10-23-2007, 10:37 PM
In week 6 I reverted back to the week 1 “expert”, CBS’ Jamey Eisenberg. This is coming a week late because I've been unmotivated and, well, he actually did pretty well and I couldn’t think of any way to spin it into making him look like an uneducated ass. But now I have, so here are the results:
36/62 = 58%. The best showing thus far, by quite a large margin. 16 of his 31 start ‘em’s turned out good (52%) and 20 of 31 sit ‘em’s (64%). He didn’t include the likes of Ronnie Brown or Larry Fitzgerald in his start ‘em’s and didn’t really include many obvious choices at all. Sure, I could argue that McNabb, Eli Manning, James, McGahee, Edwards, and Winslow are all good enough to start every week, but even if those six were to be excluded, the results would still be very good.
So what’s his secret? Might he actually be an expert? Has all of my work been for naught? Probably. But that doesn’t mean I’m wrong or that I’m going to stop. Why did this guy do so goddamn well last week? Once again Eisenberg had more success picking sit ‘em’s. So I decided to analyze who he chose as his sit ‘em’s. My hypothesis here is that at least half of his picks (let’s call it 16) are players that are owned on less than half of all fantasy teams. In order to test this, what I ideally wanted to do was look at all 31 sit ‘em’s and what percentage of teams own them across two fantasy sites – CBS and ESPN, and then take the average of the two. However, CBS only provides this information for players who are already owned in the league you are in. Since nine players are not owned at all in my league, I could not obtain the stats on these players. So instead I just looked exclusively at ESPN’s own percentage. Now, these percentages are a bit skewed because they were taken today as opposed to at the time these picks were made. But let’s just chalk this up as part of the 5% margin of error inherent to any statistical experiment.
The results show that not only are 18 of his 31 picks owned in less than 50% of all ESPN fantasy leagues, but that 11 of the 31 are owned in less than 25% of all leagues. 58% of his sit ‘em’s are players that are owned in less than half of all ESPN leagues. Ironically, that’s the same percentage of good picks he made all week. So what does this prove? That Eisenberg takes marginal players, waiver wire fodder that most teams don’t even own in the first place, and suggests not starting them. So if these players are owned at all, it’s probably to fill out the bench, or to serve as an injury or bye replacement, in which case they have no choice but to start the marginal player. In other words, this advice looks all good on the surface, but when digging deeper it turns out that it doesn’t take much skill or expertise to tell people not to start the likes of Gus Frerotte or Roddy White or Olindo Mare or any other player that on most teams would be a third or fourth option at best.
I didn’t look at anything in week 7 because this is all getting annoying and tedious. I now have 6 weeks of data though. To keep me interested in this thing and spice things up a bit, what I will do starting next week is provide my own start em/sit em picks each week and at the end of week 13 compare 6 weeks of “expert” advice vs. 6 weeks of my advice. I will make the same number of picks each week as the expert made in the corresponding week – for example, week 8 will mirror week 1’s picks, week 9 will mirror week 2, and so on. Through six weeks of data, these “experts” produced with 48% accuracy, going 133 for 274. This is higher than my hypothesized 40%, but I’ve already addressed the contributing factors that have led to this. Besides, anything less than 50% is pretty poor when it comes to making “expert” decisions on things. My hypothesis for the next six weeks is that I will match or better their 48% performance, +/- 5%.
36/62 = 58%. The best showing thus far, by quite a large margin. 16 of his 31 start ‘em’s turned out good (52%) and 20 of 31 sit ‘em’s (64%). He didn’t include the likes of Ronnie Brown or Larry Fitzgerald in his start ‘em’s and didn’t really include many obvious choices at all. Sure, I could argue that McNabb, Eli Manning, James, McGahee, Edwards, and Winslow are all good enough to start every week, but even if those six were to be excluded, the results would still be very good.
So what’s his secret? Might he actually be an expert? Has all of my work been for naught? Probably. But that doesn’t mean I’m wrong or that I’m going to stop. Why did this guy do so goddamn well last week? Once again Eisenberg had more success picking sit ‘em’s. So I decided to analyze who he chose as his sit ‘em’s. My hypothesis here is that at least half of his picks (let’s call it 16) are players that are owned on less than half of all fantasy teams. In order to test this, what I ideally wanted to do was look at all 31 sit ‘em’s and what percentage of teams own them across two fantasy sites – CBS and ESPN, and then take the average of the two. However, CBS only provides this information for players who are already owned in the league you are in. Since nine players are not owned at all in my league, I could not obtain the stats on these players. So instead I just looked exclusively at ESPN’s own percentage. Now, these percentages are a bit skewed because they were taken today as opposed to at the time these picks were made. But let’s just chalk this up as part of the 5% margin of error inherent to any statistical experiment.
The results show that not only are 18 of his 31 picks owned in less than 50% of all ESPN fantasy leagues, but that 11 of the 31 are owned in less than 25% of all leagues. 58% of his sit ‘em’s are players that are owned in less than half of all ESPN leagues. Ironically, that’s the same percentage of good picks he made all week. So what does this prove? That Eisenberg takes marginal players, waiver wire fodder that most teams don’t even own in the first place, and suggests not starting them. So if these players are owned at all, it’s probably to fill out the bench, or to serve as an injury or bye replacement, in which case they have no choice but to start the marginal player. In other words, this advice looks all good on the surface, but when digging deeper it turns out that it doesn’t take much skill or expertise to tell people not to start the likes of Gus Frerotte or Roddy White or Olindo Mare or any other player that on most teams would be a third or fourth option at best.
I didn’t look at anything in week 7 because this is all getting annoying and tedious. I now have 6 weeks of data though. To keep me interested in this thing and spice things up a bit, what I will do starting next week is provide my own start em/sit em picks each week and at the end of week 13 compare 6 weeks of “expert” advice vs. 6 weeks of my advice. I will make the same number of picks each week as the expert made in the corresponding week – for example, week 8 will mirror week 1’s picks, week 9 will mirror week 2, and so on. Through six weeks of data, these “experts” produced with 48% accuracy, going 133 for 274. This is higher than my hypothesized 40%, but I’ve already addressed the contributing factors that have led to this. Besides, anything less than 50% is pretty poor when it comes to making “expert” decisions on things. My hypothesis for the next six weeks is that I will match or better their 48% performance, +/- 5%.