Craps, and the Dont pass line odds. good?BAD?

Quote: skrbornevrymin

An odds bet will never help the player "win more/lose less." The reduction in the house edge comes because the house edge is a ratio of expected loss to the amount wagered. Just like increasing the denominator of a fraction reduces the amount that the fraction represents, the odds bet is in effect increasing the denominator of the ratio without changing the numerator. In other words, an even money bet, such as an odds bet, would not affect the expected value only the volitility of the results over many trials.

Also, reducing or limiting the number of trials, i.e. limiting the bankroll, does not change expected results. If we were to graph the results, each point on the graph would tend to vary more wildly when betting with odds, but the mean or average of all the points in each case would remain pretty much the same.

The only way to "win more/lose less" would be to quit at a point when the volitility is favorable, which is a temporary condition that only exists in the very short run, statistically speaking.

I believe I can demonstrate that you are wrong using Wincraps. It is a matter of putting the same amount of money against a smaller HE.

PS: I am a long time advocate of the idea that you can change the HE, but not the EV. So the stipulation revolves around the bettors doing just this, putting the same amount of money against a smaller HE. Thus EV also changes. The odds bettor makes fewer line bets, you see.

Quote: odiousgambit

I believe I can demonstrate that you are wrong using Wincraps. It is a matter of putting the same amount of money against a smaller HE.

Of course you can. Very easily. Good Luck with that.

That is not what both monkeys are doing. It really is an unfair comparison.
One monkey making $10 bets and the other making on average $50 bets.
Their total action is way different.
For those that say it is the same, they can live in their own world.

Now a race comparing the same amount of action should be a no-brainer since the odds layer has most of his $$$ on the odds.

OG, Continue and prove the point about who does win more/lose less with the same actual $$$s of action.

Quote: 7craps

That is not what both monkeys are doing. It really is an unfair comparison.
One monkey making $10 bets and the other making on average $50 bets.
Their total action is way different.

I agree. And "total action" is the expression I was looking for.

Quote: 7craps

OG, Continue and prove the point about who does win more/lose less with the same actual $$$s of action.

then I shall [g]

Quote: 7craps

Of course you can. Very easily. Good Luck with that.

That is not what both monkeys are doing. It really is an unfair comparison.
One monkey making $10 bets and the other making on average $50 bets.
Their total action is way different.
For those that say it is the same, they can live in their own world.

Now a race comparing the same amount of action should be a no-brainer since the odds layer has most of his $$$ on the odds.

OG, Continue and prove the point about who does win more/lose less with the same actual $$$s of action.

Multiply the amount bet by the house edge in each case and you will end up equal. The decrease in the amount of the house edge exactly mirrors the increase in the amount bet because you are increasing only the odds (even money) portion of the bet.

Quote: skrbornevrymin

The decrease in the amount of the house edge exactly mirrors the increase in the amount bet because you are increasing only the odds (even money) portion of the bet.

But the increase in the amount bet limits the action on the line bet, since you run out of money. I have put this into action myself, as when I bet 10x odds I compensate by making fewer line bets than at 3x4x5x. The total action I keep the same [edits]

Here is my math.
The no odds $10 player makes 100 such bets to have a total action of $1000

$1000 action
$10 bets
100 wagers
edge = -3/220 (I use a push as a decision here)
expected loss: -13.636


A little more math for the odds layer
$1000 action
$10 flat bets
345X lay odds (6x)

$50 average bets
(1/3*$10 + 2/3*$70
10/3 + 140/3 = 150/3)

20 wagers subject to the HE of -3/220
$200 * -3/220
expected loss: -2.727


see if your sims match

Quote: odiousgambit

But the increase in the amount bet limits the action on the line bet, since you run out of money. I have put this into action myself, as when I bet 10x odds I compensate by making fewer line bets than at 3x4x5x. The total action I keep the same [edits]

In the original example, there was only one pass line bet and in one of the two cases an odds bet. If you start mixing and matching bets its too hard to make a direct comparison between two styles of play - although you could if you were consistent with a betting formula.

Quote: 7craps

Here is my math.
The no odds $10 player makes 100 such bets to have a total action of $1000

$1000 action
$10 bets
100 wagers
edge = -3/220 (I use a push as a decision here)
expected loss: -13.636


A little more math for the odds layer
$1000 action
$10 flat bets
345X lay odds (6x)

$50 average bets
(1/3*$10 + 2/3*$70
10/3 + 140/3 = 150/3)

20 wagers subject to the HE of -3/220
$200 * -3/220
expected loss: -2.727


see if your sims match

If all your bets were pure single line bets, this would be true, but since we are only adding an odds bet to one of the line bets, the house edge in that case would not be a constant between the two cases.

Quote: 7craps

A little more math for the odds layer
$1000 action
$10 flat bets
345X lay odds (6x)

$50 average bets
(1/3*$10 + 2/3*$70
10/3 + 140/3 = 150/3)

20 wagers subject to the HE of -3/220
$200 * -3/220
expected loss: -2.727


see if your sims match

for the odds layer put another way.
The 20 average bets all have a $10 flat bet.
1/3 of the time, on average, it is resolved on the cor (come out roll)
say 7 times.
The other 13 times, 2/3, we have full odds subject to 0% HE

Those 20 flat bets * $10 is $200 total action subject to the HE
The other $800 total action is always subject to 0% HE.

That makes it more clear to me

Quote: skrbornevrymin

If all your bets were pure single line bets, this would be true, but since we are only adding an odds bet to one of the line bets, the house edge in that case would not be a constant between the two cases.

Ah! Just the point!

We are not just adding an odds bet to what we are betting anyway, not in either case! The OP [latest topic] stipulated something different, and in Wincraps I am certainly simulating something different: total action the same! Repeating myself, I endeavor to do this when playing 10x odds vis a vis 3x4x5x.

Quote: odiousgambit

The OP [latest topic] stipulated something different,
and in Wincraps I am certainly simulating something different: total action the same!
Repeating myself, I endeavor to do this when playing 10x odds vis a vis 3x4x5x.

Yeah, comparing some bets to others the only fair way is by total action.

Many craps folks say it is by HE per roll, the lowest making for the best bet when comparing.
They look the other way and do not think about the EV.

We should know this to be absolutely wrong.

The Field 3X and Place 5,9 for example.
Place 5,9
HE per bet resolved: 4.00%
HE per roll: 1.11%

Field 3x pay
per roll and bet resolved: 2.78%

So the Place 5 or 9 is a way better bet, per HE roll, than a 3X Field.
But these bets resolve at completely different times, so...
Not by EV and total action

Make 100 $5 Field bets
Field EV = 100*$5*(-1/36) = -$13.89

Make 100 $5 Place5 or 9 resolved (total action here)
PL5,9 EV = 100*$5*(-1/25) = -$20.00

when comparing two or more bets against each other, you better better make it fair.

Most times, that is the ev from total action, not HE per roll or one average bet vs another.
Good Luck WinCraps

Quote: skrbornevrymin

Also, reducing or limiting the number of trials, i.e. limiting the bankroll, does not change expected results.
If we were to graph the results, each point on the graph would tend to vary more wildly when betting with odds, but the mean or average of all the points in each case would remain pretty much the same.

Well said.
and where one falls in that graph is up to chance.
Players have no control, or do they??

Here is what 90 don't pass bets look like
same ev, the center point of the graph,
but the odds layer can either win way more or lose way more than the no odds layer.

Some see that as more FUN and more entertaining than just flat betting.

Beautiful explanation, 7craps.

Especially showing the relationship between the TINY difference for 90 events from the edge compared to zero edge and the huge difference in the variance.

I think too many people think you are WAY more likely to win just because the edge is lower.

It's not a big deal like I think a lot of people think. Certainly not during 90 events!

Quote: 7craps

The EV is the same but not the sd.

Quote: mrclean

Quote: Beethoven9th

I think you missed his point. He's saying that he'd lose less than someone who plays the Don't without laying odds.

Two well trained monkeys go up to a craps table with $1000. The table is $10 min with 3-4-5 times odds.

One has been trained to always only bet the $10 minimum on the don't pass. No odds.

The other has been trained to always bet a $10 minimum on the don't pass with max odds.

At the end of a few hours of play - who walks away with more money?

A few hours isn't really enough to decide anything, but their EV is the same, so, in the long term, they walk away with the same amount of money.

It's not really a fair comparison, though, since they are risking very different amounts of money. One monkey only has $10 in play for each point, while the other one has $10 when the resolution happens on the come-out roll and $70 when there is a point established! So, #2 is gambling a lot more than #1 but losing the same amount.

A more fair comparison would be that Monkey #1 bets $50 on the don't, while monkey #2 bets $10 on the don't with $60 in odds whenever a point is established. In this case, they are both risking similar amounts, but #1 will lose 5x as much as #2.

In other words, if you only want to bet the table minimum, then don't take odds. But if you want more action than just the table minimum, you are better off getting that action through odds rather than through higher line bets.

Remember that house edge does not determine how much you will lose. The house edge multiplied by the amount you bet will determine how much you lose. The 2nd factor is just as important as the first.

I almost forgot to post this. Used the awesome auto-bet file someone made and named "18655 games". Specified total action of $1000 per game and ran 50,000 games for each situation, one player taking odds, and one not, $10 bet on Pass , action the same

Yes, I know this is not the situation the OP of the recent thread was outlining, but does address a statement some make about free odds "not helping you win more/lose less" in the face of the critical stipulation that Total Action remain the same.

_


_

Quote: odiousgambit

I almost forgot to post this. Used the awesome auto-bet file someone made and named "18655 games". Specified total action of $1000 per game and ran 50,000 games for each situation, one player taking odds, and one not, $10 bet on Pass , action the same

Nice work.
The riskofruin auto-bet file can also be used after the pass line is changed to the don't pass

The difference in my sim results, for 345x odds,
and yours could be the overshoot on $1000 total action.
Many times my last bet did not have max odds but still had correct odds.
That caused the total action to be slightly more than $1k.


Results of 1 million session sim
parameters
$1000 total action
100 wagers
$10 Don't Pass bets
No odds
Pushes not counted towards total action

Bankroll decreased . . = 51.608% of the time
Bankroll increased . . = 40.466% of the time
Avg (mean) end bankroll = 98.60 (change of -1.40)
Broke-even: 7.926% of the time

Results of 1 million session sim
parameters
$1000 total action
$10 Don't Pass bets
345X odds

Bankroll decreased . . = 49.193% of the time
Bankroll increased . . = 49.045% of the time
Avg (mean) end bankroll = 99.71 (change of -0.29)
Broke-even: 1.762% of the time

Note the final bankroll mean and the HE.
right there