When a successful player at Fantasy Sports Live dishes out knowledge, you should definitely pay attention. Buffalo66 did just that earlier this week:
The area where I think we probably disagree a bit is on the utility of minor roster variations. Not to put any words in Buffalo66's mouth, but I think he's generally against roster variation, period. Use your best combination, all the time.
Now I'll admit it: I vary my rosters. I don't think I reach the level of "wide" variation (usually and with one caveat, which I'll get to in another post), but I have been known to vary on a limited basis as a hedge against variance.
Let's say, for example, that you have done your homework for a week's games and you have come up with four receivers, all of whom are pretty close in expected production. The logical thing to do is to pick the best three and go with them. But - and again, this requires a close case - I believe that you can reduce your overall variance by varying your picks and effectively using all four players.
Assume that you have graded out Randy Moss and Wes Welker similarly. You figure each is most likely going to score something in the range of 10 points. But you know that they're totally different players - Welker will get his points by getting his usual six to ten catches and no touchdowns. Moss will have fewer catches but is much more likely to score. Then again, he's also much more likely to put up a zero. If you graphed expected results, Welker's would be tightly bunched around 10, while Moss would be much flatter:
So who do you choose?
If you play Welker in all your contests, you're taking the lower variance option, obviously. You can predict with a much higher level of certainty what your team score is going to be. However, if it's a wild scoring week in the NFL, that may not be enough to propel you to the cash. If you play Moss in all your contests, you have no idea what you're going to get. You're taking a chance - if Randy has a big week, you may get a clean sweep, if he puts up a near zero you may get swept when even a modest point total would have made a difference. You can project results all you want, but it is an imperfect science because of the vagaries of human performance. If you saw Moss play last week you know where I'm coming from.
So, anyway, the theory is that by - in this example - taking Welker in some contests and Moss in others, you can reduce your overall variance somewhat. Can I prove this? Nope. I've started messing around with the statistics, but I'm not sure that I can eliminate enough variables and simplify things enough to actually figure it out. I should also mention that I think the utility of variance differs between heads up, six player and ten player contests but I haven't gotten anywhere with that math either. Stupid math.
I'll typically use this type of player variation only on marginal decisions and, honestly, I don't have the level of discipline in making selections that Buffalo66 or kaiseroll13 do. It's a leak.
So, in conclusion, you might want to take the fantasy pro's words to heart. Go read them again. As I see it, roster variance is like playing with gasoline around a fire. Sometimes you'll get a warm glow, but you're much more likely to get totally burned.
2) Is it a viable strategy to shake up your picks and vary your rosters across the board? I know others disagree with with me, but I say a resounding "NO".If you ask me, I agree with about 99% of this. Widely varying rosters across different contests is definitely -EV over the long run. Even in the short run - which is what a football season or a week of contests is, really, it's difficult to make consistent +EV variations and his point is well taken - if you have a best combination, how can you expect to come out ahead with different players? If you think you can, you should be playing those guys instead.
Widely varying your rosters is a sure fire method to lose money over time. While you are probably guaranteed to win a few, you will never win them all. You are "playing not to lose" instead of "playing to win".
If you practice good bankroll management and can withstand a big losing day, then there is nothing wrong with taking your best guess to score points and playing that roster across the board.
If you follow my picks, you notice I never change my rosters much (if at all). I will tell you Kaiseroll13 does the same. If the top two overall players at FSL use this strategy, you would think there is something to it. Just food for thought.
The area where I think we probably disagree a bit is on the utility of minor roster variations. Not to put any words in Buffalo66's mouth, but I think he's generally against roster variation, period. Use your best combination, all the time.
Now I'll admit it: I vary my rosters. I don't think I reach the level of "wide" variation (usually and with one caveat, which I'll get to in another post), but I have been known to vary on a limited basis as a hedge against variance.
Let's say, for example, that you have done your homework for a week's games and you have come up with four receivers, all of whom are pretty close in expected production. The logical thing to do is to pick the best three and go with them. But - and again, this requires a close case - I believe that you can reduce your overall variance by varying your picks and effectively using all four players.
Assume that you have graded out Randy Moss and Wes Welker similarly. You figure each is most likely going to score something in the range of 10 points. But you know that they're totally different players - Welker will get his points by getting his usual six to ten catches and no touchdowns. Moss will have fewer catches but is much more likely to score. Then again, he's also much more likely to put up a zero. If you graphed expected results, Welker's would be tightly bunched around 10, while Moss would be much flatter:
So who do you choose?If you play Welker in all your contests, you're taking the lower variance option, obviously. You can predict with a much higher level of certainty what your team score is going to be. However, if it's a wild scoring week in the NFL, that may not be enough to propel you to the cash. If you play Moss in all your contests, you have no idea what you're going to get. You're taking a chance - if Randy has a big week, you may get a clean sweep, if he puts up a near zero you may get swept when even a modest point total would have made a difference. You can project results all you want, but it is an imperfect science because of the vagaries of human performance. If you saw Moss play last week you know where I'm coming from.
So, anyway, the theory is that by - in this example - taking Welker in some contests and Moss in others, you can reduce your overall variance somewhat. Can I prove this? Nope. I've started messing around with the statistics, but I'm not sure that I can eliminate enough variables and simplify things enough to actually figure it out. I should also mention that I think the utility of variance differs between heads up, six player and ten player contests but I haven't gotten anywhere with that math either. Stupid math.
I'll typically use this type of player variation only on marginal decisions and, honestly, I don't have the level of discipline in making selections that Buffalo66 or kaiseroll13 do. It's a leak.
So, in conclusion, you might want to take the fantasy pro's words to heart. Go read them again. As I see it, roster variance is like playing with gasoline around a fire. Sometimes you'll get a warm glow, but you're much more likely to get totally burned.
Bad analogies for the win!
1 comments:
Roster variation can be a winning strategy if you are only playing multis. This assumes one of your combinations will win multiple buy-ins by finishing 1st.
IMO varying rosters in HU (outside of maybe 1 WR or TE) is a slow death sentence to the rake.
Post a Comment