Actually, I doesn't matters whether or not BR1 can see the double-down card. In all cases, BR1's best play is to follow with a double-down. I believe that this is the FREE HIT (DOUBLE-DOWN in this case) situation. This is the situation where winning a single bet will do BR1 no good. If BR2 loses to the dealer (via BR2 bust or dealer making a better hand), then it does not matter what BR1 does. BR1 will win. If BR2 beats the dealer, then it does matter what BR1 does. If BR1 justs hits, then BR1 can not catch BR2. If BR2 doubles-down, then BR1 may catch BR2. BR1 DOUBLE-DOWN has no down-side, but has the potential to turn a possible loss into a possible win.
The BIG problem with online tourneys I was thinking about this on my morning run today and rembered a post a few weeks back where someone (could have been Barney or fgk42) mentioned that it was likely that someone will develop some software (if it's not already available) that will be able to tell you the statistically right play for every hand during a live game. This can't be pulled of in a casino for the obvious reasons but online it's not a problem. This post clearly illustrates the point. If I was on Global and had my "TBJ Playmaker v1.2" running in the background all I would have to do is input the other players and dealers cards, the bets made and the software could tell me what the optimum bet was and how to play the hand. We can see in the example that this post uses that the statistically "right" play is often not the intuitive one. When this software is developed (and if it's not already available it will be) online TBJ will become almost pointless because players who use "TBJ Playmaker v1.2" will have a massive advantage and I for one will not want to play against that sort of unfairness. Obviously it's not going to guarantee 100% win rate but on Global all you need to do is come 2nd 65% of the time to make a profit. If you come first 20% of the time you place in the top 2 the "in-the-money" rate comes down to 55% and if you can get 1st 40% of the time you're in the money it comes down to 50%!! What do you think? Is online TBJ ultimately doomed? How can we safeguard against this sort of thing? Cheers Reachy Ps. There is no such software as "TBJ Playmaker v1.2" but you've got to admit it'd be a good name!
Reach I was the one thinking about how online tournament could be taken over by advantage players teaming etc. Figuring the perfect play each and every time time after time. I think Ken's math example is just scratching the surface. He was using unlimited decks, a continuous shuffler I take it. The real problem would come from a counting program that would make Ken's math example even more exact of a deck with a discard pile. ( BTW the outcomes from the math example were close to negligible. You could be right doing the second best play almost as often as the first.) BUT, if you had a counting program running and a team member data entering and crunching numbers in mili seconds, or a little more you lol, could get the perfect basic strat play. For example in Ken's example what does an 11 mean when you figure all the cards. 8-3 7-4 9-2... with the dealer 10 and br1 10-6 8-8.... and so on. But lets place this "team" scenario with the games at blackjack21. If your team member was entering the data all along and figuring ratios of each card a perfect basic strat would evolve. If the dealer had 10 br1 had 10-6 and you had 9-2 the team member would enter these figures and the computer would easily calculate all the burned cards plus the new deal. This could change the numbers Ken and aralik came to and I mean change them big time. You could pull away from the negligible probability and maybe even get to a couple % advantage via basic strat. Barney
arlalik, I posted my numbers for hitting hard 9 just now, but after doing so I decided to double-check the hit-to-17 numbers against the dealer outcome chart. They didn't match, so something is out of whack on my end. I deleted the post with my erroneous results, and I'll post again after taking a closer look.
I found my problem... I was hitting to a HARD 18 or better only. In other words, if the first card was an Ace for soft 20, I was hitting again. CORRECTION TO RESULTS POSTED EARLIER IN THIS THREAD: Arlalik is right... With a hard 9, hitting to 18 or higher is the best strategy, yielding the 31.02% result. (Rounded from arlalik's 31.01). Hard 9 or Hard 11: Hit to 18. Hard 13 or Hard 17: Double. As a reference, here are complete numbers for hard 9, treating soft and hard totals as the same... Hit to 12: (Stiff: 41.42%, 17: 8.28%, 18: 8.28%, 19: 31.36%, 20: 8.28%, 21: 2.37%, Bust: 0.00%) Hit to 13: (Stiff: 35.69%, 17: 8.92%, 18: 8.92%, 19: 32.00%, 20: 8.92%, 21: 3.00%, Bust: 2.55%) Hit to 14: (Stiff: 28.82%, 17: 9.61%, 18: 9.61%, 19: 32.68%, 20: 9.61%, 21: 3.69%, Bust: 5.98%) Hit to 15: (Stiff: 20.69%, 17: 10.35%, 18: 10.35%, 19: 33.42%, 20: 10.35%, 21: 4.43%, Bust: 10.41%) Hit to 16: (Stiff: 11.14%, 17: 11.14%, 18: 11.14%, 19: 34.22%, 20: 11.14%, 21: 5.23%, Bust: 15.99%) Hit to 17: (Stiff: 0.00%, 17: 12.00%, 18: 12.00%, 19: 35.08%, 20: 12.00%, 21: 6.08%, Bust: 22.84%) Hit to 18: (Stiff: 0.00%, 17: 0.00%, 18: 12.92%, 19: 36.00%, 20: 12.92%, 21: 7.01%, Bust: 31.15%) Hit to 19: (Stiff: 0.00%, 17: 0.00%, 18: 0.00%, 19: 36.99%, 20: 13.92%, 21: 8.00%, Bust: 41.09%) Hit to 20: (Stiff: 0.00%, 17: 0.00%, 18: 0.00%, 19: 0.00%, 20: 16.76%, 21: 10.85%, Bust: 72.39%) Hit to 21: (Stiff: 0.00%, 17: 0.00%, 18: 0.00%, 19: 0.00%, 20: 0.00%, 21: 12.83%, Bust: 87.17%) On your other question... If BR2 makes 20 (or 21), BR1 should still stand with 16.At least, I think so! Thanks for the chance to find my error. Very helpful.
20 vs 16 Thank you Ken for double checking the numbers. This thread is taking us too far. Now about the case when BR2 = 20, BR1 = 16, Dealer = T (the same case but BR2 hit and stood). If BR1 Stand with 16, he advances only when dealer bust = 22.98% If BR1 hit once, he advances when: Dealer(bust)*BR1(17-21) = 22.98*5/13 = 8.84% Dealer(17)*BR1(18-21) = 12.07*4/13 = 3.71% Dealer(18)*BR1(19-21) = 12.07*3/13 = 2.79% Dealer(19)*BR1(20-21) = 12.07*2/13 = 1.86% Dealer(20)*BR1(20-21) = 37.07*2/13 = 5.70% Dealer(21)*100% = 3.74*100% = 3.74% Total..............................................= 26.64% (better than 22.98%). I checked hit to 18 and is worst than hit to 17. Now if BR2 = 20, Dealer = T, BR1 HIT TO 17 The second case when BR2 = 21, BR1 = 16, Dealer = T Stand = 22.98% Hit once = 20.34% (worst than stand).
There's our discrepancy. BR1 advances when the dealer busts (22.98%) but also when the dealer makes 21 (3.74%). The total of 26.72% is slightly better than the 26.64% obtained by hitting.
Little help here - please This helps when looking at the hard 11 scenario but maybe I'm just not getting this - I'm not used to looking at all those probabilities so I'll just ask in words. In the hard 9 & hard 11 cases DD by BR2 - BR2 CANNOT bust (That's a given) Forget about the outcome of BR2 vs dealer BECAUSE this forces BR1 to HIT - as previously mentioned. Now when BR1 has to double vs. Dealer 10 those are the odds that we should be concentrating on (26.72% per your figures) should we not? If BR2 doesn't double then the only way BR2 advances is BR2 beats dealer and BR1. So that probability = 26.64 that BR1 hits and advances. The probability that BR2 will beat the dealer & BR1 = ??? Arlick, thanks for your explainations. I know that this may seem "basic" to you but I look at this as a good learning exercise and appreciate you and Ken's time
Any more coming? Ok Ken, I admit I didn't get this and I don't want to beat a dead horse so I just accept the answers. How about another teaser? Maybe a weekly feature? Come on PLEASE
fgk42, I'll try to take your previous comments one by one... True, BR2 has non-bustable doubles here, but the percentages work better if he does NOT double. I admit this doesn't jive with intuition here, since you're forcing BR1 to double and he has a hard 16. Still, I'll try to explain why it's true in this case... Doubling means you get only one card. If you end up stiff, then BR2's double is a free shot. If he doesn't bust, he's a lock. If he does bust, he's still a big favorite because the dealer must bust as well for him to lose. That factor is the biggest contributor to this oddity in optimal strategy. Actually, BR2 doubling forces BR1 to double, not hit. Here you've taken a number out of context. (That's VERY easy to do in this confusing thread, with errors and corrections.) The 26.72% you mention is the chance that BR1 will advance after BR2 hit (not doubled) and ended up with exactly 20. BR1 should then stand with 16, and he has a 26.72% chance of advancing. That particular number became prominent in the discussion only because arlalik and I had a different result at one point and we wanted to resolve the discrepancy. Here you've also used a number (26.64) that ONLY applies when BR2 has hit and stood with exactly 20. Again, I think the multi-threaded conversation is the culprit here. I don't know if it's really worthwhile to even have you go back and try to assemble the thoughts of this thread into the separate concepts. It's been pretty tangled, and at the end of it, the decisions it produces are somewhat unusual. What you need, and what you requested, is more puzzlers of this sort, and particularly, one that's not chosen to be unusual. I'll post a more vanilla sort of question later today.