Very Newbie Questions

[DCWildcat]

Thanks for taking the time to read this.

I don't know a whole lot about sports betting, but I do know a decent amount about statistics, as well as a decent amount about basketball. I was wondering if you betting cognoscenti could help me with some hypothetical stuff.

1. How do you react when a score is drastically different than the line? I'm sure you have all seen cases where the line is -3, the sharps agree that it's pretty much the true line, and then lo and behold the underdog wins by 22. In this case, what conclusion do you come to, if any? Some ideas:
"We never really have enough data to make accurate projections on these lines, so that's why everyone was wrong."
"We over/underestimated certain factors; we had the data, we just didn't use it right."
"We're still confident in that line. If they played again tomorrow, with no conditions changing, we'd still go for the favorite at -3 (or maybe a little different but not much), because the underdog just got incredibly lucky."

2. This is somewhat related to the first because it might be a possible answer to it. Basically, I've been kicking around the idea that if two teams were to play over and over again, without any conditoins changing (injuries, etc.), the results of those games would form some sort of a normal distribution that clusters around some score. For example, suppose we've got a college basketball game, Kentucky vs. Louisville, and Kentucky is a 3 point favorite (and the line is correct). Might it be possible that that scoring margin (3) would be the mean/median/mode point of the bell curve (where 100 is for IQ), while scores of UK winning by 2, or 0, or losing by 5, 10, etc., would fall farther left, while UK winning by >2 would fall on the right?

I feel like this idea is so obvious (not necessarily correct, just intuitive) that someone else has probably already come up with it. After all, player's points, team points, etc., are normally distributed, so it seems plausible that game results could be the same way. If someone else already has come up with it, does anyone have any information on it?

Sorry the post is so long, but any help is greatly appreciated.

[Lori]

In cricket betting you quite often get situations such as:

Team A needs to score 350 to have a first innings lead priced at even money.

However, the spreads on what score that team might get are priced at 360-365 because on the 50% of the time they score over 350, they score a lot more, but on the times they score less, they don't score much less.

That means their average score is higher than 350, but they still only have a 50% chance of getting 350.

You could construct a similar situation in football where a star quarterback is 50% likely to play the game and he's so amazing that he's worth 14 points to the team (exaggerated for effect) If the two teams were equal when he plays, you might see the other team as a smallish favourite and yet regularly lose by 14.

I'm not sure that either of these things address your points directly, but I have bet many times on situations where if I lose, I'm really going to lose, but I still feel I'd win over half the time.

Lori

[Lori]

[ QUOTE ]
Might it be possible that that scoring margin (3) would be the mean/median/mode point of the bell curve (where 100 is for IQ), while scores of UK winning by 2, or 0, or losing by 5, 10, etc., would fall farther left, while UK winning by >2 would fall on the right?


[/ QUOTE ]

In general, I suspect that this is correct. Of course, factors like injuries or ejections can put big lumps in your curve.

Lori

[rjp]

What you have to realize is that we're talking about ODDS, not what is going to happen. Odds are a blowout doesn't happen in the case you mentioned, but IT CAN AND WILL HAPPEN.

Embrace this, understand it, and enjoy it. It's like getting sucked out on the river in Hold'em, or the closer blowing the game in the bottom of the ninth. You expect this type of stuff to happen, and everything is possible, though most of it is improbable.

OK, after that rant, everything can and will occur at some point, but if you consistently have the odds in your favor then you have a good chance at winning money.

[DCWildcat]

Rjp, I don't think that really answers my question. I understand what odds are, I'm a gambler. I'm asking about the probability distribution underlying sporting events is.

[Lori]

[ QUOTE ]
Rjp, I don't think that really answers my question. I understand what odds are, I'm a gambler. I'm asking about the probability distribution underlying sporting events is.

[/ QUOTE ]

I suspect also that it depends on the sport. Something like Basketball I would expect to follow a much more 'tidy' curve than something like Football that can have huge turnarounds on a handful of plays.

Lori

[rjp]

If you know the odds then what exactly is your question?

The answer I gave is that the odds are very small that it happens, but it does happen. The margin of victory distribution has roughly half of games decided by 10 points or less, and half of games decided by more than 10 given that a team wins the game. The underdog doesn't win as much, but they too have a chance of winning by more than 20 points.

[johnp158]

I don't have gambling experience, but do have a comment about your second question. Though it seems intuitively correct to think that you could have repeated sports events falling into some kind of normal distribution around a “correct” line, doesn’t that thinking assume that each event is independent? In reality, a team in this situation would be constantly adjusting according to results of past events, which, in this case, are a part of the same distribution. Hmm…this probably isn’t particularly relevant to reality, since you’ll never see the same form of the same team play against the same form of the same team repeatedly. Even in a 7-game series, you’ll have injuries, etc.

So I guess I’m coming back around to agree that it does make sense (to me) to think of a single game in the sense that a good line represents the hypothetical result of many repeated instances of the same event, assuming all conditions are kept constant. Am I on the right track?

[Performify]

[ QUOTE ]
I suspect also that it depends on the sport. Something like Basketball I would expect to follow a much more 'tidy' curve than something like Football that can have huge turnarounds on a handful of plays.


[/ QUOTE ]

Most definitely.

Many casual handicappers forget the importance of turnovers in the NFL. A drive killing turnover that results in a score can be a 14-point swing: reference Favre's INT in the end zone yesterday where New Orleans then drives down to score a TD in a few plays. That's a 14-point swing to the game. which is why most generally include the probability of turnovers and the corresponding probability of point swings resultant from turnovers.

With NFL football, empirically, you've got a clustered distribution but I don't think its right around the spread, its around the key numbers 3,4,7. A high percentage of favorites win, regardless of covering. So you've got a high percentage of score distributions on the winning side: most of the time its just 3,4,7 points regardless of the spread. Then occasionally its a much larger blowout: the 13 point favorites that win by three or four touchdowns. Occasionally the underdog pulls the complete upset, again usually by 3,4,7 points and very occasionally by a large blowout margin, but these are generally pretty rare empirically.

i'd hypothesize that if you graphed the data you'd end up with something like:

Favorite's margin of victory:
>-7: ====
-7 : =========================
-4 : =========
-3 : =========================
+3 : ================================================== =======
+4 : ===================
+7 : ==================================================
+10: ===================
+14: =========
>14: ====

Bars not exact obviously, but should give some idea of what i'm trying to convey. I don't think the spread factors in as an center to the distribution, most games will cluster around one team winning by 7 or by 3 regardless of the spread. but the historical data is certainly out there if someone wanted to put this information together, or start tracking it from the beginning of this season

[rush66]

Performity,

So based off that. Wouldn't teasing favorites to say -7, +3, or +7 be very profitable (in theory).