tomarken
Footballguy
jon_mx said:Actually he has been dead on most weeks, and this week is still very good. The smaller sample size in the least likely to succeed groups makes variations much more probable. All of the groups under 50 percent contain less than 100 people. Just look at the group 30-34...there are 38 people in that group. In order for his prediction to come about, 12 or 13 teams would have had to survive. Just because 16 teams survived, it does not mean his predictions understated the odds, it just means something unexpected happened. Probably 1 or 2 players had better weeks than expected a lifted a few teams up. Besides, if you flip a coin 100 times, and heads comes up 57 times, you can't conclude that your prediction of 50 was understated. That is just how random events work.CalBear said:This confirms my intuition on the simulator--that high survival percentages tend to be overstated, and low survival percentages tend to be understated. Looks like the crossover is around 70%.Simulator accuracy report (now broken down a bit further):
Prob # surv pct-------------------------- 100 1 1 100.095--99 1262 1226 97.190--94 1280 1182 92.385--89 945 807 85.480--84 647 522 80.775--79 470 372 79.170--74 362 263 72.765--69 270 185 68.560--64 210 137 65.255--59 173 101 58.450--54 124 73 58.945--49 96 52 54.240--44 69 35 50.735--39 68 24 35.330--34 38 16 42.125--29 29 6 20.720--24 28 12 42.915--19 13 1 7.710--14 21 2 9.5 5-- 9 7 0 0.0 0-- 4 7 0 0.0
If you predict 50 and it comes up 57, you did understate it. Besides, a coin-flipping analogy doesn't work here because a coin has predetermined odds. We know that it's 50/50 and could arguably verify that outside of a Monte Carlo simulation.Drinen's predictions and the final results are slightly different in the way they are being interpreted (the former is an individual team's chances of survival, while the latter is the overall survival rate for a collection of teams) but ignoring that for a moment, if Drinen says 50% and it turns out to be 57%, he did understate it. That's not a criticism of Doug, of course, no one would expect him to get them exactly right - and besides, if he's wrong it's really Dodds's fault.
I have a 24 man roster and I have 4 of the players from the high side of your comparisons and no players from the low side. This is not where the difference in the rosters is typically going to be. I think most people did like I did when picking their rosters. Pick the studs they want first, then the cheap bargains, then fill in the rest with middle of the road players. The difference between a 20 and 24 man roster is more typically going to be something like one $14 player vs. five $1-$6 players. It's not going to be Shaub vs. Edwards or Rodgers vs. Romo.
Twenty four is the only answer............
Twenty four is the only answer............
-QG
Still have a 90.9% chance of advancing despite my 7 players on a bye.
ollars spent, on average, on your most expensive player, your second-most expensive, and so on. Broken down by roster size.