Playerlink Projected Balance....

[Raine05]

I just want to make sure I got this right....

the Proj. Bal for a player is what he is expected/projected to do the rest of the way, so this is what I should look at to make future decisions(for the rest of this year) on things like a possible trade, or the value of 1 player over another, etc. So, if Player A has better YTD stats and/or better Proj Total stats but the Proj Bal of Player B is better = I should go with Player B, because he will (projected too, I should say) do better from this point on?

Also, is this just a some kind of subtraction of YTD stats from the original Proj for a player, or do you take other considerations?

Thanks.

[RAY@HQ]

Yes, you have the concept of Projected Balance correct.

I'm not sure I understand the second question.

[Tom_T]

My take on his second question: do we just subtract YTD stats from the original projection to get the balance-of-year numbers? No, because the projections change.

[Raine05]

Thanks Ray and Tom (yes, that was what I meant ....glad the answer was NO).

One more question (well, for this thread anyways), I remember reading in the beginning of the year that the Projections were at best around 70% to be accurate, I assume that the Proj. Bal stats have to be higher then that, based on the amount of games already played, and that the closer and closer we get to the end of the season, the stronger and stronger (closer to 100%) the Proj. Bal will be, correct?

Sorry, if I might be stating the obvious, but just wanted to make sure.

Thanks.

[Ed@HQ]

Thanks Ray and Tom (yes, that was what I meant ....glad the answer was NO).

One more question (well, for this thread anyways), I remember reading in the beginning of the year that the Projections were at best around 70% to be accurate, I assume that the Proj. Bal stats have to be higher then that, based on the amount of games already played, and that the closer and closer we get to the end of the season, the stronger and stronger (closer to 100%) the Proj. Bal will be, correct?

Sorry, if I might be stating the obvious, but just wanted to make sure.

Thanks.

Raine, I don't think it's ever been measured in the same way, but I would actually assume the OPPOSITE to be true. As the projected time frame becomes smaller and smaller, the accuracy will be lower and lower. Even though the balance of year projections at this point are more immediate than the pre-season projections (trying to only project 2 or 3 months into the future rather than 6 months), as the time frame in question becomes smaller the expected fluctuation in performance simply overpowers the expected level of performance, so much so that HQ doesn't even offer projections for the last two weeks of the year when the season reaches that point.

Hope that helps.

-Ed

[RAY@HQ]

I'm with Ed on that point.

[Raine05]

I will re-read and try to think about your answer tomorrow again (it's been a long day) but I am a little confused. My 1st thought would have been if Player A averages 3 basehits a week after the 1st 5 months of the season, and nothing changes like playing time, injury, etc., it would be easier to predict that Player will finish with 12 hits in the last month to go. Compared to Player B averaging 3 hits a week after the 1st month, because of the short period of time, he could really finish say with 5 hits a week (avg) or maybe 2. This is why I thought the Proj Bal would be more accurate the further into the season.

Thanks.

[RAY@HQ]

The problem is that if you take a look at a player's weekly or monthly stats, they aren't that linear. They don't get 3 hits a week... they get 8, 1, 4, 7, 3 in a 5-week stretch. Even months aren't really much better in many cases. Take a look at Teixeira's HR totals by month: 3-13-4-1. Projecting the next number in that series is going to be imprecise at best.

[Ed@HQ]

My 1st thought would have been if Player A averages 3 basehits a week after the 1st 5 months of the season, and nothing changes like playing time, injury, etc., it would be easier to predict that Player will finish with 12 hits in the last month to go.

I agree that it might be appropriate to expect Player A to finish with 12 hits in the last month, but to predict with any precision that of Player A's next 40 or so trips to the plate, he will strike the ball and have it that land somewhere in fair territory or over the fence (a.k.a. "hits") is very difficult. Three great defensive plays bunched together and you're suddenly 25% off your prediction.

Compared to Player B averaging 3 hits a week after the 1st month, because of the short period of time, he could really finish say with 5 hits a week (avg) or maybe 2.

Yes, but we likely know a lot more about Player B than what he did in that first month, such as what he did last year or the five years preceding it.

This is why I thought the Proj Bal would be more accurate the further into the season.

Your thinking is correct as it relates to expectation; again it's the fluctuation around that expectation that becomes unpredictable over shorter time frames. Assuming that we do in fact have a larger baseline from which to work (vs. only knowledge of the current season) when setting our expectation, though, then we should be able to predict with greater accuracy what will happen over 5 months than 1 month, because over 5 months the fluctuations can and often do even out, whereas over 1 month we could catch a high or low tail (as Ray noted with the Teixeira example) and miss badly with our prediction.

Thanks for asking the question. I'm sure that there are others on the boards thinking the same thing ...

-Ed

[makelele]

I agree that it might be appropriate to expect Player A to finish with 12 hits in the last month, but to predict with any precision that of Player A's next 40 or so trips to the plate, he will strike the ball and have it that land somewhere in fair territory or over the fence (a.k.a. "hits") is very difficult. Three great defensive plays bunched together and you're suddenly 25% off your prediction.


Yes, but we likely know a lot more about Player B than what he did in that first month, such as what he did last year or the five years preceding it.


Your thinking is correct as it relates to expectation; again it's the fluctuation around that expectation that becomes unpredictable over shorter time frames. Assuming that we do in fact have a larger baseline from which to work (vs. only knowledge of the current season) when setting our expectation, though, then we should be able to predict with greater accuracy what will happen over 5 months than 1 month, because over 5 months the fluctuations can and often do even out, whereas over 1 month we could catch a high or low tail (as Ray noted with the Teixeira example) and miss badly with our prediction.

Thanks for asking the question. I'm sure that there are others on the boards thinking the same thing ...

-Ed

This is a question that could be easily answered with data instead of speculation. Simply check previous seasons' in-season projections and compare them to actual performance. I can see where the OP is coming from - just because you have a larger body of data to draw on doesn't make it more meaningful. For example, when September rolls around, a player's performance may be better predicted by his last 5 months rather than his last 5 years.

One way I feel I can surpass the magical "70%" threshold is to use RotoLab to keep track of performance in-season. When I look at league totals and standings this time of year, my confidence in the balance-of-season projected totals improves as we get closer and closer to the end. That is also a relevant question that fantasy managers should make sure to answer for themselves as we head down the back stretch.

[Raine05]

Thanks for the responses.

Ed - "it's the fluctuation around that expectation that becomes unpredictable over shorter time frames"...............I understand this point now (Ray, your exapmle of stats not being linear also helped). By the way how do you copy quotes in a white box like you guys do, yeah I am not a big poster (I hit the small check box marked "Quote message in reply?", but nothing happens) ?

Anyway, after reading the responses it comes back to my 1st question - should I be strictly looking at the Proj Bal to determine which players to start or go after, etc.? And I guess the answer should really be "no" (because the fluctuation around that expectation becomes more unpredictable over shorter time frames). But possibly use them in combination with YTD and the trend of Last 31 and Last 7 days (which for beginners like me, is another can of worms)?

Thanks.

[RAY@HQ]

Yes, that last approach (weighing BoY projection against recent performance, although last 7 days won't tell you much) is the way to go.

If you hit the "quote" button, it should take the post you want to quote and put it in the Reply box wrapped in quote tags... then you just respond below that section.

[Ed@HQ]

This is a question that could be easily answered with data instead of speculation. Simply check previous seasons' in-season projections and compare them to actual performance. I can see where the OP is coming from - just because you have a larger body of data to draw on doesn't make it more meaningful. For example, when September rolls around, a player's performance may be better predicted by his last 5 months rather than his last 5 years.

One way I feel I can surpass the magical "70%" threshold is to use RotoLab to keep track of performance in-season. When I look at league totals and standings this time of year, my confidence in the balance-of-season projected totals improves as we get closer and closer to the end. That is also a relevant question that fantasy managers should make sure to answer for themselves as we head down the back stretch.

I hear what you're saying, and I agree that our general expectation of balance-of-year performance will become clearer as the season rolls on (because the YTD data is both relevant and recent). But it's not quite speculation to say that our predictive ability is diminished as the time frame of what we're trying to predict becomes too small. To your point, maybe we can still be accurate over 2-3 months (and yes perhaps even more accurate, just as we may be more accurate when predicting one year instead of three years even though it is a smaller window), but when we get down to that last month, or that last week, or that last game, we do reach a point of just not giving the natural fluctuations a chance to even themselves out.

I can't think of the link right now (maybe from The Book), but there have been studies that show how many plate appearances it takes for certain statistics to stabilize. If I recall correctly, most are in the hundreds of PAs, so to predict anything smaller than that with reliable accuracy may just not be realistic.

[makelele]

I hear what you're saying, and I agree that our general expectation of balance-of-year performance will become clearer as the season rolls on (because the YTD data is both relevant and recent). But it's not quite speculation to say that our predictive ability is diminished as the time frame of what we're trying to predict becomes too small. To your point, maybe we can still be accurate over 2-3 months (and yes perhaps even more accurate, just as we may be more accurate when predicting one year instead of three years even though it is a smaller window), but when we get down to that last month, or that last week, or that last game, we do reach a point of just not giving the natural fluctuations a chance to even themselves out.

I can't think of the link right now (maybe from The Book), but there have been studies that show how many plate appearances it takes for certain statistics to stabilize. If I recall correctly, most are in the hundreds of PAs, so to predict anything smaller than that with reliable accuracy may just not be realistic.

I agree with that. It's hard to be confident in projections for a time frame smaller than 1-2 months. But if we have to make those types of projections (obviously we do, to some extent, in season), it would be nice to have the data on how much weight to give to in-season vs. prior season performance.

[Raine05]

[QUOTE=RAY@HQ;486700]Yes, that last approach (weighing BoY projection against recent performance, although last 7 days won't tell you much) is the way to go.


Sorry, but what does BoY stand for? Thanks.