[Bsims]

I use a system that generates 325+ likely scores for a MLB game. Using these scores, I can evaluate potential wagers by calculating the probably of success and then combine this with the odds to compute an expected return. I use this for MLB run line wagers as discussed elsewhere. I also use it to evaluate parlays. On occasion, it will find more than one parlay with a positive expected return on the same game. Today there were 3 games like this.

Parlay	Prob	Exp Ret
bal-133 Over 9-115	30.0%	$1.090
cws+127 Und 9-105	22.9%	$1.064
hou-125 Over 8-105	25.3%	$1.005
cle+116 Und 8-115	27.1%	$1.149
bos-132 Over 7.5+105	27.9%	$1.006
tba+122 Und 7.5-125	28.8%	$1.152

This creates a dilemma. What to bet? There are 3 choices on each game.

1. Only bet the parlay with the highest return.

2. Bet both since they both have positive expected return.

3. Bet 1 unit on the one with the highest return and 1/2 unit on the other .

What would you do?

[u21c3f6]

In these situations I calculate the Kelly %. I would wager on the highest Kelly result first and then the next highest and so on as long as the resulting % was significant enough. I do not wager the full Kelly amount. I wager closer to half Kelly.

Joe.

[ClippersSux]

Have you calculated the possible third outcome? Ties. Ties will be reduce your payout odds tremendously. About 10% MLB games end in a total of 9 runs scored; and another 7% end in a total of 8 runs scored.

[Bsims]

Have you calculated the possible third outcome? Ties. Ties will be reduce your payout odds tremendously. About 10% MLB games end in a total of 9 runs scored; and another 7% end in a total of 8 runs scored.

Yes, I did not include them in the post because they might be confusing. But I do include in the expected return calculations the win-push occurrences. Here is the table including tie probabilities.

Parlay	Prob Win	Prob Win-Push	Exp Ret
bal-133 Over 9-115	30.0%	5.8%	$1.090
cws+127 Und 9-105	22.9%	2.5%	$1.064
hou-125 Over 8-105	25.3%	6.0%	$1.005
cle+116 Und 8-115	27.1%	3.0%	$1.149
bos-132 Over 7.5+105	27.9%	0.0%	$1.006
tba+122 Und 7.5-125	28.8%	0.0%	$1.152

[jtoler]

I like this guy, always playing with the numbers, good luck

[Miz]

i'd probably choose the one with the highest growth percentage and wager on it. I'd forego the conflicting wager to reduce simultaneous events and allow a slightly larger stake (I like 0.4 kelly or so, and maybe less than that with the lower prob success of a given wager). I often have to choose the simpler option for things like this (and skip a detailed analysis) just to make life and time more manageable.

More importantly, since the conflicting events are correlated (albeit negatively) they aren't truly independent (so Kelly may not apply as easily anyway).

Good luck! Sounds like an interesting model!

Did your research show that the dog/under and fav/over bias was a strong tendency over the games you looked at?

[oilcountry99]

The problem is finding a book to take correlated parlays. I would wager on the one with the higher EV

[Miz]

Pretty sure that mlb total/side correlation is weak enough that it is allowed

[Bsims]

Pretty sure that mlb total/side correlation is weak enough that it is allowed

Some books allow this, some don't. There is some very weak correlation. I'll be posting some data in a thread here, and probably on my blog.

[Bsims]

Did your research show that the dog/under and fav/over bias was a strong tendency over the games you looked at?

There is some bias towards those two combinations. Not enough to stand alone. I'm working on some write ups now.

[Miz]

how can i read your blog? You post some good data-driven things.

[Miz]

found it in a different thread. Thanks Bill.

[Barrakuda]

There is some bias towards those two combinations. Not enough to stand alone. I'm working on some write ups now.

link?

[KVB]

...
1. Only bet the parlay with the highest return.

2. Bet both since they both have positive expected return.

3. Bet 1 unit on the one with the highest return and 1/2 unit on the other .

What would you do?

Bet one unit, the same unit, on each.

You might consider a separate bankroll or subset of your bankroll to calculate a unique size unit, but stick with unit size, a flat percentage of starting bankroll, and don't readjust until that bankroll has increased nearly 50% or adjust down about 50%.

Because even two pick parlays are a bit more volatile, with slightly expected longer losing streaks and higher payouts, the bet size changing limit should be thought about a little closer.

But you get the idea. In the end, try to keep the risk flat on these for all bets. In my opinion, a form of fractional Kelly sounds attractive, but it has more downside than upside in this instance.

[Bsims]

I resolved my own dilemma by looking back at my history files for situations involving the two somewhat contradictory suggested wagers. I've documented the results in my latest blog post at ole44bill.blogspot.com. The bottom line is that I will now only wager on the parlays with the higher expected returns of the two.

[ChuckyTheGoat]

Keep bangin out the profits, Bsims.