Guy was bragging in the poker room picking 60% winners was 'easy' in reagards to our football pool (Pinnacle lines). I offered odds off the top of my head and wanted to see how much I have the best of it, assuming I do.

We pick 5 games a week for 17 weeks, so 85 picked. He needs to go 51-36 (or better) to win. So, assuming his picks are all coin flips, I'm only losing if he gets lucky. What are the odds he gets 51 winners or better?

TIA

Is this against the spread?

My bad, yeah, using the spread, and he's just an average football guy at best.

This can be answered using the binomial distribution. The following site makes the calculation:

https://www.danielsoper.com/statcalc...tor.aspx?id=71

The answer is given as 4.1% chance of 51 or more correct picks.

Thanks for the link. Multiple searches on my own came up empty. I could only find results for exactly 51 wins.

I laid 4-1 so getting the $$ in good anyway.

60% against Pinnacle lines, lol. Either he's the best picker in the world (better than professional sports bettors), or he's an idiot who needs to get very lucky.

He might even be worse than 4.1%. If the lines are early lines, he's slightly worse than 50/50 each game, because he's more likely to take the dumb side than the sharp side.

Edit: Since he only has to beat the spread and is never laying odds, whenever one side of the line is laying extra odds, you should move the line a half point. For instance, if it's -2.5(-115), move the line to 3. You're still using the Pinnacle line, it's just two forms of the same line.

Even the best picker in the world cannot achieve 60% picking five games a week. All the best picker in the world can do is come up with the TRUE pointspread, ie the number that would split the difference if the game could be played multiple times under the exact same conditions. Then he would have to come up with five games a week where his line was different from the bookies by an average of about three points per game. I am almost positive bookies lines are not off from the theoretically true line by that much that often.

Exactly, "better than anyone" was an understatement, it's really "better than possible". 60% might be achievable if you hardly ever make a pick, but not 5 picks a week. There definitely aren't 5 lines per week that move 3 points (or the equivalent of 0 to 3) from open to close, and databases show that closing lines are accurate.

Week 1 is where the most of the wacky lines are to be found if you place your bets during the summer. After that, if you see a 3-pt movement it's almost guaranteed to be from an injury update.

That's not exactly right. Databases show that closing lines are accurate in aggregate. But you are implying that closing lines don't offer good bets. If green coins come up heads 45% and red coins do it 55% and there are an equal number of each, you will put them at pickem if you don't know about the colors. And if I am the only bettor who knows it I will win, the public will lay 11-10 and lose to you, and the statisticians will say your line was accurate when its not.

My point is that if a line only moved 0.5 or 1 point from open to close, it's pretty safe to assume the closing line wasn't off by 2 to 2.5 points*, so the lack of 3-pt line movements in the NFL is a good indicator that there aren't five 60% picks to be made each week.

*Not saying it's impossible, but even if closing lines are that soft and just waiting for some person/AI to come along and exploit them, OP's friend isn't that person.

My quarrel is with the statement that "databases show that the closing lines are accurate." Based on what? Would the test that supposedly shows them to be accurate say that opening lines are not?

The outcomes are compared to the spreads. If a line is accurate, 50% of the non-pushes should cover and 50% shouldn't.

Opening lines are accurate too, but not as accurate as closing lines. Injury statuses and weather forecasts become clearer, but even without new info, there's the influence of smart money. Professional bettors are better handicappers than the oddsmakers, so when they move the line, the line becomes more accurate.

But again you are talking in aggregate. I am thinking that the technique used to come to this conclusion is fallacious and that it may be similar to a similar technique that results in the conclusion that the stock market is efficient.

Every result could be fixed while the lines still appear “accurate”

Or more practically, if the lines are accurate under one set of variables, that doesn’t mean there aren’t other variables that could be understood by an expert bettor to gain an edge.

I don't think anything like a true pointspread exists. Why do you think that "if the game could be played multiple times under the exact same conditions" you would have different outcomes?

Naturally he goes 5-0 the first week.

Moves him to 11% to win.

Good time to double the bet

Remember that the 4.1% figure was calculated under the assumption that he's exactly 50% to correctly call each game. 60% is way overconfident, but it's possible he does have a slight edge on his picks. For example if he has a true pick rate of 52.5%, then it moves his chance of winning the original bet all the way up to 10%, and the chance of winning the bet after having picked the first 5 correctly is 21.7%. If his true pick rate is 52.5% then it would be -EV to double the bet right now at 4:1 odds against him.

Mainly because by "exact same conditions" I meant all the info available to humans before the game were the same. I didn't mean it literally (for example the wind speed 17 minutes into the game.)

BUT even if I meant it literally subatomic uncertainty plus the butterfly effect would mean very different outcomes. No different than if you trip before you reach your poker table, your results for the day is likely to be very unlike what they would be if you didn't trip.

My thoughts exactly. I was going to offer him 3-1 from here on out. My buddy suggested offering him odds weekly to reduce variance.

Even if this was correct the 3-1 offer would still put me in a +EV position,
but unless he is some kind of savant, naturally intuiting when teams are on or off, there is a close to zero chance he could pick at 52.5% or better. He's a recreational poker player, casual football watcher.

Plus, I'd think anyone who could actually pick over 50% would realize just how hard it is to pick at a 60% clip. The fact that he was willing to take such bad odds leads me to believe he is just another coin flip bettor (or worse).

It's possible that the closing lines are consistently off by 3 points, but undervaluing favorites and underdogs equally often. For instance, if the line is +3 then 50% of the time the "true" line is 0 and 50% it's +5.* That would make lines appear accurate in aggregate when in fact there is a treasure trove of available information not being factored into handicappers' formulas. I don't know how likely that is. I'm pretty sure that if you ask the people in the 2p2 Sportsbetting forum, they'll say it's very unlikely.

If the lines were off by that much in both directions, it would mean the margins of covers have a higher variance than they would if the lines were correct. I wonder if there's evidence of this.


*5 isn't a typo; I think the worth of moving from 0 to +3 is closer to that of moving from +3 to +5 than to +6. And by "true line" I mean the most accurate line possible without knowing all the unknowable variables (much like a coinflip is 50% to us despite the process being deterministic).

In real life you original comment is almost certainly almost completely true. There is no way half the lines are off by a lot in one direction and half in the other. I just quarreled theoretically with your contention that final lines have been shown to be correct. The fact that favorites cover 50% doesn't prove it. A few lines are probably off a little in one direction and a few in the other. But that fact is overwhelming evidence that the lines are in general close enough to the "true line" (using the definition we both agree on) that there are not five 60% bets per week.

Side note, closing lines can be soft when enough people are betting on an event. Certainly it was evident in the Mayweather/McGregor line, the dumb money drowned the smart money and the line moved in the direction of even odds, closing at something like Mayweather -500 or lower. Betting Mayweather was possibly the most +EV sportsbetting opportunity of all time (aside from arbitrage and inside info of fixes).

That might happen (to a much lesser degree) in NFL betting too (e.g. perhaps in the post-season and some MNF games).

When the line move in football or basketball is small it could easily be dumb money because, unlike the stock market, professional bettors, having to lay 11-10, have no reason to bet into slightly incorrect lines.

In boxing the syndrome can be even more pronounced because the bookies vig is usually so much greater AND because the mere existence of a weird line or weird line move makes pros worried about shenanigans. If the Mayweather fight had been certain to be honest the dumb money would not have drowned out smart money. Warren Buffet alone might have bet a billion.

Reality sets in. He picked 5 favorites (3 road favs) and went 0-5 in week 2.

He's now at 3% to win the bet.