Todd's Scarcity Blog

[viper]

I'm never sure which area is best to place these posts. Hopefully Todd will move them as he seems best.

Todd's blog states correctly that replacement player scarcity, aside from the catching position, does not exist in NFBC sized leagues to include those 12+ leagues of AL-only and NL-only. That final UTIL slot can essentially be from any non-catching positions. What he talks about is, what I will refer to as, "interim scarcity". Within each position, there are areas that interim scarcity exists. Within every draft, category scarcity will exist. And as Todd says, each position contributes differently to the sum of all the available fantasy points. About a week ago, I started examining the contribution value mid-point for each of the four positional groupings [CA, CO, MI, OF/UT].

My other "study" is on category tiers but that is for another post.

Bring it on Todd

[Todd Zola]

Jumping ahead, what we are going to find is what I will call "fact". This will be represented using the tiered approach discussed elsewhere. These "facts" will revolve around how each of us perceive/value the player pool.

That said...

1) My "facts" may differ from your "facts", based on how each of us value the players. An example is how I presently evaluate the closers. I see risk in assuming great things from Bell/Marmol/Devine/Broxton/Corpas/Perez/Francisco/Qualls etc -- others may trust them more and therefore see a very plush second tier.

2) Even if we agree on the "facts", our strategy to best take advantage may differ.

[IronMarshal]

Viper,

I would add that scarcity can be created during the draft, but in risky areas such as saves.

I think of scarcity in catagories as well as positions.

Some guys will try to corner the market in certain countable catagories. This can create a scarcity when the player pool, evenly distributed, does not have scarcity.

Todd's blog piece really touched on where I think he is going. That is to look at valuation in terms of a big picture plan rather than the usual focus on the usual chasing of stats that many of us do.

I catch myself sometimes chasing stats rather than looking at what would be available if I examined stats from positions and what will be available later. Tiering can be very helpful with that.

I have to really look at this this year as each year I need to approach the auction differently because the guys like to follow my lead on auction strategy. I need to stay one step ahead of them, and this year, my valuation may be different than in the past because of that.

[Markus_mb]

Todd's Example:
POOL A: Red 45, Blue 40
POOL B: Green 20, Black 10

I think this is an *excellent* illustration of the ONE dimensional problem, that dimension being players "mean value".

But the problem is not 1-dimensional - players do not automatically return their mean value. At the least there is a second dimension of "variance" around that mean that has to be considered. Let me adapt Todd's example incorporating this second dimension:

EXAMPLE 1
POOL A: Red 43-47, Blue 38-42
POOL B: Green 10-30, Black 0-20

Todd, how to do you view this problem? Do you still want to take a very "likely" $5 loss in Pool A for a somewhat less likely chance at a $10 gain in Pool B (since great overlap of Pool B distributions)?

EXAMPLE 2
POOL A: Red 30-60, Blue 38-42
POOL B: Green 18-22, Black 0-20

This time I've made one choice from each pair highly variable and the other narrowly confined. In this case I think I would prefer the upside of the more variable players. Because the first player in Pool A has a $5 higher mean it limits the cost of his downside. The cheaper player in Pool B still has a chance to earn what the expensive one does, and at these prices may not be that hard to replace.

How would you reason example B?


And to add a third dimension - shape. Most players have a negative skew - most outcomes near a value but a few tailing way off (injury, role, etc). But some players negative skew is a lot more than others......Top 4 round players tend to have more limited negative skew because they are established and proven. On the other hand, some players will have positive skews aka upside - perhaps a Colby Rasmus or Kendry Morales type. And those skews of distributions around the mean can change the analysis also.

My Take: Position scarcity *as a rule* is foolish because there is no general principle governing all cases. Rather you have to look at the specific mean expectation, likely range of outcomes, and probability shape/skew for those outcomes on a case by case basis. Sometimes taking the scarce player will be the right moves, other times not.

Let me add that for Pedroia and Kinsler there is both a large variance and a negative skew. And for that reason taking them early on the basis of scarcity is a mistake in my view.

[viper]

Taking those two player based on scarcity is not exactly what Todd is suggesting.

On the bigger scale, I think all Todd is trying to say is that it is not correct to willy-nilly take the best available player even if scarcity does not exist at the replacement level of a draft.

I personally dislike this particular 2x2x1 argument even though I understand its use as a learning tool. It makes a complex subject seem too simple. A baseball draft is likely the single most complex draft in fantasy sports. Consider the NFBC. There are 14 positional players, 15 teams & 5 categories. As if this 14x15x5 matrix wasn't complex a nut to crack, fantasy baseball adds a second simultaneous draft for pitching which is a 9x15x5 matrix. Plus there are no rules as to the order that each owner can fill his pair of matrices. You may start a draft with 5 hitters while another might start with 5 pitchers. I'm not saying that is the best way to go but I suspect not many have actually tried that approach.

[Todd Zola]

They key to this admittedly myopic view is the only time it is "correct" is at the end of the season, using a simplistic single category league. Other than that, it is more a principle or consideration, to be factored into strategy along with a bunch of other factors. A few months ago, I visited the good people at the Rotojunkie/Bullpen forum and talked about this a little -- to quote myself after someone wondered if this was really applicable to multi-categories so a mythical replacement level player is needed to calculate useful stats....

Short answer is it won't be completely self-consistent as the replacement level player then becomes a mythical being as he must be a composite across all the categories.

Which really only further goes to demonstrate why these values are a guide, and should not be treated as a be-all-end-all quantification of value from which you build your team. There are a myriad of other factors involved, some intuitive, some can be learned and some you just have to gain from experience. They involve risk and category balance and knowing your league's drafting and trading tendencies etc. They involve knowing your strength in terms of identifying certain types of players and setting a strategy to best take advantage of those traits. They involve studying each season's player pool to see where there are pockets of value and where there are dropoffs.

And there's more too, but I'm pretty sure people get the idea.

And this is just in a REDRAFT league. Keeper and or dynasty formats are a completely different animal.

Todd's Example:
POOL A: Red 45, Blue 40
POOL B: Green 20, Black 10

I think this is an *excellent* illustration of the ONE dimensional problem, that dimension being players "mean value".

But the problem is not 1-dimensional - players do not automatically return their mean value. At the least there is a second dimension of "variance" around that mean that has to be considered. Let me adapt Todd's example incorporating this second dimension:

EXAMPLE 1
POOL A: Red 43-47, Blue 38-42
POOL B: Green 10-30, Black 0-20

Todd, how to do you view this problem? Do you still want to take a very "likely" $5 loss in Pool A for a somewhat less likely chance at a $10 gain in Pool B (since great overlap of Pool B distributions)?

EXAMPLE 2
POOL A: Red 30-60, Blue 38-42
POOL B: Green 18-22, Black 0-20

This time I've made one choice from each pair highly variable and the other narrowly confined. In this case I think I would prefer the upside of the more variable players. Because the first player in Pool A has a $5 higher mean it limits the cost of his downside. The cheaper player in Pool B still has a chance to earn what the expensive one does, and at these prices may not be that hard to replace.

How would you reason example B?

The short answer is I would have to think these through -- assuming the rules are you are stuck with these players and cannot replace them, I don't have an answer simply upon inspection.

What I do know is in a "real" fantasy league, you have 21 other picks to consider in concert with this one which speaks to the quote above.

And to add a third dimension - shape. Most players have a negative skew - most outcomes near a value but a few tailing way off (injury, role, etc). But some players negative skew is a lot more than others......Top 4 round players tend to have more limited negative skew because they are established and proven. On the other hand, some players will have positive skews aka upside - perhaps a Colby Rasmus or Kendry Morales type. And those skews of distributions around the mean can change the analysis also.

My Take: Position scarcity *as a rule* is foolish because there is no general principle governing all cases. Rather you have to look at the specific mean expectation, likely range of outcomes, and probability shape/skew for those outcomes on a case by case basis. Sometimes taking the scarce player will be the right moves, other times not.

Let me add that for Pedroia and Kinsler there is both a large variance and a negative skew. And for that reason taking them early on the basis of scarcity is a mistake in my view.

There are some that approach the Kinsler/Pedroia variance using a weighted average to come up with the single-value projection. It is then up to the individual to determine how they want to handle the risk/variance aspect.

With Kinsler, health is more the issue. For the sake of this discussion, let's pretend his rate of performance will be the same over a range of AB, even though one may argue an injury will alter the rate. He hits HR at a rate of about 1 per 27.4 AB.

400 AB = 14.6 HR
450 AB = 16.4 HR
500 AB = 18.2 HR
550 AB = 20.1 HR
600 AB = 21.9 HR
650 AB = 23.7 HR

Now you factor in a subjective estimate of the chance he accrues each of those totals

400 -- 5%
450 -- 15%
500 -- 30%
550 -- 25%
600 -- 15%
650 -- 10%

(for entertainment purposes only, not saying this is how I would do it)

.05 x 400 + .15 x 450 + .30 x 500 + .25 x 550 + .15 x 600 + .1 x 650 = 530 AB

530/27.4 = 19.34 or 19 HR

so the "projection" will say 19 HR, but there is a defensible range between 15 and 24.

You can do the same thing with Pedroia, but in his case, the playing time is stable but the rate of performance is variable.

I'm not sure you can or even if you should WANT to quantify all this. With 23 roster spots divided into hitting and pitching, and each of those further divided into subsets (speed/power and strikeouts/saves), I think the ART of drafting supersedes the SCIENCE of drafting.

[Markus_mb]

I personally dislike this particular 2x2x1 argument even though I understand its use as a learning tool. It makes a complex subject seem too simple. A baseball draft is likely the single most complex draft in fantasy sports. Consider the NFBC. There are 14 positional players, 15 teams & 5 categories. As if this 14x15x5 matrix wasn't complex a nut to crack, fantasy baseball adds a second simultaneous draft for pitching which is a 9x15x5 matrix. Plus there are no rules as to the order that each owner can fill his pair of matrices. You may start a draft with 5 hitters while another might start with 5 pitchers. I'm not saying that is the best way to go but I suspect not many have actually tried that approach.

Well said.

[Markus_mb]

I think the ART of drafting supersedes the SCIENCE of drafting.

I basically agree. But I have to say, once they programmed a computer to beat the world masters in chess my thinking on art vs science changed a bit. I suspect that if someone wanted to invest the (enourmous) time and money to do it, most of the "art" we use in our decisions could be modeled effectively by a computer.

[Todd Zola]

I think the ART of drafting supersedes the SCIENCE of drafting.

I basically agree. But I have to say, once they programmed a computer to beat the world masters in chess my thinking on art vs science changed a bit. I suspect that if someone wanted to invest the (enormous) time and money to do it, most of the "art" we use in our decisions could be modeled effectively by a computer.

I was actually thinking of the chess analogy earlier. When it comes to chess, I know the rules, I have played casually, that is the extent of my chess experience, so please feel free to correct any fallacious assumptions I make.

I believe the fundamental difference is chess is static, whereas fantasy baseball is fluid and has the types of risk variance discussed already.

Again, I am naive to this, but something I have always wondered about is why there is not only 1 game of chess ever played between Masters. Seems to me there must be "best" first white move, the "best" first black move, the "best" second white move, etc. Intuitively, my guess is the present computers are capable of discerning this series of "best" moves, but even the best players can only know/learn/remember a finite amount of these moves so either eventually they veer off from the "best" moves or the purposely veer off and hope their opponent is unable to figure out the best counter to their "not-best" move.

In real simple terms, in tic-tac-toe, there is a best X move, followed by a best O move, etc. That said, there is no series of "best" moves that assure either X or O of winning, both are waiting for their opponent to make a mistake. I wonder if this is the same with chess, that if black counters each white move perfectly, will the game end in a draw?

Jumping to fantasy baseball, yeah, there probably is a way to incorporate these risk issues into an automated determination of the "best" selection -- meaning more often than not, you are making the best decision based on the range of possibilities.

In tic-tac-toe and I assume chess for the WHITE pieces, if you make the "best" move every time, YOU WILL NOT LOSE, every time.

In fantasy baseball, if the "best" pick gets injured, you are in trouble -- which of course is accounted for in the rick. If you make that "best" pick 100 times, more often than not you will do fine.

Where it gets fuzzy is I believe what is "best" in fantasy baseball is transient, as you cannot predict how your opponents may choose their players, how they balance their hitting and pitching and the subsets therein. This is what I call the art of drafting, doing a better job of foreseeing this "human element" -- the whims of your opponents. The computer may say Smith is the best pick in round 4, but there are a series of events that could occur that in hindsight, someone would have been better than Smith.

[viper]

If all the projections were 100% correct, there would probably be a best pick at every turn but I seriously doubt if anyone would have the time. Plus, any best pick can be thwarted by bad picks by another drafter.

Call this the VIPER Challenge. I have always wondered how a draft would turn out using LAST YEAR's NUMBERS. You would know exactly how your player would preform. You would be able to determine your current standings position after each and every pick. And my guess is that every draft would be different. Without computers to track your results, my guess is that no one would be willing to bet a whole lot that they had the best draft at the end.

[Guest]

The computer may say Smith is the best pick in round 4, but there are a series of events that could occur that in hindsight, someone would have been better than Smith.

OK, given everything that's been said, we agree that if enough energy were taken, a supercomputer could calculate the theoretical "best pick".

I agree with what Todd says, but I think he'd agree that the standard isn't mind-reading, its taking into account what one knows when that round 4 pick is made, and if things change later, well thats life. The art is figuring out what needs to be taken into account, the science is somewhat the strategy coming from that artwork,

I personally dislike this particular 2x2x1 argument even though I understand its use as a learning tool. It makes a complex subject seem too simple.

It outlines the problem, while being too simplistic in aiding to a solution.

4 team league, 2 positions. HR derby.

Pos 1 - 20,15,15,1
Pos 2 - 20,3,2,1,

Just because the best player at both positions have 19 marginal HR, one cannot suggest that the distribution of stats within a position doesn't have some effect on the "right" player to take.

And to add a third dimension - shape. Most players have a negative skew - most outcomes near a value but a few tailing way off (injury, role, etc). But some players negative skew is a lot more than others......Top 4 round players tend to have more limited negative skew because they are established and proven. On the other hand, some players will have positive skews aka upside - perhaps a Colby Rasmus or Kendry Morales type. And those skews of distributions around the mean can change the analysis also.

Markus - and for the record, pumped to see your name back here - I ask a very unscientific question - a $30 player could lose $30 of value (assuming a floor of 0, which I shouldn't, a $10 can lose 10). Assuming I agree on variability, isn't the reality that the more I pay for someone, the more downside they offer? I'd be glad to hear Im wrong, but isn't the variability you speak of already taken into account in their "mean value". I understand that really reliable players might have a smaller expected variance, but I'd like to think a solid valuation/projection set at least takes this into account somewhat (ours does in its own way).

My Take: Position scarcity *as a rule* is foolish because there is no general principle governing all cases.

See my example above. Postion scarcity calculated as players in the positively valued player pool is extremely incomplete. Looking at value throughout the player pool by position can point out where "scarcity" - in quotation marks because Im not defining it the same way.

I'll define scarcity at a point of time in a draft as "no available players at this time to select". That could mean SB guys, 2B, P, whatever. If its the fourteenth round and I need SB and the highest projection guy has 12, then at that point there is SB scarcity. If I need HR and a second baseman and the highest HR second baseman has 7, then thats scarce there.

Pre-draft planning like that of the great NFBC players keeps you away from scarce situations as much as possible - in my opinion the mark of great players - making the move in round 8 that doesn't force you into a bad move in round 19.

If all the projections were 100% correct, there would probably be a best pick at every turn

Theres a best pick knowing what you know to the best of your information - why hold yourself to a higher standard?

[cwk1963]

I was 5th board on my high school chess team many years and many brain cells ago. What Todd says is very true. Chess is very static and baseball (and, by association, fantasy baseball) is very fluid. While there may be a best first white move and best first black move, etc programmed into the computer, each of those moves in and of themselves have no outside variables. That is, if whites first move is deemed to be king's pawn to k4, the final outcome will happen. There is nothing that can alter the result until black makes his move. In baseball, you can determine that the best first pick is, for example Arod. And you can make that determination based on your assumption he will produce a line of 40/110/125/310/20 in a 5x5 scenario. However, there are many things that can happen until Arod reaches k4 - he can get injured; other team-mates can get injured resulting in fewer r and rbi; he can under-produce. Each of those scenarios would result in him only reaching k3. On the other hand, he can run more; his team-mates can all stay healthy and each of them produce more; he can simply over-produce himself. Each of those scenarios could result in him reaching k5 which, in chess, is illegal on an opening move. In chess, a rook can only move as a rook; a bishop can only move as a bishop; a knight can only move as a knight, etc. In baseball, however, any player can wind up looking like a different piece based on a number of factors. So while you could probably program a computer for the science and some of the 'art' of fantasy baseball, because of all the mitigating factors you would never be able to program a computer to win a fantasy baseball league season after season.

[Markus_mb]

In real simple terms, in tic-tac-toe, there is a best X move, followed by a best O move, etc. That said, there is no series of "best" moves that assure either X or O of winning, both are waiting for their opponent to make a mistake. I wonder if this is the same with chess, that if black counters each white move perfectly, will the game end in a draw?

Yes.

In fantasy baseball, if the "best" pick gets injured, you are in trouble -- which of course is accounted for in the rick. If you make that "best" pick 100 times, more often than not you will do fine.

Yes, there is randomness in roto that isn't in chess. This randomness would prevent a computer program from winning 100% of the time. The argument though is that such a program, using probabilities built in, would fair better than any human over the long run of many trials (which reduces chance). But it took them a hell of long time to create the chess program.....I'm not suggesting this something easy or at all near. Just using my imagination!

[Markus_mb]

Call this the VIPER Challenge. I have always wondered how a draft would turn out using LAST YEAR's NUMBERS. You would know exactly how your player would preform. You would be able to determine your current standings position after each and every pick. And my guess is that every draft would be different. Without computers to track your results, my guess is that no one would be willing to bet a whole lot that they had the best draft at the end.

I have wondered too. I think the issue here is category point distributions. If player 5 takes that 1B he passes you in HR and you lose a point. If he takes that MI he passes the other guy in SB and he loses a point. Whether you or the other guy wins depends on something neither of you control - what player 5 (or all of the others) do.

I think there would be much to learn by studying a post-season draft of this kind. It would be really interesting to do one over and over and over with the same season and same drafters and see what strategies emerge and which ones do well!

[viper]

and finally realize that chess is against one opponent with a known goal. In a fantasy draft there are multiple people drafting. Some may trying to do better in one category and other favor a different category. The variables are mind staggering.

God, I love this game.

[Todd Zola]

The VIPER CHALLENGE is basically a retro draft, a few of us in "the industry" have talked about putting one together but have never done so. I did once way back when. At the time I was a bit of a hot-head, someone thought they had the best value system, they trash talked me into joining their draft, I kicked their ass and retired undefeated

[Todd Zola]

FYI -- for a future blog piece, I have been doodling with more of those HR derby scenarios like Gary proposed. The results are quite interesting and not necessarily intuitive.

Let's look at the one he just posted...

POOL A 20,15,15,1 (19,14,14,0)
POOL B 20,3,2,1 (19,2,1,0)

My first blush pick is 20B. Let's play it out...

TEAM1 20B
TEAM2 20A
TEAM3 15A
TEAM4 15A
TEAM4 3B
TEAM3 2B
TEAM2 1B
TEAM1 1A

TEAM1 21
TEAM2 21
TEAM3 17
TEAM4 18

What if I went 20A?

TEAM1 20A
TEAM2 20B
TEAM3 15A
TEAM4 15A
TEAM4 3B
TEAM3 2B
TEAM2 1A
TEAM1 1B

TEAM1 21
TEAM2 21
TEAM3 17
TEAM4 18

Same thing....

What if it were

POOL A 20.1,15,15,1 (19.1,14,14,0)
POOL B 20,3,2,1 (19,2,1,0)

TEAM1 20.1A
TEAM2 20B
TEAM3 15A
TEAM4 15A
TEAM4 3B
TEAM3 2B
TEAM2 1A
TEAM1 1B

TEAM1 21.1
TEAM2 21
TEAM3 17
TEAM4 18

or...

POOL A 20,15,15,1 (19,14,14,0)
POOL B 20.1,3,2,1 (19.1,2,1,0)

TEAM1 20.1B
TEAM2 20A
TEAM3 15A
TEAM4 15A
TEAM4 3B
TEAM3 2B
TEAM2 1B
TEAM1 1A

TEAM1 21.1
TEAM2 21
TEAM3 17
TEAM4 18

To me, this helps demonstrate why value should be determined using USEFUL stats as opposed to something using mean or average value or SD or have a 'percentage of the pool' component.

I have some examples where you don't want the first pick, you can't "win".

[Markus_mb]

Markus - and for the record, pumped to see your name back here -

Good to be back and see all you guys too!!! To quote a movie: If I may be so bold, it was a mistake for Todd to accept the promotions to [admiral]. Mastersball is his first/best destiny. Anything else is a waste of material.

And to add a third dimension - shape. Most players have a negative skew - most outcomes near a value but a few tailing way off (injury, role, etc). But some players negative skew is a lot more than others......Top 4 round players tend to have more limited negative skew because they are established and proven. On the other hand, some players will have positive skews aka upside - perhaps a Colby Rasmus or Kendry Morales type. And those skews of distributions around the mean can change the analysis also.

I'd be glad to hear Im wrong, but isn't the variability you speak of already taken into account in their "mean value". I understand that really reliable players might have a smaller expected variance, but I'd like to think a solid valuation/projection set at least takes this into account somewhat (ours does in its own way).

To this part I'd say definitely not. "Variability" is a fundamentally different concept than "Mean Expected Value". No matter how precisely you calculate that mean point there will be variability around it - so it can't account for that.

I ask a very unscientific question - a $30 player could lose $30 of value (assuming a floor of 0, which I shouldn't, a $10 can lose 10). Assuming I agree on variability, isn't the reality that the more I pay for someone, the more downside they offer?

I understand you're reasoning here and it's hard to explain my point clearly (without a novel)......probably because it's not clear enough in my head.

One point is that (a) injury, (b) role (e.g, batting 8th vs 3rd), and (c) PT are the biggest source of variance. And the range for these variables can be GREATLY reduced for the elite stars versus those lower in the pool. Injury is the hardest to reduce but generally the stars have a longer track record and we know a lot more about them because they have been followed closely by many people (i.e., interrater reliability).

Put more mathematically, when using inference to estimate what sample outcome will result from a population the variability is StandardDeviation/Root(SampleSize). So larger samples lead to a much more precise estimate. The stars generally have smaller SD to begin with (in numberator) and larger samples to judge from (in denominator).

My Take: Position scarcity *as a rule* is foolish because there is no general principle governing all cases.

See my example above. Postion scarcity calculated as players in the positively valued player pool is extremely incomplete. Looking at value throughout the player pool by position can point out where "scarcity" - in quotation marks because Im not defining it the same way.

I'll define scarcity at a point of time in a draft as "no available players at this time to select". That could mean SB guys, 2B, P, whatever. If its the fourteenth round and I need SB and the highest projection guy has 12, then at that point there is SB scarcity. If I need HR and a second baseman and the highest HR second baseman has 7, then thats scarce there.

Pointedly (just for emphasis), if you "need" anything in R14 you've probably executed your strategy poorly (or have a poor strategy to begin with). Early in the draft choices are plentiful. If you make these choices wisely (from many possible configurations) then when you reach the point where choices are not plentiful your team is already insulated against those. This is part of the reason why I believe you have to let the draft come to you and not try to force your strategy on the draft.

What Viper said is correct - there is a hitter matrix and a pitcher matrix interlaced throughout the draft. The pitcher matrix has great position flexibility. If you took 5 OF and 3 1Bs in the first 8 rounds, and in round 9 there are no C,MI,3B worth the round, you simply take a pitcher there and "bridge" yourself down the draft to the point where MI/C/3B are at value again. I don't recommend 5/3 in those 8 rounds, but I wouldn't "reach" to avoid it either.

[Todd Zola]

Pointedly (just for emphasis), if you "need" anything in R14 you've probably executed your strategy poorly (or have a poor strategy to begin with).

Or perhaps you "need" something every round, with the more times you are able to draft specifically towards that specific need, you have a good strategy and have executed it well.

Of course, your strategy might be not to "need" anything specific in the 14th round and be able to take the best value on the board. Jumping ahead a bit to a different strategy topic, this might be optimal, as we will demonstrate that most everyone has their desired pick of the litter at the end of the draft. If there are 5 guys you like that NO ONE else likes, you can get them in rounds 19-23 in a standard snake draft.

[JP Kastner]

There are other fluid systems where the dynamics have been modeled. The classic example is the financial markets. You basically make the same decisions picking stocks that you make picking fantasy players with two caveats. First, when you make an investment in a fantasy player, you have a monopoly in that player. The second is that all fantasy players depreciate in value over time.

I would argue that the same computer methods used to predict financial markets could be used to predict fantasy players. The application would calculate the risk of investment and make recommendations on future moves. ("Now is the time to draft a closer. There is a 75% chance that the top three closers will be selected before the next pick.")

There is a PBS Nova episode that discusses the development of such an application for the financial world. I'm sure you can find a link at the PBS website.

I suppose there is an additional caveat. There is a huge financial motivator to develop a tool in the financial market and absolutely no financial motivator to do it in the fantasy baseball world.

[viper]

using financial modeling for success seems like less than a great idea right now

[viper]

things are a bit more active here than on other boards Todd used to be involved with.

[Guest]

Pointedly (just for emphasis), if you "need" anything in R14 you've probably executed your strategy poorly (or have a poor strategy to begin with).

If you need an individual player, I'd agree. But in a 15 teamer, a 14th round pick (for example) would be between picks 196-210. Or on average the 7th best player on each MLB team. I'd say you're in the late teens before players get pretty fungible. Again, I get your point, but one could make a very logical argument that a well-laid plan for using picks in the teens might allow more upside/flexibility in the earlier rounds.

[Markus_mb]

Pointedly (just for emphasis), if you "need" anything in R14 you've probably executed your strategy poorly (or have a poor strategy to begin with).

If you need an individual player, I'd agree. But in a 15 teamer, a 14th round pick (for example) would be between picks 196-210. Or on average the 7th best player on each MLB team. I'd say you're in the late teens before players get pretty fungible. Again, I get your point, but one could make a very logical argument that a well-laid plan for using picks in the teens might allow more upside/flexibility in the earlier rounds.

Generally at that point in the draft I'm looking for the best player available with upside. I don't care about the position or stat upside - I'll take more HR if that's what's there.....or SB instead....or BA.....or K's. Mostly I just try to take a high quality pick.

My drafts tend to go....
- Take best player first several rounds....
- Fill in for need tactically in the middle rounds...
- Take best available player in the later rounds....

[IronMarshal]

I was 5th board on my high school chess team many years and many brain cells ago. What Todd says is very true. Chess is very static and baseball (and, by association, fantasy baseball) is very fluid. While there may be a best first white move and best first black move, etc programmed into the computer, each of those moves in and of themselves have no outside variables. That is, if whites first move is deemed to be king's pawn to k4, the final outcome will happen. There is nothing that can alter the result until black makes his move. In baseball, you can determine that the best first pick is, for example Arod. And you can make that determination based on your assumption he will produce a line of 40/110/125/310/20 in a 5x5 scenario. However, there are many things that can happen until Arod reaches k4 - he can get injured; other team-mates can get injured resulting in fewer r and rbi; he can under-produce. Each of those scenarios would result in him only reaching k3. On the other hand, he can run more; his team-mates can all stay healthy and each of them produce more; he can simply over-produce himself. Each of those scenarios could result in him reaching k5 which, in chess, is illegal on an opening move. In chess, a rook can only move as a rook; a bishop can only move as a bishop; a knight can only move as a knight, etc. In baseball, however, any player can wind up looking like a different piece based on a number of factors. So while you could probably program a computer for the science and some of the 'art' of fantasy baseball, because of all the mitigating factors you would never be able to program a computer to win a fantasy baseball league season after season.

I have been a pretty serious student of chess and a decent player. Chess is not a static game, it is very dynamic and fluid. The only static part may be considered the pawn structure. The pieces are constantly maneuvering to make slight postional improvements (especially at the highest levels) that can lead to tactical or strategic opportunities and advantages. There is no "Best" first move, although many will tell you e4 (K4) is the "best." Many great players have prefered d4 (Q4) as an opening (World Chapions like Botvinnik and Kramnik come to mind). Some even prefer Nf3. Chess is very complex, far beyond what many patzers think. Most people believe it is complex, but have no idea how very complex it is.
That said , there are parallels between the Fantasy baseball draft and chess. Both require a lot of preparation, intuition, knowledge, tactical sense and Strategy.

[cwk1963]

I was 5th board on my high school chess team many years and many brain cells ago. What Todd says is very true. Chess is very static and baseball (and, by association, fantasy baseball) is very fluid. While there may be a best first white move and best first black move, etc programmed into the computer, each of those moves in and of themselves have no outside variables. That is, if whites first move is deemed to be king's pawn to k4, the final outcome will happen. There is nothing that can alter the result until black makes his move. In baseball, you can determine that the best first pick is, for example Arod. And you can make that determination based on your assumption he will produce a line of 40/110/125/310/20 in a 5x5 scenario. However, there are many things that can happen until Arod reaches k4 - he can get injured; other team-mates can get injured resulting in fewer r and rbi; he can under-produce. Each of those scenarios would result in him only reaching k3. On the other hand, he can run more; his team-mates can all stay healthy and each of them produce more; he can simply over-produce himself. Each of those scenarios could result in him reaching k5 which, in chess, is illegal on an opening move. In chess, a rook can only move as a rook; a bishop can only move as a bishop; a knight can only move as a knight, etc. In baseball, however, any player can wind up looking like a different piece based on a number of factors. So while you could probably program a computer for the science and some of the 'art' of fantasy baseball, because of all the mitigating factors you would never be able to program a computer to win a fantasy baseball league season after season.

I have been a pretty serious student of chess and a decent player. Chess is not a static game, it is very dynamic and fluid. The only static part may be considered the pawn structure. The pieces are constantly maneuvering to make slight postional improvements (especially at the highest levels) that can lead to tactical or strategic opportunities and advantages. There is no "Best" first move, although many will tell you e4 (K4) is the "best." Many great players have prefered d4 (Q4) as an opening (World Chapions like Botvinnik and Kramnik come to mind). Some even prefer Nf3. Chess is very complex, far beyond what many patzers think. Most people believe it is complex, but have no idea how very complex it is.
That said , there are parallels between the Fantasy baseball draft and chess. Both require a lot of preparation, intuition, knowledge, tactical sense and Strategy.

It is a fluid game in that the players can make a variety of different moves on each turn - the same as a baseball draft. However, it is very static in that you have to follow the rules set up for piece 'production'. A pawn can only move 1 space at a time (except on the first move) and can only move in a forwards direction. A bishop can only move diagonally and must stay on its own color. A rook can only move horizontally or vertically, etc. In baseball and, therefore, fantasy baseball, a pawn can act like a bishop and a knight can act like a queen. That is, you have projections that you think a player will fall into but any given year any player can fall short for a number of reasons (including injury) or someone can come out of the blue to beat what was projected for them for any number of reasons (including skill-set improvement, just having a light bulb go on, or even PED's).