Re: Fixed vs float and precision and conversions

["Björn Lundin" <b.f.lundin@gmail.com>]

Den 2020-07-08 kl. 23:16, skrev Shark8:
> On Wednesday, July 8, 2020 at 12:08:16 PM UTC-6, björn lundin wrote:
>>
>> hmm no. valuies in global_odds_table are fixed_type
>>     Global_Odds_Table : array(Tics_Type'Range) of Fixed_Type := (
> 
> Except you don't *have*, in the things you are getting [from JSON], fixed-point.
> You are getting IEEE-float, as a consequence of JSON's definition.
> In order to find the proper values for a non-'Round-able you have to compare Floatingpoints to the acceptable-values in the table: for simplicity this means using the table-values as floats. [Conceptually they are universal-numbers.]
> 


hmm, it is difficult to condense a case to a simple case that 
demostrates the problem - and then does not over-simplify it, compared 
to showing lots of things that removes clarity - but then seems to be 
needed anyway.


> By putting the global-table in the convert function, as an array of floats, I got rid of the need for Tics_Type, as well as simplifying the translation (if-needed).

Yes, but Tics_Type is needed anyway.
Say I want to green up (put my self were I win no matter outcome of event))

I say runner 1 wins @ 1.85 and risk €10
then later - when runner 1 seems to be winning, odds have dropped to 
1.25 i say runner 1 does NOT win @1.25 risk €16

if it wins I get €10 *(1.85 -1.0) = €8.5
I also loose the lay-bet = - €16 *(1.25 -1.0) = -€4.0
total win = €8.5 - €4.0 = €4.5

if it looses
I get the €16
and loose the €10 => €16 -€10 = €6

Sometimes you want to calculate how much to risk the second bet
and using the array is nice then.
Sometimes I can say I want to greenup on a delta-tics = 20
That makes calculations easy

> 
> Take a look at the conversion function you provided, and the one I provided; there are two big differences: mine will return a valid Odds value, and that value will be that value from the table which is the closest value to the input-float.


yes but if the odds are outside 1.01 -1000.0 I'd like to throw an exception.
You may be right that your function gets the closer value.
I will re-examine.

Thanks again


-- 
Björn