you have a bench, which many leagues do.  But you get the point that this is only complicating the illustration and that ALL systems will have these add on options that adjust your initial dollar values, which you use stats to generate.

As to your second example I started doing exactly that this year as well.  When I price Chipper Jones I price him plus however long I estimate he will be on the DL worth of replacement stats.  But again, this is an add on to that original price generation based on Chipper’s stats.  My league does allow pickups of course so if a guy gets injured it does make sense to add this to the dollar value.

As to your third problem.  You do have to use an original z-score to decide who the “draft worthy” players are.  But you can also just eyeball the bottom of the player pool too.  Once you have established the player pool of 200 players or whatever who will be drafted, you can run the z-scores against only those players.  If you haven’t seen this much then I think that may be the problem, it seems like this is an essential step in making the \$ value a 1:1 calculation.  And yes I agree its probably rare, but that’s why we have you expert guys!

Eligibility has some value yes, but that is yet another option you have to add on to tweak the values as you say.  I’m ok with that.  When you price the player you put them in their position that they can be the best over replacement.  If I have Mark DeRosa I’m marking him a 2nd baseman for the initial calculation as he will have his highest total above replacement at 2nd instead of replacement over an outfielder.  That’s where their value is maximized at draft time.  Its nice that you can move them later and you can add whatever amount you feel that is worth.  Now that worth I can see as a place simulations could help quantify!  Run a league 1000 times and see how often DeRose ends up being needed at short or OF and how he enhanced his teams value in doing so.

So no, you shouldn’t go on, but what I was saying regarded the price before all that.  Which is the same price you are generating when you come to the conclusion that Halladay is worth \$35.  The price before all that.

Posted 09/22  at  01:25 PM
John Burnson said...

“When I spoke I spoke of a league in which you could not drop or add anyone and in which there is no bench slots, everyone starts.  As most auction roto-leagues are as far as I know.”

I was not introducing bench slots. I meant player movement via injury or promotion/demotion.

“Any system you create will have to reward players who play the full year more then players who do not.”

In a world of “slots,” there need not be a reward for playing full time. There can be value in missing time (that’s why I root for my bad players to be sent down!).

“eyeball”

I just wanted to mention that you included the word “eyeball” in your writeup. Burnson’s Law: Any algorithm with the word “eyeball” can be improved.

“So no, you shouldn’t go on, but what I was saying regarded the price before all that.  Which is the same price you are generating when you come to the conclusion that Halladay is worth \$35.  The price before all that.”

I still don’t agree that that’s necessarily so. The z-score method trusts that the only thing that matters is the distance in standard deviations from the mean. I can imagine curves where that doesn’t suffice.

Is it impossible that the simulated leagues could tease out other variables? You seem to be ruling that out, whereas I am not. I’m not ruling it in, I’m just not ruling it out.

Posted 09/22  at  02:07 PM
KY said...

i don’t agree with any of that.

“I was not introducing bench slots. I meant player movement via injury or promotion/demotion.”  Bench slots are for all intensive purposes the same as injury time.  the time in which you can replace the player with someone else.  again, this is an add on benefit to your original calculation so its not the point.

“Any system you create will have to reward players who play the full year more then players who do not.”
I mean only that a Chipper Jones who plays 162 games is worth more than one who plays 140 + 22 replacement player games.

“eyeball”
I specifically said you don’t have to eyeball it, since your eyeball is almost as good as calculation, you save the time in this case.  you could calculate it, as you do in taking an average of the bottom 30 players or so to define replacement level.

Yes, simulations would help with lots of “variables”.  And that’s what I said above when I said a simulation would be a good place to figure out how much position flexibility is worth.  But I also said that all of those above things are not part of the original generation of dollar values from a set of projections.

If you leave all of those things out as add on problems, which they are, a simulation will not beat a calculation.  What’s more, once a simulation has helped you quantify the effect of slotting/position flexibility/ect, and you incorporate those results into your \$ calculations as add-ons, a simulation will no longer be able to beat the calculation again.

It also did not appear that your simulation was addressing any of these add on problems, only the initial, “Given Roy Halladay’s stats, how many \$1 player stats does it take to balance him?”  Where you have defined Halladay and \$1 Guy’s stats already.  I.E.  Whatever you defined for them are what I call “projections”.  And it appears to me that is the one part the simulation can not improve, because it is a calculation.

You could calculate the value of position flexibility in a given league too if you took the time, but a simulation may give you almost the same value faster, and therein would be its value.

Posted 09/22  at  02:22 PM
KY said...

I’m curios what a “I can imagine curves where that doesn’t suffice.” player pool in a league with no add drops using end of year stats would look like that would cause the z-sore method to generate dollar values that can be beat?

If there are no add drops and you use end of year stats all of those above problems of position and slotting are removed I believe.

Posted 09/22  at  02:25 PM
John Burnson said...

KY, As I said a while ago, you have complete faith in z-scores, which means that you have complete faith that player values can be no better described than as the sum of the number of standard deviations from the mean. Whether my lack of a similar conviction is a strength or weakness remains to be seen, and won’t be answered here. Adieu.

Posted 09/22  at  02:35 PM
KY said...

And it would be really nice for this conversation if you had provided a single reason z-scores would not give you perfect dollar values using end of year stats with no substitutions.  Since that what I spent a day asking for.

But instead you just keep saying, “I don’t think they do that.”

Posted 09/22  at  02:49 PM
John Burnson said...

KY, Distributions of numbers are characterized by more than their mean and standard deviation. Do you grant that? You must, because it’s true (for example, skewness). If you do grant it, then why should we assume that a calculation that considers only mean and standard deviation is perfect?

The nice thing about the simulation is that, in theory, it should pick up mean and standard deviation as part of its laborings, along with anything else of import. So it shouldn’t have any LESS information than z-score, and it may have more.

I should have stated this plainly 20 posts ago.

Posted 09/22  at  03:02 PM
KY said...

Yes, thank you very much!

Posted 09/22  at  03:09 PM
Peter Kreutzer said...

I enjoyed your supernerdy story John, and am back here hoping for Part II. Let it rip.

I actually understand KY’s point, but I think s/he’s giving too much credit to the efficiency of the drafting process.

I’ve conducted auctions using after-season stats. That is, 2006 season auction after 2006 is done. The fact is that the roto system is too complex to give you a single \$ value. The way the game is played, your values for the remaining players vary quite a bit depending upon who you have already rostered.

Prices are dynamic. With each player who is chosen, the value of all the remaining players changes. And they change in different ways for different teams, depending on who each team has rostered.

This is true, KY, even in an Auction and Hold format of post-season drafting.

What John is doing that is interesting is coming up with another model to derive values. This is how we learn stuff. I don’t know yet if there is good info coming out of his modeling/simulating. But thinking about things in different ways, and applying different approaches, gives us a chance to learn things we would never otherwise learned.

What I learned from conducting a post-season draft is that strategic agility is way more important than the actual values we give players on draft day. Melding a team is of the utmost importance, not only ticking the spreadsheet.

Posted 09/27  at  11:59 PM
Page 1 of 1 Commenting is not available in this weblog entry.

<<Previous Post:  Waiver Wire: NL