Non-snaking draft order

[KSB2424 3,148]

Quick answer... You still have a 47.4% to hit either red or black (on a double 00 wheel). The wheel has no memory.

 

Remember, you're talking to a chemical engineer and a professional poker player here. I know odds and statistics.

So why do you believe just because in 2006 the top 3 picks had an advantage over the last three picks that somehow predicts in 2007 the same relsult will happen? Just like the roulette wheel where each spin is seperate from the previous spin, this NFL year is a sepearate event from last year.

 

There are different coaches, different players, different weather, different offense schemes, different defense schemes, different home/away schedules, different offensive and defensive linemen which causes a million different matchups, different guys getting injuried, different guys not getting injured, different playing surfaces, different player moods, ect. ect. ect. ect.

 

How can you honestly say that you can predict the future with so many unknown variables in play?

[Mangatang 0]

How can you honestly say that you can predict the future with so many unknown variables in play?

 

You don't have to predict the future (exactly). You just have to have a decent estimate of values.

 

Are you saying that you think FFToday's average auction values for the players are COMPLETELY off? Those values are a fair market estimate of what the players are worth to fantasy football participants this year. If the snake method allows the first team to get more of that estimated talent than the last spot, then it's an unfair start.

 

Your argument should be that the variance of the predictions is so great that any reasonable advantage that the first pick has gets negated by the shear noise of the unpredictable outcome. If that was your point, I could at least acknowledge that it has some merit. Of course, in the long run (maybe dozens of years, playing many leagues a year), the advantage would start to become more significant, albeit with a high level of variance.

[ratbastard 0]

Quick answer... You still have a 47.4% to hit either red or black (on a double 00 wheel). The wheel has no memory

yes and no. in statistics we often speak of the possibility in relation to a group of events. Looking at your responses, I now know why so many americans are stuck in sub prime mortagage hell. They simply do not grasp, and/or understand mathmatics and statistics.

[Fozzy4 0]

ONE YEAR! One year's worth of data in a year where the top scorer was so far above the rest of the pack.

 

Using 2003 data, (why not, it proves my theory); The difference between the number one player and the number ten player (using my leagues scoring) was 46 points. The difference from number eleven to number twenty was 69 points. So using only one year's data, I can say that the teams drafting at the beginning of the snake draft have only a slight advantage that it was debatable and easy correctible through team management through out the season.

 

Team 1 – 1075

Team 2 – 1080

Team 3 – 1068

Team 4 – 1058

Team 5 – 1054

Team 6 – 1055

Team 7 – 1061

Team 8 – 1070

Team 9 – 1066

Team 10 – 1064

So there is now a second year’s worth of data showing the advantage of drafting first is dependent on how much better the top few players are in any given year. One year (like last year), it might be a huge advantage. Another year, it might not be.

 

Since you seem to think 2003 is such a great year for snake draft, I will instead provide you with the results of the value totals for each team, as opposed to the total point production numbers that you did use. The primary difference is that values as calculated for VBD can be used to compare how much advantage a player gives you regardless of his position, where FFPts cannot do that. A player is not necessarily a better pick than a different one if he will make more fantasy points during the year, but a player with higher calculated value always is the better pick because he gives you a bigger advantage at his position than the other guy with lower value does at his position. One simple example of this from '03 (since I have the data set handy) is Chad Johnson versus Edgerrin James:

 

3. Chad Johnson CIN 195.5 96.6

10. Edgerrin James IND 220.9 78.9

 

You can see that CJ is the #3 receiver in '03, while Edge is the #10 RB. The first number is their FFPt totals. If you used that to judge, you would certainly pick Edge. However, the second number is their value, and represents how much they outplayed the last starter at their position by. From that, we can see that CJ outplayed the reference WR by 96.6 FFPts, while Edge only outplayed the reference RB by 78.9 FFPts. That means we should take CJ, because he gains the team a bigger advantage than Edge does, even though Edge produces more fantasy points during the year. CJ makes a bigger difference at WR than Edge does at RB, and so is the better pick because he piles up more advantage for your team.

 

Having covered that, note how entirely different the VBD totals come out than do the highest-production-first totals previously shown by joelmrut:

 

Team 1 444.8

Team 2 414.3

Team 3 407.1

Team 4 383

Team 5 387.5

Team 6 364.6

Team 7 352.7

Team 8 345

Team 9 342.7

Team 10 321.5

Team 11 322

Team 12 320.5

 

Spread 124.3

 

As with '06, we see a big discrepancy between the teams on opposite ends of the draft, and again the same trend of sinking value as you move further into the draft order. Also, note the direct correlation between the first rd of perfect draft in '03, and the spread of the top teams:

 

01. Priest Holmes KC 373 231

02. Ahman Green GB 345 203

03. LaDainian Tomlinson SD 339 197

04. Jamal Lewis BAL 311.1 169.1

05. Randy Moss MIN 267 168.1

06. Torry Holt STL 242.1 143.2

07. Clinton Portis DEN 274.5 132.5

08. Shaun Alexander SEA 269 127

09. Deuce McAllister NO 263.7 121.7

10. Chad Johnson CIN 195.5 96.6

11. Fred Taylor JAC 236.2 94.2

12. Anquan Boldin ARI 189.7 90.8

 

The ~30 pts that separate Holmes from Green are almost exactly what the two teams end up separated by in total value. Similarly, team 12's 140.2 pt drop-off is only a bit above the overall difference between the teams. In short, the trend seems to be that the advantages gained in rd 1 of a snake draft end up dictating the overall advantage differences between the teams. That implies that the remaining rounds function as intended: the changing directions more or less balance the advantages afforded to each team - it is almost entirely the first rd where the advantages that show up in the end are dictated. That's what I've speculated all along: that the first round drops off so much that it is responsible for a lot of the advantage inherent in snake drafting. As it turns out, the data indicates it is responsible for nearly all of it, with only very minor exceptions (like team 5 ending up slightly ahead of 4, which appears to be due to the almost non-existent value difference between their first round picks).

 

Given that finding, the question for '07 becomes "Are LT2, SJax, and LJ a big value advantage over Peyton Manning and/or Laurence Maroney, two typical guys you can get in 12th position?" Just about every source you'll look at says they are, and I think they are too, so the only conclusion I can make is that '07 is setup to be another year of snake draft unfairness. The FFT projections for instance have LT2 more than 90 value pts above either of those guys, so judging from that it's pretty reasonable to assume the value gap between 1st and 12th will end up around 90 pts. Considering that is more than 1/4 of the total value the later teams tend to accrue, it's a pretty hefty advantage we're talking about. In terms of FFPts, team 1s starters would outperform team 12's by an average of ~5.5 points per game. I don't think you could call it acceptable until you were below a 2 point per game discrepancy (32 value difference), and I'd really like to see it below 1 (for a 16 point value dif). Interestingly enough a 5% difference from 320 equals 16, so below 1 point per game is likewise the level at which the difference becomes statistically meaningless.

 

yes and no. in statistics we often speak of the possibility in relation to a group of events. Looking at your responses, I now know why so many americans are stuck in sub prime mortagage hell. They simply do not grasp, and/or understand mathmatics and statistics.

 

Sorry chief, you're wrong on this one. There aren't two ways to view it. It doesn't matter at all what the wheel did, nor how many times in a row it did it. Your odds to hit any given thing on the wheel are still exactly the same as always the next time it is used. Your contradiction to that is akin to saying you can flip 15 heads in a row, and have the next flip somehow affected by that. It isn't affected by that no matter how you look at it, so you are simply wrong to say it could be.

[KSB2424 3,148]

So if in 2007 it turns out that the bottom 3 draft picks had a statistcal advantage over the top 3 draft picks (which is entirely possible), are you gonna implement a "front filled" draft method to compensate in 2008?

[KSB2424 3,148]

Remember, you're talking to a chemical engineer and a professional poker player here.

Either you used to be a professional poker player and are now a chemical engineer, or your used to be a chemical engineer and are now a professional poker player. You can't be both.

 

From reading your posts, my guess is that you are a middle school chemistry teacher who plays online poker in the evenings.

[Mangatang 0]

Either you used to be a professional poker player and are now a chemical engineer, or your used to be a chemical engineer and are now a professional poker player. You can't be both.

 

From reading your posts, my guess is that you are a middle school chemistry teacher who plays online poker in the evenings.

 

Good point. I guess you would call me a part-time poker pro. I'm still a process controls engineer at an oil refinery (for 10 years). But, I also make an additional $15k-$20k per year playing online poker (for the past five years).

[KSB2424 3,148]

Good point. I guess you would call me a part-time poker pro. I'm still a process controls engineer at an oil refinery (for 10 years). But, I also make an additional $15k-$20k per year playing online poker (for the past five years).

Lemme guess. You also have a trophy wife, make 800K a year at your job, sport a 12 inch pecker, and can bench press 285 like you are allergic to it!

[Mangatang 0]

Lemme guess. You also have a trophy wife, make 800K a year at your job, sport a 12 inch pecker, and can bench press 285 like you are allergic to it!

 

No. Why would you say that? Do you know of any $800k/yr job openings? My resume is ready to go.

[KSB2424 3,148]

No. Why would you say that?

I'm just messing with you. This is a common joke on these boreds because a lot of people tend to over-exagerate how much money they make, how good at poker they are, how many fantasy titles they have won. ect. It's a running joke round these parts.

 

Now answer my question dagnabit!

 

So if in 2007 it turns out that the bottom 3 draft picks had a statistcal advantage over the top 3 draft picks (which is entirely possible), are you gonna implement a "front filled" draft method to compensate in 2008?

[Mangatang 0]

Now answer my question dagnabit!

 

 

QUOTE

So if in 2007 it turns out that the bottom 3 draft picks had a statistcal advantage over the top 3 draft picks (which is entirely possible), are you gonna implement a "front filled" draft method to compensate in 2008?

 

It is possible that after the 2007 stats are finished, that the snake method would benefit the late spots. For that to happen, there would need to be the top 11-12 player all have similar stats. Then there would need to be a steep decrease in talent in the 2nd round, then leveling off there after. Now, that type of results have never happened (that I remember seeing).

 

Every year, there is a very steep slope at the top of the player talent which levels off after the first round or so. We can only go by what history tells us.

[KSB2424 3,148]

It is possible that after the 2007 stats are finished, that the snake method would benefit the late spots. For that to happen, there would need to be the top 11-12 player all have similar stats. Then there would need to be a steep decrease in talent in the 2nd round, then leveling off there after. Now, that type of results have never happened (that I remember seeing).

 

Every year, there is a very steep slope at the top of the player talent which levels off after the first round or so. We can only go by what history tells us.

You didn't answer the question!

 

You tapped around it like a well trained politician. It is entirely possible that picks 1-10 in a 10 team league to be somewhat similar (i.e. small variance) and there a big drop off at 11,12 or 13. This IS possible. Unless of course you can predict the future.

[KSB2424 3,148]

It is possible that after the 2007 stats are finished, that the snake method would benefit the late spots. For that to happen, there would need to be the top 11-12 player all have similar stats. Then there would need to be a steep decrease in talent in the 2nd round, then leveling off there after. Now, that type of results have never happened (that I remember seeing).

 

Every year, there is a very steep slope at the top of the player talent which levels off after the first round or so. We can only go by what history tells us.

History told me at that roulette table it only lands on red as well.

[Mangatang 0]

History told me at that roulette table that it only lands on red as well.

 

Keep talking. You're bringing me good luck. Check out this hand I just got while reading your post.

 

http://www.kingscascade.com/StraightFlush.JPG

 

By the way, I got ShOrTy to commit all of his chips on that one.

[KSB2424 3,148]

Keep talking. You're bringing me good luck. Check out this hand I just got while reading your post.

 

http://www.kingscascade.com/StraightFlush.JPG

 

By the way, I got ShOrTy to commit all of his chips on that one.

Nice!

 

You better start making a move there "suicide king". You only got 10 times the blinds. Don't keep letting ngppa push you around.

[Mangatang 0]

Nice!

 

You better start making a move there "suicide king". You only got 10 times the blinds. Don't keep letting ngppa push you around.

 

Yeah, but after that hand, I had 30 big blinds. I'm still hacking away, it's still early.

[KSB2424 3,148]

Yeah, but after that hand, I had 30 big blinds. I'm still hacking away, it's still early.

A8 takes down pocket 4's.

I was rooting for you.

[Bert 1,128]

My god. I don’t know where to start.

 

Again, the actual outcome doesn’t justify making bad picks when you draft. If Michael Pittman ends up winning the rushing title this year, does that mean it was a dumb play not to draft him? NO! You know why? Because the odds of him having a year like that are astronomical. And anyone who bets against astronomical odds (without getting the proper payout in return) is crazy.

 

Just like in poker. I flopped top set on a rainbow board (Me = AcAh, Board = As9d6c). My opponent only has middle pair (9c2d). I call his all-in bet. He catches runner-runner 9’s (about 0.1% chance on the flop). Now, he beats me with quad 9’s, but does that make my call on the flop a bad play??????????

 

Now do you see what I mean when I say the ACTUAL outcome does not determine whether a play was correct or not? What matters is the estimated probability at the time that the decision is made.

 

My god. I don’t know where to finish.

My God I understand your point you are unable to grasp mine.

 

First let me start with what we trying to prove. We play fantasy football. We use football stats to score fantasy points. One person says the serpentine produces unfair results and the back fill produces more even point distribution. The other person says the variability of what an individual will produce is so great the bias created by the draft is eliminated.

 

Let me try to play off your analogy. Since you’re a poker pro you know that there are a fixed number of combinations of cards. You know what the exact odds are of your opponent catching 9 9 to beat you (0.1%). Because the card combinations are finite if you play an infinite number of hands you will win 99.9% of the time and lose 0.1% of the time. These odds are fixed, they are a given, they do not change. You are not projecting that the odds of getting a 9 9 is 0.1%. Because the probability is a given you play according to those odds. The problem with fantasy football is the odds aren't fixed. You use projected value to decide if a draft is fair. I prefer to use actual results to determine if a draft is fair since the probability that the projections are correct is virtually zero.

[zhinz 0]

I would say the problem with this analysis is that it relies on the VBD number. There are many different ways to calculate the VBD baseline, and each works well for some situations, and not so well for others.

 

Making the last starter as the baseline overrates the highest players, especially in deep leagues.

Making the last drafted player as the baseline overrates players at the positions with the least depth.

I like having the baseline be the average starter.

 

The best system would probably be one that used different baselines at different times. But no one's ever tried to figure out how to do that.

 

The point is that VBD is an approximation for value. At the extremes, it's not as correct. Personally, I think you'll know when your cheatsheet is in the right order when the #1 drafter's points are as close as possible to the #12 drafter.

 

In other words, I don't think your numbers say anything about one spot being better than the other. Instead, they prove that VBD isn't accurate enough.

[Fozzy4 0]

My God I understand your point you are unable to grasp mine.

 

First let me start with what we trying to prove. We play fantasy football. We use football stats to score fantasy points. One person says the serpentine produces unfair results and the back fill produces more even point distribution. The other person says the variability of what an individual will produce is so great the bias created by the draft is eliminated.

 

Okay, this is a topic I tried to cover before, but admittedly might not have done a good job of. This time I'll try by explaining the concepts and why variance can't remove bias.

 

In order for the bias to be destroyed by variability, there must be absolutely no basis at all for creating projections - you must accept that projections have no correlation at all to actual performance, and therefore that there is no projection in existence that can help you draft better than anyone else.

 

If there are such projections that can help you draft better, though, then the extent that they generally correlate to the actual data is likewise the extent to which my projected bias will correlate to actual bias since the calculations for this season are entirely based on the projections. That is, variance from projections to performance creates similar variance in bias, but cannot simply cancel it, nor change the average amount of that bias. We can all easily tell that some projections can help you draft better than you would without any projections, so likewise we know that my math correlates to actual bias to some degree. From there we could graph the variances from given projections to the corresponding actual production of the player. Adjusting those variances across the number of starters (9 is standard) would then yield a data set that gives average variance for a whole team's production from its projection (by extending an average for a player across 9 players). Perhaps you see where this is going? You could then use that team variance to project a range of values for the team's actual production, and use maybe 2nd standard deviations on the low side for team 1 and 2 standard deviations on the high side for team 12 to determine the smallest gap you're likely to see between them.

 

However, that's not all. The projections will remain the average output of the team unless there is a significant trend for projections to be higher or lower than actual performance across the board. My experience has been that good projections like FFT's tend to be about in the middle, being about equally prone to over- and under-shoots. That would mean that the variance allows for some different outcomes, but does not change the average (expected) outcome. The long and short is what I said before: variance can make it more interesting and allow disadvantaged teams to win sometimes, but it can't actually change the expected outcome. The expected outcome maintains the same spread as it did before, meaning even with variance accounted for, we expect to see the same amount of bias on average that we would have if projections simply were equal to production. It may be higher a given time, it may be lower, but the average amount for a league will be precisely what I calculated.

 

The short version: variance can't change the average outcome, so doesn't affect the expected outcome, and therefore doesn't affect bias or fairness at all. My calculated difference between team 1 and 12 (and all the others that I didn't explicitly calculate) will remain the average outcome regardless of how much variance there is, which means we can under any level of variance expect bias of exactly that much.

 

Now what if there is a significant trend for projections to be either higher or lower across the board than actual performance? Ahh, there's the beauty: again, it won't matter. It will mean that the average output for each team shifts up or down by a given amount, but that amount is the same for each team as the projection trend is equally likely to affect each player. Thus the entire data set would shift a given amount to account for the trend of high or low projections, but again the differences I calculated would remain as the average (and thus expected) outcome.

 

As you should now see, as long as the projections correlate to the actual data somehow, variance is measured with regard to them, and being equally likely to affect any given player (and thus logically any given team) it affects the average outcome not at all. Sorry, but that's just the nature of variance. The only way it could actually alter the average mathematical advantage I calculated is if the projections didn't correlate to performances even remotely, and I think we can all agree that they do correlate some. They do so to varying degrees to be sure, but all reasonable projections have some relation to the actual performance data, and thus maintain the average bias I calculated before. The only difference is that instead of just having a hard number, we now have an average result of that number, with results dispersed around it. Anyway, point is that a biased expected outcome equals a biased system, and since no source of variance affects the expectation at all it can't alter the fact that snake drafting is biased.

[KSB2424 3,148]

Why are ya'll dodging my questions?

[KSB2424 3,148]

Okay, this is a topic I tried to cover before, but admittedly might not have done a good job of. This time I'll try by explaining the concepts and why variance can't remove bias.

 

In order for the bias to be destroyed by variability, there must be absolutely no basis at all for creating projections - you must accept that projections have no correlation at all to actual performance, and therefore that there is no projection in existence that can help you draft better than anyone else.

 

If there are such projections that can help you draft better, though, then the extent that they generally correlate to the actual data is likewise the extent to which my projected bias will correlate to actual bias since the calculations for this season are entirely based on the projections. That is, variance from projections to performance creates similar variance in bias, but cannot simply cancel it, nor change the average amount of that bias. We can all easily tell that some projections can help you draft better than you would without any projections, so likewise we know that my math correlates to actual bias to some degree. From there we could graph the variances from given projections to the corresponding actual production of the player. Adjusting those variances across the number of starters (9 is standard) would then yield a data set that gives average variance for a whole team's production from its projection (by extending an average for a player across 9 players). Perhaps you see where this is going? You could then use that team variance to project a range of values for the team's actual production, and use maybe 2nd standard deviations on the low side for team 1 and 2 standard deviations on the high side for team 12 to determine the smallest gap you're likely to see between them.

 

However, that's not all. The projections will remain the average output of the team unless there is a significant trend for projections to be higher or lower than actual performance across the board. My experience has been that good projections like FFT's tend to be about in the middle, being about equally prone to over- and under-shoots. That would mean that the variance allows for some different outcomes, but does not change the average (expected) outcome. The long and short is what I said before: variance can make it more interesting and allow disadvantaged teams to win sometimes, but it can't actually change the expected outcome. The expected outcome maintains the same spread as it did before, meaning even with variance accounted for, we expect to see the same amount of bias on average that we would have if projections simply were equal to production. It may be higher a given time, it may be lower, but the average amount for a league will be precisely what I calculated.

 

The short version: variance can't change the average outcome, so doesn't affect the expected outcome, and therefore doesn't affect bias or fairness at all. My calculated difference between team 1 and 12 (and all the others that I didn't explicitly calculate) will remain the average outcome regardless of how much variance there is, which means we can under any level of variance expect bias of exactly that much.

 

Now what if there is a significant trend for projections to be either higher or lower across the board than actual performance? Ahh, there's the beauty: again, it won't matter. It will mean that the average output for each team shifts up or down by a given amount, but that amount is the same for each team as the projection trend is equally likely to affect each player. Thus the entire data set would shift a given amount to account for the trend of high or low projections, but again the differences I calculated would remain as the average (and thus expected) outcome.

 

As you should now see, as long as the projections correlate to the actual data somehow, variance is measured with regard to them, and being equally likely to affect any given player (and thus logically any given team) it affects the average outcome not at all. Sorry, but that's just the nature of variance. The only way it could actually alter the average mathematical advantage I calculated is if the projections didn't correlate to performances even remotely, and I think we can all agree that they do correlate some. They do so to varying degrees to be sure, but all reasonable projections have some relation to the actual performance data, and thus maintain the average bias I calculated before. The only difference is that instead of just having a hard number, we now have an average result of that number, with results dispersed around it. Anyway, point is that a biased expected outcome equals a biased system, and since no source of variance affects the expectation at all it can't alter the fact that snake drafting is biased.

If there was ever a good use for the saying: "You can't see the forest for the trees" this would be it.

[Mangatang 0]

Why are ya'll dodging my questions?

 

This question?

 

So if in 2007 it turns out that the bottom 3 draft picks had a statistcal advantage over the top 3 draft picks (which is entirely possible), are you gonna implement a "front filled" draft method to compensate in 2008?

 

I thought I did answer it with:

 

It is possible that after the 2007 stats are finished, that the snake method would benefit the late spots. For that to happen, there would need to be the top 11-12 player all have similar stats. Then there would need to be a steep decrease in talent in the 2nd round, then leveling off there after. Now, that type of results have never happened (that I remember seeing).

 

I said there that it IS possible that the last few draft spots could have an advantage in the snake format, IF the top 11-12 players all have the same production. I don't ever remember seeing that happen though. Every year there is a steep slope of value at the top of the player list. I am assuming that this will hold true from year to year.

[Mangatang 0]

A8 takes down pocket 4's.

I was rooting for you.

 

Thanks. Yeah, I was just below 10 big blinds, and I got all-in with a coin flip. That's about the best I could hope for. Do you play on Stars?

[KSB2424 3,148]

I am assuming that this will hold true from year to year.

Exactly

 

You are assuming that through predictions/projections (ADP) or data based on the past few seasons (a small sample) that you can predict the future which in and of itself is a fallacy. Because your theory is based on a fallacy, all of these numbers and reasoning both you and fozzy are spitting out are completely flawed.

 

You do know what they say about people who ASSUME don't you?

[Mangatang 0]

Let me try to play off your analogy. Since you’re a poker pro you know that there are a fixed number of combinations of cards. You know what the exact odds are of your opponent catching 9 9 to beat you (0.1%). Because the card combinations are finite if you play an infinite number of hands you will win 99.9% of the time and lose 0.1% of the time. These odds are fixed, they are a given, they do not change. You are not projecting that the odds of getting a 9 9 is 0.1%. Because the probability is a given you play according to those odds. The problem with fantasy football is the odds aren't fixed. You use projected value to decide if a draft is fair. I prefer to use actual results to determine if a draft is fair since the probability that the projections are correct is virtually zero.

 

The poker example I gave was one where you have complete information (you knew what the other guy held because I told you). Most poker decisions, you don't know for sure what the other guy has, so you have to estimate as best you can what he might have and compare those possibilities with the odds of those situations. Let's say I estimate that there is a 50% chance that my opponent is behind with top pair, in which case I am a 4 to 1 favorite. I estimate a 20% chance that he flopped a set and is ahead of me, in which case I am a 5 to 1 dog. Then there is always a 5-10% chance that he is flat out bluffing. Etc, etc.

 

This is very similar to making fantasy football projections. I can use all the information I have (last year's results, scouting reports, injury reports, etc.) to produce a best-guess estimation of a player's production. This estimation HAS value. This is better than just randomly picking any player from a 200 player pool. The estimation, although not perfect, has merit.

 

Again, the best way to look at it is to use auction values. Would anyone argue that FFToday's average auction values for players this year are completely worthless? If not, then they must be a pretty good estimation of the player's values this year in a completely fair auction. Now apply those auction values to the snake draft, and see what each draft spot ends up with (like I showed in a previous post). It clearly shows that the snake format gets a much bigger advantage than the backfill method.

 

This, in no way, means that the first draft pick is guaranteed to win the 2007 season. It just means that he has a better chance of fielding a good team than the later picks. That's the definition of an unfair situation.

[Mangatang 0]

Exactly

 

You are assuming that through predictions/projections (ADP) or data based on the past few seasons (a small sample) that you can predict the future which in and of itself is a fallacy. Because your theory is based on a fallacy, all of these numbers and reasoning both you and fozzy are spitting out are completely flawed.

 

You do know what they say about people who ASSUME don't you?

 

No, I am assuming that it will happen because it has happened every year for the past 10 years.

[KSB2424 3,148]

No, I am assuming that it will happen because it has happened every year for the past 10 years.

So you have done the analysis for the last 10 years? Let's try to be honest now. And even if it did....SO?

 

That is where I was trying to go with the coin flip / roulette wheel analogy. Even if an event happend 10 times in a row (i.e. a coin landing on heads) it has no bearing on the the 11th coin flip because it is a seperate event.

[Mangatang 0]

Why are ya'll dodging my questions?

 

Now, please answer my question that you have dodged all along. Do you think the FFToday average auction values are a good estimation of palyer values for 2007?

[Mangatang 0]

So you have done the analysis for the last 10 years? Let's try to be honest now. And even if it did....SO?

 

That is where I was trying to go with the coin flip / roulette wheel analogy. Even if an event happend 10 times in a row (i.e. a coin landing on heads) it has no bearing on the the 11th coin flip because it is a seperate event.

 

Granted, the sample size of 10 years may not be statistically significant from an acedemic standpoint, but if we show that the snake format has been unfair for the past ten years, don't you think it's worth considering a new draft format? From the very first post, you flat out said there is no better draft format than the snake. It's almost like you are putting your hands over your ears and screaming (like a child does) so that you don't hear any arguement that might hold truth. You are clinging to your initial instinct, but not backing it up with any thing other than number 1 and 24 averages out to the same as 12 and 13.

[KSB2424 3,148]

Now, you answer my question that you have dodged all along. Do you think the FFToday average auction values are a good representation of palyer values for 2007?

I like mine better, but they are just as good as any.

 

I think I know where you are going with this and let me draw the distinction. Average auction values are based on predictions/projections. This is all we have to go on to DRAFT players and these can vary from site to site, from person to person, as they are a guess. This is what we use to try and field the best team we can with limited information. However we should not assume that these predictions will be so true that we go around and modify a base snake method to "fit" these predictions. That is the difference.

[KSB2424 3,148]

Granted, the sample size of 10 years may not be statistically significant from an acedemic standpoint, but if we show that the snake format has been unfair for the past ten years, don't you think it's worth considering a new draft format? From the very first post, you flat out said there is no better draft format than the snake. It's almost like you are putting your hands over your ears and screaming (like a child does) so that you don't hear any arguement that might hold truth. You are clinging to your initial instinct, but not backing it up with any thing other than number 1 and 24 averages out to the same as 12 and 13.

I've never said the "back filled draft" has absolutely no merit, and I have read every post. Both you and fozzy are implying that based on a few years past data you assume the same result will happen. Which it very well may, I do not deny that. However it very well may NOT happen. That is why I do not get in the business of trying to assume and/or predict anything. I am not Ms.Cleo. Now if you and fozzy are so clairvoyant, could you please let me know the lotto numbers for the Mega Million next week?

[mikekoop 0]

Because a snake draft isn't as fair as other options. Here's the total VBD (Value Based Drafting because it allows you to compare different positions to each other) points for each draft position after 8 rounds of drafting, using conserative player projections and my league's scoring rules.

 

1. 394

2. 375

3. 345

4. 344

5. 328

6. 313

7. 296

8. 294

9. 296

10. 297

11. 295

12. 291

 

Having the #1 pick and then being able to pick two players in a row at 24/25 is too much for the 12th guy to overcome.

 

If you use the "Backfill" method, which is first round draft 1-12, then every round after that draft 12-1, you get a more even talent distribution among the teams. The 12th guy gets picks 12 and 13 as usual, but then in the third round he gets the first pick (instead of the guy with LT). It works out much better.

 

1. 361

2. 344

3. 317

4. 325

5. 318

6. 309

7. 301

8. 304

9. 315

10. 325

11. 326

12. 325

 

As you can see the differential in the snake method is 103 points, where the backfill method is only 60 points. It's still weighted in favor of the early drafters, but not near as much as the snake method.

 

Of course, auction is the only completly fair way to draft, but the backfill method is better than traditional snake and it's easy to remember the order (unlike some of these methods that switch the order every round).

[Mangatang 0]

I've never said the "back filled draft" has absolutely no merit, and I have read every post. Both you and fozzy are implying that based on a few years past data you assume the same result will happen. Which it very well may, I do not deny that. However it very well may NOT happen. That is why I do not get in the business of trying to assume and/or predict anything. I am not Ms.Cleo. Now if you and fozzy are so clairvoyant, could you please let me know the lotto numbers for the Mega Million next week?

 

4 8 15 16 23 42

[Bert 1,128]

I've never said the "back filled draft" has absolutely no merit, and I have read every post. Both you and fozzy are implying that based on a few years past data you assume the same result will happen. Which it very well may, I do not deny that. However it very well may NOT happen. That is why I do not get in the business of trying to assume and/or predict anything. I am not Ms.Cleo. Now if you and fozzy are so clairvoyant, could you please let me know the lotto numbers for the Mega Million next week?

 

KSB2424 I'm with ya but I can't debate anymore with "the actual results don't matter". But I will test their back fill draft against the serpentine this year using ADPs from right before the start of the season. I'll draft 12 teams using a serpentine and a backfill and I will add up the results at the end of the year. We will see what happens over the next several years.

 

Looking at the registration dates I know you and I will be around we will see if they stick around.

[Jimmy 0]

It is possible that after the 2007 stats are finished, that the snake method would benefit the late spots. For that to happen, there would need to be the top 11-12 player all have similar stats. Then there would need to be a steep decrease in talent in the 2nd round, then leveling off there after. Now, that type of results have never happened (that I remember seeing).

 

Every year, there is a very steep slope at the top of the player talent which levels off after the first round or so. We can only go by what history tells us.

 

Exactly (almost). I'd say that there'd need to be 15 in a 12-team draft to create a significant statistical advantage for the bottom three draft slots. Which I think makes your point even stronger.

 

 

So if in 2007 it turns out that the bottom 3 draft picks had a statistcal advantage over the top 3 draft picks (which is entirely possible), are you gonna implement a "front filled" draft method to compensate in 2008?

 

From a theoretical standpoint you're 100% correct. But, that's the thing, we can use educated guessing and a whole lot of projections to show us how positively remote the chance of this is. It's not like we're all blindly following the stats here. I think that the important thing to take from all this is this: Snake drafts give an advantage to every player that picks above the "talent plateau". While we obviously can't predict next season with 100% accuracy, history has shown us some pretty darn strong trends we can follow and educated projections support those trends. For the Snake draft to be fair, there needs to be an equal amount of top-talent to the number of teams drafting. To tip the balance to the late-pickers, the "talent plateau" actually has to wrap around into the second round. I think you're far down the road of devil's advocate and don't believe there's much likelihood of that happening..?

 

I've not ever put a name to it before, but I've been paying attention to this "talent plateau" for a while now. Five'ish seasons ago, I would have rather picked around 5th than 1st. The reason being is that the talent was top-heavy and "high-quality" (e.g. Faulk, Priest, Ahman Green, SA, CuMart, Dillon, etc) and it was somewhat of a crapshoot to figure out which would separate themselves - so, I'd just as soon let the first four or five people thin the herd, I'd take the best remaining, and then have a slight advantage in round two. In hindsight we realized that Faulk stood head-and-shoulders above his peers one season and Priest did the next, but at draft time, there was a lot of debate between four or five guys being the true #1. This season is a LOT different in that LT2 is a bar-none consensus #1, and while SJax hits #2 a lot (that just sounds wrong lol), that's all there is at the top. No matter how good or bad Gore, Addai, SA, LJ end up being this season, walking away with one of them at #3 this year doesn't give you the same warm-fuzzy that Faulk/Priest/Green/SA did back in the day. I think we just happen to be in a transition point where a lot of the established talent is phasing out and/or has a lot of question marks and the guys that are eventually going to replace them (fantasy-wise) haven't stepped all the way up yet.

 

I don't think this means you should rearrange your league's draft method every season based on projections. But, I do think it's worthy of notice and understanding. Any time you can put a name/explanation to a vague sense/feeling (e.g. VBD 10 years ago), you come out of it with a little better knowledge and control of the game.

[flushmonkii 0]

KSB, you'll have no problem at all if you were cursed with always having the 11th and 12th spot every year for the rest of your life in every regular serpentine draft? Honestly? No problem at all? Knowing that every year you'll never have a chance at the LTs and LJs of the world but will gladly take the Rudi Johnson's and Ronnie Brown's every year?

[mikekoop 0]

Our 12 Team league is in its 11th year. We do a snake with a twist. Round 1, everybody rolls dice and hi roller drafts #1 - we then go clockwise to #12 and "snake" back to #1 - just like everybody. We then reroll the dice every odd number round (3,5,7,9,11, etc...) to see who goes 1st every odd round, we then snake back every even round - then roll again. This avoids the #1 and #12 guys in a normal serpentine draft from always having back to back picks every other round. This seems to create a lot of parity and we have 3 other leagues in our city doing the same thing.

[Fozzy4 0]

KSB, you'll have no problem at all if you were cursed with always having the 11th and 12th spot every year for the rest of your life in every regular serpentine draft? Honestly? No problem at all? Knowing that every year you'll never have a chance at the LTs and LJs of the world but will gladly take the Rudi Johnson's and Ronnie Brown's every year?

 

Such a simple point, yet it says everything. We all already know just from our playing experiences that 12th pick isn't as good as 1st, but for some reason several guys thinks they can argue their way around it. If it looks like a duck, and quacks like a duck, it's probably a duck. All signs point to snake draft being unfair, so that should be our working theory, as opposed to that it is fair. As a result it is required that someone show evidence it is fair in order for that possibility to be considered.

 

So given the mountain of evidence demonstrating its bias, I demand some actual evidence that it is not biased, or I simply will not continue to argue the point. There is no basis for arguing it is fair unless someone provides that basis first, and I will not feel obligated to address any further points until someone does so. Failure to do so simply means your points have no statistical merit, and as I said I will not continue defending against accusations and viewpoints that have no merit and no basis. Find something real to use, or give it up.

[KSB2424 3,148]

KSB, you'll have no problem at all if you were cursed with always having the 11th and 12th spot every year for the rest of your life in every regular serpentine draft? Honestly? No problem at all? Knowing that every year you'll never have a chance at the LTs and LJs of the world but will gladly take the Rudi Johnson's and Ronnie Brown's every year?

I'll put it to you this way. Before every draft whether I have the first pick, the 6th pick or the 12th pick, I feel I have just as good a shot to win and have done so. In some years I prefer having the 12th pick based on my cheatsheet. In other years I prefer to have the 6th pick based on my cheatsheet, and like this year I would prefer to have the 1rst pick.

 

However I wouldn't dare modify a league draft order based these assumptions and predictions.

 

When you assume - It makes an ass out of u and me.