I don't pull money out unless I have a very good reason. For me, very good reasons have been buying an investment property, maxing out retirement accounts, and things of that sort.
A very good reason can be anything, from a hot young woman ($200?) for an hour to a diamond jewelry ($10,000) or vacation for the spouse if she accidentally found her nude picture on your phone. Lol. In the end, if you have a BR that allows you to play at your chosen level without ROR, then anything over can be yours to do as you please.Originally Posted by moo321
Nyne,
Sorry I have not addressed your comment sooner. I have been very busy and wanted to take a moment to digest what you said. My understanding of your statement is that RoR is impacted if you take money out of wins. I can understand how this applies if you take money out of your bank when your are under the amount of the bank roll that you used to calculate your RoR. I am having a hard time though understanding why it would impact your RoR if you are above the amount you based you RoR on. Can you help me to understand why this would be.
As I view things you RoR is based solely upon the risk of loosing you starting bankroll (Hopefully I am not looking at things way wrong) so if you are above that amount and betting the same your RoR would actually decrease. By taking money from winnings when you are above your bank would you not just be resetting back to your original RoR figures?
I'm not Nyne, but I think I might be able to answer the question.
While your RoR is dynamic, and does change when your BR grows or drops, when you calculate your RoR for your given time it figures in all the possible outcomes. So your RoR factors in the possibility that you will win more (and thus have a lower RoR if you employ the same strategy) and keep winning as your RoR approaches 0. Likewise, it factors in the possibility that you'll hit a down turn, have greater RoR as well as the possibility that you'll keep sustaining losses till you go bust.
What is probably the confusing part, is that we have so many RoR's floating around that it can get confusing which one we are referring to. An RoR for a given BR size and strategy takes into account all the conditional RoR's of the BR sizes that your bankroll might grow or drop to.
It is a 'strategy', in a sense, when you have a planned way out taking out from the bankroll (including any of the three ways which I mentioned, although I only endorsed one of them). You are right, that if you drop down to the BR level (through any means), then your RoR is the same as your starting RoR. However, the RoR that you would have calculated (which assumes full re-investment) for your starting figures is not the same as the RoR including this 'strategy'.
Here's an example. Keep in mind, all the actual numbers are just pulled out of my ass for illustration purposes. What the actual numbers are is not important.
Ex1. Bank roll is 1000, strategy is high-low with I18. You calculate your RoR to be 10% at the start, let's call this your RoR at t1 (time 1). Your bankroll goes up to 1200. Let's call the moment when your bankroll is 1200 t2. At t2 your RoR is lower than at t1. If you go back to exactly 1000. Then your RoR at t3 (the time at which you dropped back down) is the same as at t1.
However, the RoR calculated with the above assumes a full-reinvestment 'strategy'. If you are playing with a 'take-the-winnings' strategy, then when you start with 1000, your RoR at t1 is not 10%. It's actually higher. The reason is because you are eliminating the possibilities (the conditional RoRs) where you grow your BR and have decreased RoR if you just skim all the winnings. However, the possibilities where your BR shrinks and your RoR increase still exist.
Imagine if you had a bankroll for coin flipping (whether you have an edge or not is not important in this example, but you would need to know for RoR calculations). If you had 100 units, and said 'every time I win 1 unit, I'm going to take 1 unit out of my bankroll', what do you think will happen eventually?
You win 1 unit, your bankroll is 100 (100 + 1 - 1). You lose 1 unit, your bankroll is 99 (100 - 1). You win 1 unit again, your bankroll is 99 (99 + 1 -1).
Even if you only take out some of your winnings, this hurts your bankroll growth. There's nothing wrong with paying yourself, but you have to realize that there is a trade off between growth/risk and cashing out.
Last edited by NotEnoughHeat; 04-28-2015 at 02:11 PM. Reason: grammar & puncuation; word choice & clarity
The RoR was calculated using all ranges of possible results. You expect to play and experience variance in accordance with the probabilities. That includes 3+ SD up or down. If you spend the profits from great results that allowed you to endure poor results then you didn't actually have an accurate RoR to begin with. So you go up by 20% of your BR and then have a 110% downswing of your original BR but you spent that 20% you go bust but you would still have 10% of your BR left had you not touched your BR. This would add to your original RoR. By taking profits to reduce back to the original BR you keep your RoR the same as it was but it was never what you thought it was to begin with. It was actually much higher. I hope that is clear. This is the kind of thing I can explain a while and probably never get a better explanation for those that didn't get it to begin with.
Last edited by Three; 04-28-2015 at 02:58 PM.
You're right, Tthree.
I keep hearing that your ROR includes reinvestment of your winnings. I challenge that idea. If your ROR is, say, 1%, and your bankroll is $10,000, and you go on to win $5,000, what hapens if you spend the entire $5,000 and are back at $10,000. What is your ROR now? It's still 1% of course. Your ROR is based on your starting amount.
Now, is it wise to keep some of those winnings aside? Yes! Let's say your expected win rate accounted for $1,000 of that $5,000 gain. In that case, $4,000 was positive variance. You are just as likely to have $4,000 of negative variance in the future. No, I am not saying that the money will even out-- it's the same as the law of large numbers. What I am saying is, the variance is so steep at blackjack that it's alway wise to save some of your winnings for those rainy days. This in effect decreases your risk of ruin below the 1%, if it were properly calculated in the first place.
Oh, how I have wished I had not spent some of those positive variance winnings in the past! Bad times come, and if your ROR is 1%, the worst of bad times comes every hundred starts (in all probability). All I can say is, when you are up $5,000 and you spend it all, you'd better pray you don't have a downward streak of $6,000, because now you're playing with a totally insufficient bankroll of $4,000 and you are ever more likely to go bust if you don't replenish. The only thing you can count on in the very long run is your win rate. Everything else is smoke and mirrors. With that in mind, if nothing else, spend no more than your win rate, for that money is truly yours (in all probability in the very long run if you're fortunate enough to live that long.lol).
Last edited by Aslan; 04-29-2015 at 04:12 AM.
Aslan 11/1/90 - 6/15/10 Stormy 1/22/95 - 8/23/10... “Life’s most urgent question is: what are you doing for others?” — Martin Luther King, Jr.
My posts already addressed this, but I will challenge your challenge! The issue is, that either the 1% already figures in that you will withdraw from your bankroll in a certain way or it was never 1% in the first place! A standard RoR already takes into account what I referred to as the 'conditional RoR's' of possible future bankroll levels. The potential for your bankroll to grow, uninhibited, and reduce in RoR is part of the original (standard) RoR. Taking money out during wins and nothing out during losses is like a reverse loss rebate. We have already seen how powerful this concept is, just look at Don Johnson. What's the harm in the casino giving back some of it's winnings?
For any other details of what I'm trying to get across, I refer you to my poor-accessibility wall-o-text. Dang, I'm getting too hooked on smiley's, people are going to take me less seriously...
Last edited by NotEnoughHeat; 04-29-2015 at 01:12 PM. Reason: Added sentence starting with 'The Potential', replaced 'typical' with 'standard'
I've just been lurking in this thread and I somewhat follow your reasoning here but I'm not sure I agree. I'd really like DSchles to weigh in on this.My posts already addressed this, but I will challenge your challenge! The issue is, that either the 1% already figures in that you will withdraw from your bankroll in a certain way or it was never 1% in the first place! A standard RoR already takes into account what I referred to as the 'conditional RoR's' of possible future bankroll levels. The potential for your bankroll to grow, uninhibited, and reduce in RoR is part of the original (standard) RoR. Taking money out during wins and nothing out during losses is like a reverse loss rebate. We have already seen how powerful this concept is, just look at Don Johnson. What's the harm in the casino giving back some of it's winnings
For any other details of what I'm trying to get across, I refer you to my poor-accessibility wall-o-text
This is my argument. If you start with a given bankroll and ROR, if you add to that BR the ROR will decrease. It doesn't matter whether you add funds from play or other sources, you now have a larger BR and lower ROR. Your money and your ROR calculator don't know how the bankroll increased.
Similarly, if you reduce the BR the ROR will increase and it doesn't matter whether you reduce the BR from play or take the funds out. The funds are gone either way and the ROR increases.
So if it happens that you remove funds or have losses after building your initial bankroll bringing your BR back to its beginning size what will happen is the ROR will return to its initial figure.
This is the stance I have been operating on and how I was planning on taking advantage of my winnings when I decide I want to cash out wining. To simplify I just thought your RoR was based upon your starting bankroll (and all other factors used to calculate at the time you did) and thus was a fixed point at a specific point in time. If you played a session and won some money your RoR was now a bit lower when all other variables remained the same. If you lost a little your RoR was a little higher. Thus I always recalculated me bet ramp in cases where I lost big or won big. If I broke even nothing changed.
So I figured... if I win and take those winning out... it is no different than if I had not played at all. I am now struggling with this because my simple mind tells me that that If I were a new playing with a 10k bank paying for the first time my RoR is what I calculate it to be.
However I can understand and see how looking at that looking at it in terms of a session isn't really appropriate because we are not dealing with winning sessions or loosing sessions. Rather we are just dealing with the chance of loosing your entire bankroll. Am I correct that the RoR is based on an infinite time frame? and is that why it is said that all wins / losses are factored in. Thus that is often why people look to doubling their bankroll... because once it has doubled you can no longer loose your starting bank?
Yes, it would definitely help clear up the issue. I haven't seen senior members this divided for quite sometime.
Although I agree the RoR calculator doesn't know how the bankroll increased and, all other things equal, it will give the same RoR for 1200 no matter how no matter how it got there. The key qualification however is 'all other things equal'. I think it does matter how the bankroll increased. If you are taking money out of your bankroll, you are playing with a different 'strategy' (let's call this BR strategy, so it doesn't get confused with playing or betting strategy) than with a full re-investment.
Say we have two RoR's, one for a particular playing/betting strategy, with 1000, and full reinvestment; the other for the same playing/betting strategy, also with 1000, but money taken out on win. It's your contention that these RoR's are one and the same and it's mine that they are distinct. I say that they are different, because the likelihood for the bank to grow to 1200 (or any other amount) is different. In the first case, you would have to look at the possible ways to grow an additional 200. In the second case, you have to look at the possible ways to grow an additional 200 plus whatever you would have taken out (this will vary depending on how and when you withdraw from your bankroll).
In both cases the possible routes to 800 are the same number of ways to go down 200 (since by example, the second strategy only has win contingencies, although I believe that this reasoning applies across the board).
Although assuming an advantage there will be an asymmetry between going up 200 and down 200, in the first case there is a symmetry between being 200 above expectation and 200 below expectation. The second case does not have this symmetry. If you take a flat amount out, all the above expectation gains to your bankroll will be 'shifted' down in comparison to below expectation losses. It will be just as likely for you to 200 below expectation as 300 above expectation, less a 100 withdraw. The same applies if you take out money on the percentage above expectation instead--the 'shift' will be less of a linear shift and based on percentages instead (I'm sure a math buff could supply us with the type of function).
I don't think you even have to be explicitly systematic about your withdrawals for this to apply. I think all that I have said applies to randomly taking out X dollars at random win Y, different functions and asymmetries but the same effect.
I don't think that looking at sessions is inappropriate, it's just important to look at them the right way. What matters to RoR is not happenstance but the distribution of outcomes.
From what I understand, that is correct. I actually started a thread on this very topic some time ago. I will try to find it. I remember Don pitching in. I guess although it isn't completely accurate to say that you can't lose your starting bank (assuming you're doing the exact same thing) once you double it, effectively that's the case. From what I recall, the RoR's between doubling your bank and an infinite time frame are for all intents and purposes, the same.
EDIT: here's the link http://www.blackjacktheforum.com/sho...ight=risk+ruin, there's also some other useful threads if you search for 'risk of ruin'
Last edited by NotEnoughHeat; 04-29-2015 at 02:38 PM.
For any particular playing and betting strategy, there are essentially infinite possible paths your bankroll growth may take if you play for an infinitely long time. These paths each have their various ups and downs. If your starting bank is 1000 units, your risk of ruin is simply the percentage of all possible paths that at some point drop 1000 units below the starting point. Any money you remove from the bankroll shifts that path downward from where it would otherwise be, so if you graphed that path, there would be a sharp drop at each withdrawal. Ordinarily, if you were up, say, 1000 units, then to go bust your bankroll path would have to drop 2000 units from that point. If you had removed 500 units, then you'd now only have to experience a 1500 unit loss to go bust, rather than 2000, which makes going bust much more likely. Any scheme that plans to removes money from your bankroll will move some of those paths from having a downturn that nearly busts you to actually busting you out. Different schemes will push different numbers of paths into bust territory. The standard formulas don't account for any withdrawals from your winnings, so they assume your 1500 unit drawdown after you doubled your bank means you're still in the game, not busted out. If you aren't fully reinvesting your winnings to pad your bankroll for possible future drawdowns, the RoR you calculate is wrong for your actual style of play. In particular, if you withdraw all profits above some certain threshold, then your risk of ruin approaches 100%. Large drawdowns of 1000+ units are rare, but if you play long enough, they happen, and removing all winnings above that point will ensure that when it does happen, you won't have your previous winnings to keep you in the game.
With that said, the optimal way to bet is not to simply set your betting ramp and play that way forever. As your bankroll grows to multiples of your starting bank, you should be increasing your bets. This also makes the calculated RoR numbers wrong. The mathematically optimal way to grow your bankroll the fastest is to use Kelly betting, however it's impossible to use perfectly in the real world. The swings associated with perfect Kelly betting are also hard for most people to stomach. As a bit of a compromise, we use fractional Kelly betting. I used 2/3 Kelly when I was starting out, and as my bankroll grew, I gradually reduced that to 1/2 Kelly. This was when I still had a job. I now use 0.3 Kelly. Kelly betting theory lets us calculate the Certainty Equivalent of our play. This is essentially what our play is worth if we were to trade our uncertain series of wins and losses for a certain payment. If you withdraw more than the CE of your play, you will eventually go bust, just as you would if you bet in excess of 2x Kelly. I suggest keeping withdrawals below 1/2 of your CE, which, if your bets are sized properly, will be 1/4 of your EV. This is a fairly safe level, although you still will experience more severe drawdowns than if you weren't taking money out, which is why I suggest waiting until you have room to decrease your bets a couple of times if necessary. You could relax this a little if you have the ability and willingness to put some money back into your bankroll from outside sources if you experience a bad run.
Eventually, Kelly becomes irrelevant too. You eventually get to a point like KJ is at, where you can easily afford to bet more, but you choose not to because of other concerns, like staying below a particular betting level due to heat concerns, or you are limited by table maximums, CTR threshold, limited bet size options, or any of a number of other issues, but to the extent possible, Kelly or fractional Kelly is optimal.
Posting Permissions
-*-
Reply With Quote
Bookmarks