Theoretical ROI calculation on middles

[ChuckyTheGoat]

When u try to middle a game, it should in theory be pretty feasible, to calculate your expected return. When you calculate your Return on Risk, it's expected profit/total $ risked.

For the denominator calc, do you use total $ amt in play? Or maximum $ loss?

So if you figure your expected return to be +$5 from two $110 risks (at -110 each), is your ROI:

a) +5/220, or

b) +5/10?

Thx and GL.

[Machiavelli]

I think the more pertainent number is (a) +5/220.

If you had a no juice middle, for example -7 -105 and +7.5 +105, the ROI would be infinite using method (b).

[ChuckyTheGoat]

OK, mach, thx. Makes sense.

[count zero]

Don't know anything about ROI, but what would be wrong with having an infinite ROI? If risk literally equals zero, wouldn't that be appropriate?

Also, if you use a), you get oddities like a straight bet being an order of magnitude better ROI-wise than a middle on 3. If I understand concept a) correctly, ROI for middling -2.5 / +3.5 @ -110 on each side would be (.1 * 2) - (.9 * .1) = .2 - .09 = average win of .11 units per 2.2 units in play, so by hypothesis a) ROI = .11 / 2.2 = .05. By comparison, a straight bet by a 57% bettor would have an ROI of .97. But surely a straight bet at 57% is not 20 times better than a middle on 3.

[buckeye]

I've always thought of it as A. for scalps ( small ROI with zero risk ) and B. on middles ( potentially large profit margin when compared to actual risk if you can get a -2'/+3' ). ROI is not really how I evaluate/look at middles because scalps have no "gamble" in them whereas middles do - barring any "lead" considerations. So I see the "moved $" as important to the scalp ROI and the ROI is always positive on each attempt. But not so for the middle since there is risk and a particular middle is highly likely to have negative ROI but there is +EV long term and a long term profit expectation. But I am no expert and that is just the way I look at it. Mach's example is a combination "almost" scalp ( 0 cent ) and side, so it is a bit of a grey area, admittedly.

cz,
I'd calculate the 57% as:
.57*2 - (.43 * 2.2)= .194 which is a little under 4 times the middle ratio you calculated, not sure where you got .97 but it is only fair to compare the same 2.2 units bet between the scenarios! I do look at middles like B., so 50% vs 19.4% is my comparison.

I'd think because the actual risk on a middle is .1/2.2 vs 2.2/2.2 on a straight bet you'd want to bet more UNITS per middle attempt ( or a bigger % of bankroll ) than you would on straight bets. I know that mine tend to be 10-40 times as big as my straight bets but I don't use Kelly so I don't know what it would dictate! I also VERY RARELY have .1/2.2 at risk, usually much less and correspondingly less than a 10% hit expectation on a clean middle!

JMHO

[ChuckyTheGoat]

Thx, buckeye. As you state, a true ROI calculation is probably not a good measure for these risks.

[count zero]

Good analysis, buckeye. On the diff in our results, I think you only calculated the EV, didn't you? I got EV = .194 as per your calcs, but then to get the ROI you have to divide by the risk amount, which for a straight bet of 2 units would be .2, so .194 / .2 = .97.

[PMcK]

<< Also, if you use a), you get oddities like a straight bet being an order of magnitude better ROI-wise than a middle on 3. If I understand concept a) correctly, ROI for middling -2.5 / +3.5 @ -110 on each side would be (.1 * 2) - (.9 * .1) = .2 - .09 = average win of .11 units per 2.2 units in play, so by hypothesis a) ROI = .11 / 2.2 = .05. By comparison, a straight bet by a 57% bettor would have an ROI of .97. But surely a straight bet at 57% is not 20 times better than a middle on 3. >>



You can calculate the ‘expected’ ROIs for straight wagers using: ROI=(decimal line)*probability - 1. A 57% bettor would have (1+10/11)*.57-1=.088 ROI (equal to 0.194 units won/2.2 invested).

If the bet at -2.5 is the 57% bettor’s ‘normal’ bet risking 1.1 units and later he sees +3.5 and bets again 1.1 units, his expected profit without the middle attempt would be 1.1*.088=0.097 units and with the 10% chance for a middle 2.2*.05=0.110 units. So the middling would be more profitable despite the lower ROI if in practice the bettor ‘invests’ more. On the other hand the expected ROI of the first wager might be higher than normally if the line moves from -2.5 to -3.5.

[buckeye]

<< I got EV = .194 as per your calcs, but then to get the ROI you have to divide by the risk amount, which for a straight bet of 2 units would be .2, so .194 / .2 = .97. >>



cz,
I'd say the risk on a 2.2 straight bet is 2.2, not .2! The risk on a middle of 1.1 each side is .1 because you can't lose both bets and the worst result is -.1. With a straight, even at 57% expectation, each bet "risks" the entire amount as 43% of the time you will be -2.2. So a ROI expectation is .194/2.2=8.8% (as PM stated ) compared to the "profit expectation"(not a true ROI) of 50% for that "3" middle. BTW, now that AG is gone those aren't that easy to find outside maybe taking leads or using locals - but that muddies the waters.

GL

[count zero]

Oops, brainlock! Sorry about that.

[sea biscuit]

ChuckeyTheGoat,
How do you convert decimal lines to number lines? For instance, converting .9 or -.9 to a number.Also, how do you convert a number to a decimal?
Thank you.

[buckeye]

seabiscuit,
I'm not sure what you were asking Chucky but if it is how to convert $lines to decimals and vice-versa I think the easiest way is to divide 1 by the absolute value of the number ( dividing by 100 where necessary ) and add the minus sign and and multiply by 100 where appropriate on the reverse. 1/.90909=1.10, so -110. 1/1.10=.90909.. Does that help any? Most use the decimal multiplier of 1.90909 instead of just .90909. I also think most players use charts to convert since they are so easy to find anymore!

GL

[maria sharapova]

I could explain it,but this link makes it easier,and i'm lazy.
Text
ciao from maria,linkster

[sea biscuit]

Thanks Buckeye.
Seabiscuit

[sea biscuit]

Thanks Maria, Linkster
Seabiscuit