Missed this the first time around. While most of this guy’s reasoning seems fair, the above quote, if it means what I think it means, is terrible. This is like betting on heads after tails has come up three times because “it’s due.”
That’s sort of what it sounds like, but a coin has an equal chance of landing on either side. So it’s bad logic. He qualified his statement by saying that the Cards and decent and the Lions suck, making the argument a little less absurd. Still, pretty flimsy logic though.
Still, there’s something to it. A team that loses too much grows desperate and may put in extra effort to get out of the losing streak. Similarly, a winning team may become complacent, especially against a team in the beginning of a losing streak.
You understand what I mean. I went through about five different adjectives trying to describe the phenomenon, then just went with “higher” assuming anyone reading would already know or could easily figure out what I was getting at. Which you did.
The short form is: as the knuckleheads pile money on the Lions, the line will become more attractive for Cards bettors, which is the side that (logically) is “favored” by the knowledge unavailable to the unwashed masses in the first place.
Oh, and just for the hell of it (and without commentary), the division rankings for the midpoint of the season:
#1 AFC South 16-5 (76%)
#2 NFC East 16-6 (73%)
#3 NFC North 13-7 (65%)
#4 AFC North 11-9 (55%)
#5 NFC South 10-13 (43%)
#6 AFC East 8-14 (36%)
#6 AFC West 8-14 (36%)
#8 NFC West 3-17 (15%)
#1 AFC 20-19 (51%)
#2 NFC 19-20 (49%)
I wonder if that’s actually true. It would be pretty easy to test. Just take a few dozen teams that finished the year with the same record, then see if their winning percentage following two, three, or four losses was significantly better than their overall winning percentage.
Do you want to get started on that, or should I?
You mean against the spread? We already have several years of that data compiled in Omni’s old threads, when I compiled the records against the spread and used those numbers to generate “spread-picker” picks.
Historically, the spread-picker system has a winning record, if marginally so. One thing I noticed in the many, many years of tabulating the spread-picker system is that while it works in aggregate, every single year there are two or three teams that will kill you, either by running roughshod over the spread all season or getting killed by it.
This year, for example, the Patriots would be a best bet (to lose) nearly every week, and would be dragging it down pretty hard. I’m guessing the Rams wouldn’t be doing it any favors either.
Except that, for every coin that lands heads, the next toss has a higher and higher probability of being tails, when taken as a whole. Sure, each toss itself is a 50/50 shot, but after 5 heads in a row, a sixth heads in a row is insanely unlikely. There’s a hidden counter-probability to it, while there’s a 1 in 2 chance that the sixth toss lands heads, there’s a 1/64 chance that six tosses all land heads. How I’ve always looked at it at least.
On to the football! I bet the NFL every week, and as of late (using a new betting strategy I developed) I’ve won bets every single week.
Here are the games that, without any research, jump out at me as bet games.
**Buf -3: **This one seems seems loony. Do people still think the Bills are a bad team or something? They lost to Denver, but if you watched the game, they should have won. They lost to Pittsburgh and New England, both of which are understandable losses. They lost to Dallas, but again, probably should have won. And lately, they have beaten the last three teams they have played, all by a healthy margin. Granted, all three of those teams are bad teams, but this all tells us three things:
1.) They’re on a roll. Momentum is important.
2.) They beat bad opponents by a healthy margin.
3.) Miami is a bad opponent.
I bet the Bills win by 10.
StL +12: Plus twelve? From what I’ve seen this year, Vegas tends to favor New Orleans and Philadelphia when they shouldn’t. Here’s why this game seems like a lock: Did anyone see the Rams last week against the Browns? They actually dominated the first quarter, before Steven Jackson went down again with an injury. New Orleans has a terrible defense, and St. Louis has a deceptively good pass defense. If Jackson plays, I can see St. Louis staying close in this game, running the ball effectively and shortening the game, limiting the amount of scoring the Saints can do. They absolutely can move the ball with Jackson in the game. The Saints will likely win, but I’m not sure they win by 13. Probably 7-10.
GB -6: Green Bay passes heavily, Minnesota can’t defend the pass, at ALL. Of Minnesota’s losses (one of which was to Green Bay already, in Minnesota), four out of five were to teams that heavily feature the pass (Detroit, Green Bay, Philly, Dallas). Neither team will be able to run the ball (including Minnesota, Green Bay’s run defense is very good), and the Minnesota air attack won’t do a single thing against a very good Green Bay secondary. I say Green Bay forces 3-4 turnovers, Favre throws for 3-4 TDs, and Green Bay wins going away.
I’m not sure Chicago is 7 points better than Oakland. Fargas will be starting and he has been very effective, and Chicago likely won’t be able to do anything offensively. Oakland is good against the pass, and Chicago can’t run. Bad matchup there, but I’m not sure I’m confident enough to bet the Raiders.
I’m not sure I’m confident enough in Arizona or Detroit to take either in a bet, but I wouldn’t be surprised if this game was a lot lower scoring than people expect. Detroit exploded last week, but it was largely due to two defensive touchdowns and a running score, so their total was inflated. Detroit’s defense causes a lot fo turnovers and Warner hasn’t been very protective of the football as of late. I expect an ugly, turnover laden game, featuring a lot of field goals. I’d have to look at it, though.
Ind -4: Seems like a statement game to me. I don’t think San Diego’s offense can muster enough against that Colts defense to win.
groan
What you’re missing in that logic is that 5 heads in a row only has a 3.125% chance of happening. So after that 5th consecutive head gets flipped, you’ve already painted yourself into a 3.125% corner. The next flip has a 1.5625% chance of being a head, a 1.5625% chance of being a tail, and a 96.875% chance of never having gotten to that point in the first place.
Really? That it’s entirely unlikely for that occurrence to have happened at all (when it had just happened) is even more irrelevant to the discussion than what the weather was during the flips. You’ve backed yourself into some weird time warp conundrum when you start basing your decisions on the probability that an event, which already happened, shouldn’t have happened.
I know how probability works. I know that every coin flip is independent of any other coin flips, and that if you flip a coin a hundred times, the previous hundred flips don’t influence the 101st. That’s all common knowledge. I’m saying that if a result is extremely unlikely, it is far more likely for that unlikely event to not continue than it is for it to continue into even more improbable territory. I don’t think that 101st flip is even money. At some point, it’s so astronomically high to roll another head, you just have to start betting tails. Works for me, I guess.
No, you don’t. Or you wouldn’t have continued on to say this:
No, it’s actually completely relevant. Try to internalize the concept, because for some it’s the only way to cut through the fog of the gambler’s fallacy. And judging by the rest of your post, you’re deeply mired in it.
As an aside, never bet against a trend, but rather always that it will continue. Flipping a coin, shooting craps, spinning roulette, handicapping football, whatever. There is never any causative agent that will make something “due”, but there’s always something that could be skewing results (however minorly) that could be helping to cause the streak in the first place. Maybe the coin is a trick coin from a magic shop, or the dice are loaded, or the wheel is biased, or the team keeps failing to cover because it’s in a huge market where tons of delusional fans keep distorting the spread.
Nothing will ever make anything hit because it’s due, but plenty of things can contribute to bucking the odds by continuing the trend. This is generalized advice for any 50/50 system. (All of which allow you to systematically lose.)
This:
. . . flatly contradicts this:
The second statement is the one that it wrong.
No offense, but I’d be very surprised if it did.
No, I meant straight up. Haven’t heard of “spread-picker.” Link?
Duplicate post deleted.
I first mentioned it on the SDMB in [post=6641125]this post[/post]. Look for followup later in that thread.
To see weekly results, search for “spread picker” (no quotes) with me as the author.
THIS FUCKING BOARD SUCKS WITH THE CONSTANT FUCKING TIMEOUTS
Interesting theory. What’s the system’s winning percentage, and over how many games?
I said:
So, I went and did this. I looked at every 5-11 team in NFL history – there have been 62 of them since the league went to a 16 game schedule in 1978. A 5-11 team has an overall winning percentage of .313. I wanted to see if these teams were more likely to win following a losing streak of two or more games than they were in general. If their W% in such games was approximately the same as their overall W%, the theory posited by Team of Scientists would be proven false.
The actual won-loss record of 5-11 teams in games following two or more consecutive losses was 145-241, for a W% of .376. This result surprised me, and I’m not sure how to look at it. A few possibilities:
-
It really is true that teams on a losing streak try extra hard to snap out of it. I find this hard to believe, but given the information I have at the moment I have to seriously consider it.
-
There’s a problem with my methodology that hasn’t occurred to me. For instance, should I be discounting two lossess from the 5-11 record, since I can’t start counting results for a team until two of its losses are off the table, and its record from that point on is actually 5-9 (.357)? If this is the case, it would go a long way toward making the next possibility correct. . .
-
The nature of NFL schedules tends to stop losing streaks. Teams rarely play three consecutive road games, and the NFL makes some small effort (I assume) to alternate between strong and weak opponents on the schedule.
Thoughts? Especially about #2?
It is absolutely not relevant what the chances of something NOT happening are, when they are happening.Saying that the 6th flip has a 1.whatever chance to be heads and the same chance to be tails, and then a 95% chance to have not happened is absolutely, insanely, irrelevant. It matters none what should not have happened when it has happened. This is like saying “there is a 35% chance of rain this hour, but it won’t rain because there was a 95% chance of it not raining last hour.” Insane. I actually can’t comprehend the logic of this, and I’m really trying.
As an aside, I only gamble football, and I ride as many hot streaks as I can, but this is different.
Here’s how it works. You flip a coin 1000 times. Logic and probability law dictate that the coin will eventually end up flipping very close to 500 heads, 500 tails. It doesn’t matter how many times you flip the coin, it has a 50/50 chance of heads and tails, right? So following that logic, any number of times flipped, the coin should come very close to 50/50. The larger the sample sixe, the more likely it will eventually come very close to a 50% split.
Now what if the first 300 flips were 300 heads in a row. Insanely, unimaginably remote, I know, but go with it. We know, by rule, that over any large number of flips, the coin should come out 50/50. That means that at some point the tails have to far outnumber the heads in order to even out to its own probability. Over the next 700 flips, you’re far more likely to get a large proportion of tails than heads. In fact, it should be close to 200/500. It won’t of course, but it WILL even out. This is an idea of how something being “due” works. If you saw someone flip a coin heads 20 times in a row, and they said they would flip it 80 more times, you’d be insane to keep betting heads, because of this, right?
This is really silly. We’re obviously talking about a fair environment, that’s just a given. To say the game is slanted, however it may be, is completely irrelevant too. You’re definitely grasping for straws with this, even if the overall advice to ride hot streaks has merit. You make it seem obvious and simple to always follow a hot streak of any kind, obviously anyone running a gambling game will want betting to continue along 50/50 lines (like a football game’s spread). It isn’t profitable to stack the deck or load the dice in a game where you don’t actually compete against someone. If you load the dice, eventually people will figure it out and ride the dice. You’re gping to lose money. There isn’t any point to bring up loaded dice, magic coins, stacked decks, any of that. I mean, the entire point of this threadjack is talking about statistical anomalies and how they are bound to reverse themselves eventually. Statistical anomalies, not cheating.
There’s a reason you won’t flip a coin ten times and hit ten heads in a row. It’s too unlikely. You could do it a thousand times and probably not hit ten heads in a row. I wonder how that isn’t proof in itself?
Absolutely they do, that’s the point. Common, simple statistical probability indicates coin flips are independent of one another. Common, simple geography also once indicated that the world was flat, and that the Sun orbited the Earth.
None taken. Obviously coins are different than any other type of game because the coin has a set, defined probability of 50/50, and you can only pick one of two outcomes (unlike a die where you can be right that a 6 will not come up an 11th time in a row, but still lose because you picked 3 instead of 1). It’s stupid to bet a losing team in football because there isn’t a defined set of odds, there’s a reason they are losing over and over and it isn’t a statistical probability that it will eventually even out. Teams go 1-15 all the time. It’s stupid to do in slots, or anything like that. But coins? 50/50.
Ok, I’m going to try to explain as clearly as I can why this is wrong. However, if I fail and you still don’t see it, I beg you to take my word, and the word of the 20 other people who are going to join in and explain why you’re wrong … because you are, absolutely and unequivocally, incorrect.
Yes, over 1000 flips, the probability is extremely high that the eventual split will be very close to 50/50. However, some infinitesimally small percentage of the time, the first 300 flips will come up tails. Assuming the coin is fair and that there isn’t anything supernatural going on, this tells us nothing about what’s going to happen over the next 700 flips. It was a freak occurence, that’s all.
To illustrate, let’s make it even simpler. Instead of 1000 flips, make it 600. By your logic, if the first 300 flips are tails, then almost all of the next 300 should be heads. Doesn’t that seem insane? It should, because the chance of flips 301-600 being all heads is 1/999,999,999,999,999,999 (or so), just like the first set of 300.
If you don’t belive me, please test this out for yourself on a small scale. Flip a coin in sets of 10. For every set in which 4 of the first 5 flips are the same, add up the second 5 flips, counting what percentage of flips have the opposite result as the majority result from the first 5. Do this long enough, and the second 5 will come down approximately 50/50. I promise.