General Opinion on a how to?

[Daniel]

Hi guys

I have a requirment to calculate a prize pool for a competiton similar to a
poker tournament. But i am stuck on best ways or even if i have the right
ideas, so:

Players pay to play, total paid is prize pool and is distributed to top 3 or
so winners.

Variables are number of players who will win, if a lot enter i increase
number of winners. i think a standard top 30% rounded off would work for
calculating paid positions? So 10 players. 30% would be 3 of them (rounded)

Now to calculate the actual amount of prize each gets i'd like to be able to
vary it. So i could offer or be versatile enough to offer different
distribtuions. Such as, a winner takes all, a top 2 get the most and then
3rd gets a fair bit less, or a 4 out of 10 get prizes.....etc

Also i'd like to be able to offer prizes where say first place can get cash
AND a car for example. But other 2 places just cash etc etc

This is where i am stuck, do i go for a hard coded table in database and
read off it when my prog loads into a struct and just reference that when
doing calculations? One way forward, but then if a number of players enters
that i havent accounted for, then what?

Alternatively as i said above i calculate winning positions based on a % but
then how do idecide which of those positions wins what?

Very stuck and confused, i keep thinking this must be simple. any ideas or
pointers welcome?

Thanks

 

[Jamey McElveen]

Daniel,

You could create an interface that defines a winnings calculator and then
uses a factory to dynamically load and instantiate the named interface from
an external assembly. This would allow you to be flexible enough to add new
calculations to the system down the road.

Jamey

 

[Peter Bradley]

Seems to me you need to do some analysis.

Some obvious objects are:

Game
Player
Prize
PrizePool

Now, can you say what state variables and methods these objects should have?
Does thinking about that suggest to you that you may want an inheritance
hierarchy, or an interface for some of these objects? E.g.

Could there be different types of Games (Pool, Poker, Darts, Domines,
Cribbage)? Would they have extra/different state and behaviour?

What about prizes? Would an interface work best here? Then a PrizePool
could hold objects that implement the interface?

Will a Game hold a collection of players? Presumably a PrizePool will hold
a collection of Prizes. Since the Prizes may be different, will they be
different in an inheritance sense, or in an interface sense? Will a
PrizePool need a reference to a Game (i.e. need to know which Game it is the
PrizePool for)? Will a Prize need a reference to a Game, too? Or could the
same Prize be used for many Games.

There's no substitute for good analysis.

HTH, IMHO, YMMV, Just my 2c etc etc


Peter

 

[Daniel]

That part is already done and complete. The issue is more the back end
storage of the prize pool break dowm to hard code in code as your method
suggests or not?

If i have a database table that reads the variables out then that to be is
the only way forward.

As Jamey said, a interface and a factory sounds a good plan and i can
instatiate each from the database in advamce choosing the one needed at run
time.

The way i see my calculation working with the aid of Jameys tips is this:

- load a game
- Retrieve Tournament placed winning positions from database as a %, top
30%, top 1% (winner takes all), etc
- Load into a prize pool object instance for that game

My biggest issue is the distribution of that prize pool. Rather than hard
code, "if 3 prize positions then break down as follows", "if 4 etc".

I would rather some form of algorithm for or setting in my database to say
that for this game use this ditribution, for this one use another one, and
eaxh distribution be stored in a back end db table.

But how do you store a 'distribution'? I guess its an algorithm, if so i
could store in the db which algorithm to use etc? this is where i am
stuck....

Any ideas?

 

[globally_unique_identifier]

If it were me, I'd implement a rules engine for the project. Your
design must define the general rules of how the payout behavior will
operate based on the number of players. Check out the book "MDA
Explained: The Model Driven Architecture--Practice and Promise" for an
excellent example of how to implement an object-oriented rules
engines. For the varying behavior you describe, I would just use the
random number generator seeded with the current second when the payout
is calculated. Your algorithm's in the rules engine will have to be
able to use this value somehow to calculate the varying payouts.

 

[Jamey McElveen]

You could store the name of the algorithm in the database along with some
params for it. I have done this two ways in the past successfully. First,
you could store the params in a TEXT field as xml. Second you can create
another detail table with key value pairs. Either way you would not have to
change the database as you add new algorithm to the system. We used a
system similar to this for calculating holidays for a calendar. Almost all
holidays on the calendar could be calculated with a few basic algorithms.
<holiday name="CHRISTmas" algorithm ="date" day="25" month="12"/>
<holiday name="July 4th" algorithm ="date" day="4" month="7"/>
<holiday name="Thanksgiving" algorithm ="weekday" month="11" week="4"
weekday="5"/>

however easter was different so we had create a special algorithm for it.
<holiday name="Easter" algorithm ="easter"/>
<holiday name="Good Friday" algorithm ="relative" relativeHoliday="Easter"
metric="day" distance="-2"/>

All in all we have arround 5 algorithms and almost 100 holidays. This came
in very usefull this year because with daylight savings time changing we had
to only change the xml file.

Hope this helps

Jamey

 

[Daniel]

Hi Jeremy,

Thats exactly the idea i had in mind, to store the algorithm for the
calculation in the db and then have my server use that. So different prize
distributions could be calculated with ease. No need for any big design
changes to my code.

So do all calculations in the code based on the database values. I'll look
intot his methodology. My next problem is how to calculate a non uniform
spread in c#. Another post may loom i think.

thanks to all responses.