How large of a sample size is needed to suggest it is not a fluke? I have 723 bets tracked so far winning 399 so 55%. Is that a big enough sample size or do I need like 2,000?
-
-
This question gets asked 40 times a month. You probably got lucky. It is less likely if you get to 2000.Comment -
There isn't a magic number for sample size, it's a combination of win rate and sample size. For example, assuming you're talking bets around evens, the chance you're just coin-flipping is about 1 in 340. Increase the win rate, or get the same win rate at a larger sample, and that chance gets less likely.
For example, if you can have that win-rate at 1000 bets, it's 1 in 1800. At 1500 bets it's 1 in 32000, and at 2000 bets it's 1 in 554k.Last edited by HeeeHAWWWW; 12-08-18, 12:37 PM.Comment -
focus on line value not a win loss percentageComment -
It's just a simple binomial test, which you can use if betting at constant odds (if you vary odds it gets a little more complicated). For example:
binom.test(550,1000,p=0.5,alt="greater")
number of successes = 550, number of trials = 1000, p-value = 0.0008653
The p-value is the chance of 550+ wins over 1000 trials, if your true win rate is 50%.
For this scenario though, it's probably more likely the record being tested is that of a marginal winner getting a little lucky, rather than a total coin-flipper getting very lucky. Here's the above for a genuine 53% player:
binom.test(550,1000,p=0.53,alt="greater" )
number of successes = 550, number of trials = 1000, p-value = 0.1082Last edited by HeeeHAWWWW; 12-08-18, 01:35 PM.Comment -
Line closes say -5 I usually have -4.5 bearing the close by .5 point. That’s what I’m sayingComment -
You should know before even placing the first bet
Also honestly lol@waiting 1000s of bets before knowing if its working or not
I feel like none of you actually bet day to day
Imagine betting nfl and thinking ur just lucking it for 10 years because u cant get a high enough sample size
Just fking lolComment -
you were a statistically significant winner if those were your actual bets and your only bets over that period....... there are tons of binomial calculators you could play with to determine the significance of winning record vs. 50% (beating the market, but not the juice) or vs. 52.x% (beating the -110 spread).....
this does not mean you are necessarily a winning bettor today or will be in the future.Comment -
Even if it isnt luck, 55% is not the most massive edge. However, if implementing the proper bankroll management, it is actually more than enough.Comment -
If all anyone has is data, it would require a very large sample size to show if someone is truly making winning bets. Having both data and a theory means it could happen much sooner. Possibly even within the first bet (an easy example would be betting the Bulls at +160 and the Timberwolves at -150. With the right bet sizes, it's guaranteed to be a profitable strategy).
Without a theory, it means we're left wondering if playing basketball makes people taller? Or is it meaningful if the Blazers are 15-1 against the spread as home favorites following a win as a road underdog? (The data supports both of those being true)
Most people here say beat the closing line (theory) and have a winning record (data) and you absolutely are making winning bets. I make it even easier for myself and say getting the best line available at the time of the bet is usually enough (assuming it's better than market price by enough to beat the juice).Comment -
When a new casino opens and they take a bet on black at even on the very first spin on the roulette wheel, they know it's a good bet for them. All they need to know is that 18/38 = 0.47. They don't need to analyze hundreds or thousands of results to get enough of a sample size to know that is a good bet for them.
Last week pinnacle had Nuggets over 220.5 at +120. I was in a sports book that was offering the under at -110.Comment -
yes but sports betting is not like roulette. even though odds show you have an edge, you can still lose or that edge can change tomorrow. in roulette its simple math in sports betting not so much even though in the long run beating the line is the best way we can improve our chances of winning.Comment -
example say you figure out before everyone that Tom Brady is a decent Qb and there is a good chance Patriots might win, so you bet them early and magic they win, books say it was a fluke and barely adjust next week and boom they win again, books say shit happens and move the line just a hair and bam they win again, and now guess what your edge is gone. Look at this years Cleveland Browns
started 4-2 and now 3-3 edge gone! yes not an adequate sample size and not relevant but the point isLast edited by danshan11; 12-09-18, 11:35 AM.Comment -
its really simple to see how beating the line works in reality not theory.
take any 2018 MLB team and add a dime to the moneyline and you will win with 70-80% of the teams I have not done them all but yeahComment -
our job as bettors is to do only thing. We have to predict accurately what the closing line will be when the line comes out. We have to do this correctly at a high rate so we can bet the deficiencies
if you want to see how good a capper is ask him what his number is on a few games BEFORE any lines have come out and if he is real close he is probably really good.
right now I will give you my tomorrow NBA lines (without injury adjustments ) and you be your own judge if I know what I am doing in the NBA. if somebody cant do this or is not even close to the actual number they are not serious or dont understand the game. totals have 2 numbers cause I use a backup model to get another perspective so I dont get one sided. I bet after adjusting for injuries on games where the side is off by 2 or more and the total is off by 3 or more. This season I am so far at 1% clv that is about 10 20 cents too low to cover the margin but it will improve, my model has a learning part that makes it smarter as the season cruises a long. I expect by the break to be better than the margin and in the Black
12/10 Wizards 219.5 218 12/10 Pacers -4 12/10 Pistons 224 224 12/10 Seventysixers -7 12/10 Pelicans 224.5 226 12/10 Celtics -4 12/10 Cavaliers 224.5 221.5 12/10 Bucks -12.5 12/10 Jazz 215.5 213 12/10 Thunder -4 12/10 Kings -3 12/10 Bulls 225 225.5 12/10 Magic 214 208.5 12/10 Mavericks -4 12/10 Clippers -9 12/10 Suns 221.5 222 12/10 Grizzlies 203 208 12/10 Nuggets -4.5 12/10 Timberwolves 223.5 216 12/10 Warriors -6 12/10 Heat 216 212 12/10 Lakers -6 Last edited by danshan11; 12-09-18, 11:48 AM.Comment -
I believe not fact but that any thing that looks at history (system type stuff) and finds areas where they have winners is not a long term success and the reason is NOISE, Yankees win at an amazing rate on Wednesdays and have 1000s of records to show it, that is just noise and people build systems on things like this. Remember yesterdays results have very very little effect on todays game UNLESS someone is injured or does not play but if
Broncos played Toros today and lost 55-14 and the line was Broncos -3
if they played again tomorrow it would be Broncos -2.95 assuming same location, weather and rosters
and -2.95 might be too much of an adjustmentComment -
The casino loses very often on spins of the roulette wheel.
Bet the Packers at +150 and the Vikings at -140 (on the weeks when they are playing each other, obviously) and it's very similar to the house booking roulette (or card or dice games). It's pretty simply math that I'm going to win.
When those options present themselves (which they do nearly every day) I much prefer to take a position and only bet whichever side I think is stronger -- and the math is only slightly different.
I don't know what the line is going to close at. But I do know exactly what lines are available when I do make my bet. If I can beat the market price by more than what the juice is, like you say, "it's simple math."Comment -
unless you are doing arbitrage only 2 lines matter
line bet at and line closed at (pinnacle for most major sports not sure on smaller markets who is sharp)Comment -
At the time of the bet, only one of those is known. If it beats pinnacle take back at the time the bet is made, it will beat pinnacle at the close enough to win
(with some exceptions, such as "soft" openers. Which is why we do need some data other than just this theory. But a lot less than 2000 that the original question was asking about)Comment -
oh yeah if you are just looking for stale lines
yes you will win if you find enough of them to cover variance
and it proves beating the line mattersComment -
My NCAA BB #s 231-180-4 so right at 55%. A lot or the majority I bet 1H and game so that boosts my # some. Thanks for the feedback guys. Yeah I am always tweaking, but the biggest things I found helped me this year is flat betting pretty much everything and correlating 1H and game lines.Comment
