CVData Ties

[SleightOfHand]

How is the tie % calculated in CVData? Is it based on the total # of ties vs total hands? Or is it based on the total $ pushed vs total $ played? I would think there is a difference when counting ties in regards to splits and doubles.

 

[QFIT]

Hands tied out of total hands. Split hands are counted independently.

 

[Kasi]

Hands tied out of total hands. Split hands are counted independently.

Don't know why this surprised me a little lol. I think I thought CVDATA (don't have it) probably would have counted w/l/t/ %'s based on "rounds" played. And therefore maybe counted a round with 2 split hands when one loses and one wins, as a "round" with a "net push" and put it in the "tie" column. Are you saying you'd count the winning split "hand" in the win % and the losing split "hand" in the "loss" %?

Somehow I think I thought maybe CVCX (don't have that either) treated splits this way.

Would you say when alot of people cite w/l/t/ %'s, they express them in terms of "rounds" and include "net pushes" in the tie column? Would you say that is what the Wiz does? And what most people do when figuring HA since that seems to generally be based on initial bet/round?

No big deal - it just struck me as maybe a little inconsistent in how I thought maybe CVDATA figured out alot of other stuff - like maybe Stan Dev for example that I thought assumed a "net win/loss/rd"? Are some outputs using different underlying assumptions than other outputs?

Not even sure what I'm saying, and never have really cared that much about w/l/t %'s anyway. So, as usual, I don't even know what my point is anyway lol.

 

[QFIT]

Many stats are hand based and many round based. Advantage is calculated both ways. Many standard deviations are calculated. It depends on your needs. When calculating overall risk and betting, round data is required. When looking at bonuses, playing decisions, etc, we need to look at hands both independently as well as pre-split. Suppose you play three spots at once, for risk analysis you must look at all three as one bet. But you wouldn't look at w/l/t percentages based on rounds. The more data the better.

 

[Kasi]

Many stats are hand based and many round based. Advantage is calculated both ways. Many standard deviations are calculated. It depends on your needs. When calculating overall risk and betting, round data is required. When looking at bonuses, playing decisions, etc, we need to look at hands both independently as well as pre-split. Suppose you play three spots at once, for risk analysis you must look at all three as one bet. But you wouldn't look at w/l/t percentages based on rounds. The more data the better.

Thanks QFIT. I think I see what you mean. And as long as the user knows the definition of what's included or not in w/l/t %'s, fine.

Is there a reason Sleight had to ask what's included? I assume he merely missed something in the documentation? I only ask because Sleight does not seem to me like the type of guy who would ask such a question without trying to figure it out first.

You are probably right in that I have always been more focused on knowing, best as I can, overall risk and the best way to bet and therefore have worried more about "rds" than "hds".

I guess the more data the better but w/l/t% is way way down my list in the first place lol.

What would you surmise is the usefulness of such a stat anyway?

 

[SleightOfHand]

Thanks QFIT. I think I see what you mean. And as long as the user knows the definition of what's included or not in w/l/t %'s, fine.

Is there a reason Sleight had to ask what's included? I assume he merely missed something in the documentation? I only ask because Sleight does not seem to me like the type of guy who would ask such a question without trying to figure it out first.

You are probably right in that I have always been more focused on knowing, best as I can, overall risk and the best way to bet and therefore have worried more about "rds" than "hds".

I guess the more data the better but w/l/t% is way way down my list in the first place lol.

What would you surmise is the usefulness of such a stat anyway?

I was asking because I was simming my strategy to find the frequency of ties for a side bet, and wanted more precise knowledge as to how it was derived, since the help menu just said that it was tie % and not any more specific

 

[Kasi]

I was asking because I was simming my strategy to find the frequency of ties for a side bet, and wanted more precise knowledge as to how it was derived, since the help menu just said that it was tie % and not any more specific

Thx Sleight. Makes complete sense to me now lol.

Good luck on the side-bet!

 

[QFIT]

In general it's not a particularly useful feature. But, you never can tell what will be useful.

 

[SleightOfHand]

Buh?

While talking to another member on this forum about the side bet, we came across the situation where if we removed the insurance option, the probability of ties of would decrease. Does winning an insurance bet count as a tie? Comparing the push probabilities relative to the insurance index, it seems that the major discrepancies start slightly below the index, which I suppose is due to other players showing their cards after the insurance bet is taken, but nonetheless, doesn't seem to me like it should be considered a tie, but simply a won side bet. Are there other factors that affect the side bet as well, or is there a mistake on my behalf?

 

[QFIT]

All bets in a hand are counted when determining won, lost, pushed. All bets spring from your initial bet. What matters is how much you win relative to the initial bet.

 

[Automatic Monkey]

All bets in a hand are counted when determining won, lost, pushed. All bets spring from your initial bet. What matters is how much you win relative to the initial bet.

Ah, OK. So if you split a hand, win one half and lose the other, that's treated as a push.

 

[QFIT]

Ahh, sorry I misspoke. Split hands are kept separately in the won/lost tables. However, in the Hands tables, yes they are combined. It all comes down to how I figured each set of data was most likely to be used. It would require less cpu time and code to track win/lost/push by round instead of hand. But, this data by hand seemed more likely to have a purpose. Although frankly, the percentage of hands won is not usually a particularly useful stat.

 

[SleightOfHand]

Yikes!

So, this means a push occurs >9% of the time at zen TC+21, not 4! It would depend on indeces, decks, pen, etc, but does that sound right?

 

[QFIT]

Pushes are certainly common at high counts as so many hands are 20.

 

[SleightOfHand]

Lol I meant that a TC of +21 seems a little high, although I am not one to make any judgements on the statistics of the game

PS: AM go to chat if you are available!

 

[QFIT]

TC has less effect on wins/losses than people think. One of the reasons I've never found the stat of much use.

 

[SleightOfHand]

Another discrepancy

I forgot, we also noticed that there was a difference between the index defined by our custom side bets and the push rate in the percents table. Turning off insurance has definately improved that discrepancy, but there is still a bit of inconsistency. My sim was to make 5 sections of the side bet where we get paid off with a dealer and player hand of 17, 18, 19, 20, 21. First two cards is reset/off, split/dd hands count, everything else being open