Loss Rebate Very basic Strategy and some questions

[notapro]

What's your best guess for bankroll needs to play while maximizing EV
Just a rough guess, as to units needed and win/loss points.

Assuming 6D LS S17, DAS DA2 with 20 % rebate; all conditions similar to what Dr. Eliot Jacobson posted on APHeat.
Quit after winning 24 Units, and quit when down 26 units. Average hands played will be 481. EV will be 1.25 Units

So what will be the units of bankroll needed for this? I mean its about 50/50 win or lose about 25 unites for the day, and if lose then a rebate of 5 units, so my layman's guess is 100 units needed. The detailed Excel spreadsheet only provides for session ROR, and i cant for the life of me figure out how to apply the data. It could be 1000 units needed, for all I know.

Then for Roulette same thing:
20% rebate, single zero, just the roughest of guesses as to win/loss stop and how many hands to get there. 2.7 % house edge. 141 rounds, 3.4 units won but Excel says 2080 units need for ROR of under 5 %. Is this right? then this means that much more than a 100 unit bankroll is needed for 20 % loss rebate blackjack? I'm very confused.

Also craps has high Standard Deviation when it offers 3x to 5x odds, and a lower house edge, what do u think units of bankroll needed for 20 % loss rebate would be?

Thanks for reading this, and good luck

[notapro]

Bumping for visibility because i posted on the weekend

[Gramazeka]

Where do you play this game? Tell us a little about yourself...

[notapro]

Where do you play this game? Tell us a little about yourself...

How many units do you think i would need?

[G Man]

Kind of funny. I have the spreadsheet from Eliot named "The Loss Rebate Theorem full solution" and don't get the same numbers for the game you describe. For the described BJ game, I have a Win Quit Point of 21 units and a Loss Quit Point of 23 units,
The EV is 1.09 unit per session. For 1000 units the RoR is 1%, 500 units is 10% and 100 units (like you mention) is 64%

For the Single Zero Roulette game playing one number, I get a Win Quit Point of 68 and a Loss Quit Point of 73. The expected rounds to Quit Point being 145 and the EV is 3.5 units per session. A 1000 units bankroll gives a RoR of 25%, 2000 units is 6% and 3000 units 1.5%

Hope this helps.

[notapro]

Kind of funny. I have the spreadsheet from Eliot named "The Loss Rebate Theorem full solution" and don't get the same numbers for the game you describe. For the described BJ game, I have a Win Quit Point of 21 units and a Loss Quit Point of 23 units,
The EV is 1.09 unit per session. For 1000 units the RoR is 1%, 500 units is 10% and 100 units (like you mention) is 64%

For the Single Zero Roulette game playing one number, I get a Win Quit Point of 68 and a Loss Quit Point of 73. The expected rounds to Quit Point being 145 and the EV is 3.5 units per session. A 1000 units bankroll gives a RoR of 25%, 2000 units is 6% and 3000 units 1.5%

Hope this helps.

Oh it helps a lot, the differences are probably because I played around with the input numbers, and I prob made a small mistake with them.

Im just asking because Im shocked that the units needed are so high, I just assumed that it would be like a 100 units needed for blackjack.
Like if a hundred unit loss for one session, twenty gets returned. The next day its 50/50 to win and again another huge loss is another 20 units returned.
It just seems like 500 units is not going to be much less than 10 % Ror

Thanks for your help again and if you have time, please tell me how I made such an intuitive mistake. The numbers have to be wrong, right?

[G Man]

Oh it helps a lot, the differences are probably because I played around with the input numbers, and I prob made a small mistake with them.

Im just asking because Im shocked that the units needed are so high, I just assumed that it would be like a 100 units needed for blackjack.
Like if a hundred unit loss for one session, twenty gets returned. The next day its 50/50 to win and again another huge loss is another 20 units returned.
It just seems like 500 units is not going to be much less than 10 % Ror

Thanks for your help again and if you have time, please tell me how I made such an intuitive mistake. The numbers have to be wrong, right?

Well... You will not have a 100 unit loss in a single session since your Loss Quit Point is 23 units. Then you will only get 4.6 units back from the loss rebate, making your new bankroll 81.6 units. Remember that flat betting this game without the loss rebate has a 100% RoR for any number of units.

[notapro]

Screenshot (305).jpg

Well... You will not have a 100 unit loss in a single session since your Loss Quit Point is 23 units. Then you will only get 4.6 units back from the loss rebate, making your new bankroll 81.6 units. Remember that flat betting this game without the loss rebate has a 100% RoR for any number of units.

Ok, but the numbers get crazy when its a 95% rebate. I mean the ROR should be way less, like almost 0 % ROR instead its still super high, this leads me to at least consider that the graph is perhaps meaning a session ruin? rather than a ruined bankroll?

Idk, thanks again for taking the time, I really appreciate, I hope im not appearing to criticize your answers because I'm not meaning to.

[G Man]

Take a look at what's going on with other numbers...(Win Quit Point and Loss Quit Point).
You expect to play 562 hours before quitting, that's certainly why you RoR gets higher.
The RoR you see is based on 3 variables:
Bankroll of 1000 units
EV per session of 53.486
Standard Deviation per session of 374.669
Make the calculation and you get that 46.67% RoR

[notapro]

"Make the calculation and you get that 46.67% RoR"

I know, that's what I mean: the stop loss is around 400 units so when that happens around 380 units get refunded.
So to lose 1000 units would take 50 losing sessions in a row, which is effectively impossible, so I took this as a ROR of zero, meaning these numbers just aren't right.
I like you a lot, and thank you for helping me understand this

[G Man]

They don't have to be in a row, remember that you will still lose on these sessions and they will happen 60% of the time.

Maybe Eliot Jacobson could clear this up... ?

[Eliot]

Kind of funny. I have the spreadsheet from Eliot named "The Loss Rebate Theorem full solution" and don't get the same numbers for the game you describe. For the described BJ game, I have a Win Quit Point of 21 units and a Loss Quit Point of 23 units,
The EV is 1.09 unit per session. For 1000 units the RoR is 1%, 500 units is 10% and 100 units (like you mention) is 64%

For the Single Zero Roulette game playing one number, I get a Win Quit Point of 68 and a Loss Quit Point of 73. The expected rounds to Quit Point being 145 and the EV is 3.5 units per session. A 1000 units bankroll gives a RoR of 25%, 2000 units is 6% and 3000 units 1.5%

Hope this helps.

You are exactly correct in your use and interpretation of my spreadsheet. We are assuming the person plays this loss rebate multiple times, hence the need for a large bankroll.

You are also right about the increasing rebate % weirdness. The higher the rebate %, the higher the quit points, the longer you'll play and the more bankroll you will need to play. But the more you play, the higher the chance that you hit the bottom loss-quit rather than the win-quit.

It's also important to remember that this uses the CLT, which works great for symmetric distributions over lots of trials, but is not so good over short durations and very skewed distributions. That's why these theorems are not the best trick to use for something like video poker, with potentially huge payouts, though the results are still in the ballpark. Chances are you will blow way past your quit-win on your winning days. For some games, simulations work best if you want to be super accurate.

[G Man]

Thanks for the explanation!