Hold Percentages & New Game Analysis

On another thread there was a discussion of the importance of the hold statistics as it related to evaluation of new games and existing games in the pit. I have heard/read that it is preferred that new games look to hit the 20% to 26% range in hold percentages.

The hold seems simple enough as it is the dollars won divided by the dollars dropped to the cash box. But there are so many variables that impact the hold percentage. Obviously the actual house edge when factoring in player errors impacts the hold but seeming more important is the speed of the game, numbers of players at the table, how long players are playing after buying in, etc.

I was thinking that the hold percentage is really a relationship between how much play time entertainment a game offers versus the expected loss by the player. So what is a fair exchange of play time for the average 20%+ hold rate paid by the players?

Should we be analyzing our games and tweaking them to come up with a theorectical amount of play that achieves that desired hold rate in a specified period of time? Is the hold rate on a per hour basis? Is it on a number of hands basis (i.e. should the player expect a certain number of hands dealt before "paying" the average hold percentage)? Obviously the hold is based in real terms on the floor and we can only calculate theorectical hold percentages when desiging a game. Game volatility should also come in to play, but I am not sure how to incorporate that factor.

Here is one way I was looking at it:

Let's assume a game has an Element of Risk (House Edge/average units wagered) of 2% assuming average player errors.
The average units in play during a hand is 2 units due to raising options during each hand.
Let's say an average hand with a table of four players takes 40 secs to deal a round (including shuffle time)

To achieve a theoerectical 22% hold on a 20 unit initial buy in (or 4.4 units), I estimate the following:

Theo win per hand is 2 units times 2% or 0.04 units. The number of hands required to achieve a theo win of 4.4 units is 110 hands (4.4/.04). So in theory a player will play for an average of 73 minutes (110 hands times 40 secs per hand) on the game for the casino to realize the desired 22% hold.

If we change the assumptions and assume the same 2% EOR but 4 units of play per hand and a 60 secs time per round dealt you end up with more like 55 minutes of play to achieve the same 4.4 units in theo win.

So should the theorectical time to achieve the desired theorectical hold percentage really be the tool that casino's use to compare game performance? Should a game really be judged in theory on the time of play required to achieve the desired hold target? If so, what is that desired time frame for Vegas casinos? Local casinos? Is the timeframe measured in hands of play or time of play?

I think the casino would use "hands" as opposed to "time", because they can manage the first. As long as the game has a HE, the more hands a player is exposed to, the greater the amount they will surrender to the HE.

Slower games will require a higher proportion of a players bankroll to be at risk, and a higher HE in order to make up for the lack of "decisions per hour". That is just my guess, since my observation from a player perspective is that management is mainly concerned with keeping the dice moving or the cards in motion to get in as many decisions as possible.

Casino operators focus on the bottom line; if they are holding money on a slow game, and losing money on a fast hands-per-hour game, the "fast" game goes, and the "money-making" game stays. For ANY game, management wants simply no time to be wasted, and may compare various dealers as to how relatively fast they deal each type of game.
For calculating a game design, the hands per hour should and income per hand should be known as to its speed performance, but its drop and hold performance from real-world field trials or simulations would be paramount. For example, the game gets 'x' amount of "drop" or buy-in activity - it "gets action," - and keeps or "holds" 'y' percentage of that money.

Quote: Paigowdan

For calculating a game design, the hands per hour should and income per hand should be known as to its speed performance, but its drop and hold performance from real-world field trials or simulations would be paramount. For example, the game gets 'x' amount of "drop" or buy-in activity - it "gets action," - and keeps or "holds" 'y' percentage of that money.

On the other posting, every poster seems to agree, "simulations (to get hold)" are impossible. Dan, do you agree with that?

Quote: Paigowdan

Casino operators focus on the bottom line; if they are holding money on a slow game, and losing money on a fast hands-per-hour game, the "fast" game goes, and the "money-making" game stays. For ANY game, management wants simply no time to be wasted, and may compare various dealers as to how relatively fast they deal each type of game.
For calculating a game design, the hands per hour should and income per hand should be known as to its speed performance, but its drop and hold performance from real-world field trials or simulations would be paramount. For example, the game gets 'x' amount of "drop" or buy-in activity - it "gets action," - and keeps or "holds" 'y' percentage of that money.

This certainly makes sense. You can have a great game for the house, fast paced with a low HE, or slow with a high HE; but if no one buys in.. you don't make any money. My guess is that outside of expensive and time consuming qualitative/quantitative studies, any Hold figure will be a guesstimate based on known current comparative offerings, promotion level, and field trial data.

In actual practice, I always wonder why management would keep a table closed when the ones that are open are even 80% full. Most players don't want to shoehorn themselves into a "full" table, and some will take their action elsewhere rather than wait for a space. They lose 100% of the money that walks out the door.