Dalfort
Joined: Jun 23, 2008
|
 Posted:
Mar 23, 2015 - 02:17 |
 
|
**GEEK ALERT**
As many of you know I keep track of certain stats of mine. I have a spreadsheet that I like to update regularly.
I would like to know if there is a way to calculate the number of games needed to increase an average, for example;
582/393/1988 (my total game record) yields a 26% win rate. How do I work out how games I need to win (or draw) in order to effect a change upon the average. If I could then factor in the current win 26% without totally confusing myself that would be fantastic.
I only studied Mathematics at High School level so I not sure a what level this question lies (and would be interested to know if somebody does know!)
Thanks for your time, take care Dalfort. |
_________________
 |
|
jamesfarrell129
Joined: Dec 23, 2009
|
| Posted:
Mar 23, 2015 - 15:21 |
|
When you say "effect a change upon the average", what exactly do you mean?
Are you wanting to see for example how many games you'd need to win, to improve to say 30% (as a start)?
I would imagine its going to be a fairly arbitrary function, tricky to get exact and still be easy to explain.
You've currently got (w + 0.5d)/(w + d + l) ~= 0.26.
And you want is e.g. (w+x + 0.5(d+y))/(w+x + d+y + l+z) = 0.3
where w,d,l are your current games, and x,y,z are your future games.
I guess the next step would be to expand that out, and come up with some simultaneous equations to limit things (i.e. you want to get to 30% in your next 200 games, or whatever) |
_________________
NAF name: sandwich |
|
sann0638
Joined: Aug 09, 2010
|
| Posted:
Mar 23, 2015 - 18:27 |
|
Send me the spreadsheet Dal - sann0638 at hotmail dot com. |
_________________
NAF Ex-President
Founder of SAWBBL, Wiltshire's BB League on Facebook and Discord |
|
sann0638
Joined: Aug 09, 2010
|
| Posted:
Mar 23, 2015 - 22:43 |
|
|
Dalfort
Joined: Jun 23, 2008
|
| Posted:
Mar 23, 2015 - 23:20 |
|
Cheers mike, I tried to reply earlier but the pub network dropped for my browser but not facebook...
Thanks again guys I will look deeper when I get home
Take care Dalfort |
_________________
|
|
Dalfort
Joined: Jun 23, 2008
|
| Posted:
Mar 23, 2015 - 23:20 |
|
Edit: Double post from mobile. |
_________________
Last edited by Dalfort on %b %24, %2015 - %08:%Mar; edited 1 time in total |
|
xnoelx
Joined: Jun 05, 2012
|
| Posted:
Mar 24, 2015 - 00:08 |
|
The simplest spitball technique if you want to know how many games you need to make a 1% change is to divide your total number of games by 100. That gives you 1%. i.e at the moment, you need about 30 wins to get from 26.2% to 27.2%.
Obviously this is a rough and ready approach. If you're happy with rounding anything over 26.5% up to 27, you need less games, and if you want to know how many to get to exactly 26% you'd have to proportionally adjust the number of wins (in this case, 30 wins gives a 1% rise, so a 0.8% rise requires 24 wins) and every game played adds to the total number of games, making the figures constantly evolve, but it gives you a reasonable ballpark figure.
Also bear in mind that a draw is worth half a win. So 10 wins and 40 draws would achieve the same change, although if that involved playing more games, raising your total further, the effect would be slightly watered down. |
_________________
 Nerf Ball 2014 |
|
Dalfort
Joined: Jun 23, 2008
|
| Posted:
Mar 24, 2015 - 08:51 |
|
cheers noel, that is an excellent layman explanation  I googled some stuff the reminded me of growing up the Open University guys in their their brown wooly tank tops on BBC 2... very scary, and no numbers !!!! :p
Going to look over Mikes link now, hope I can follow it |
_________________
|
|
Dalfort
Joined: Jun 23, 2008
|
| Posted:
Mar 24, 2015 - 08:59 |
|
Awesome MIke, nice and simple approach I should be able to play around with that idea somewhat to produce predictions with the stats I have.
Thanks for your help everyone, James my initial Pub post thanked you straight way as well... my phone was doing my head in, it stopped the auto-orientation thing, then I kept hitting the wrong letters and having to correct 75% of the words I typed, whilst trying to hold 2 text conversations as well... that was on top of a quite stressful day all round. Better now though :p |
_________________
|
|
Roland
Joined: May 12, 2004
|
| Posted:
Mar 24, 2015 - 09:28 |
|
I think I have the formula, I'll type asap. |
|
|
Dalfort
Joined: Jun 23, 2008
|
| Posted:
Mar 24, 2015 - 09:33 |
|
Sorted it from your explanation Noel, your example spreadsheet Mike and your support James :p
I simply created a new cell to give me each win% if the game was won, drawn or lost by using a cell reference +1 in the same formula but in each different case (hmm easier to do than explain  ). My sheet goes to 2 decimal places so each game is still affecting the % at some point I will work out how to use the last 10 games as a "form" guide for the prediction thing... or something, its just to amuse myself instead of getting vexed at minor things (like snaking GFI rolls...)
Thanks again everyone ! |
_________________
|
|
Dalfort
Joined: Jun 23, 2008
|
| Posted:
Mar 24, 2015 - 09:35 |
|
cheers Roland, still interested in anything anybody has that can add more calculations to the sheet  |
_________________
|
|
Roland
Joined: May 12, 2004
|
| Posted:
Mar 24, 2015 - 10:29 |
|
| Roland wrote: | | I think I have the formula, I'll type asap. |
nope, had an algebra formula, but it doesn't work.
It must be some kind of iterative process, where you do a calculation for each game and then the # of steps is the # of games needed.
I'll rethink... |
|
|
Dalfort
Joined: Jun 23, 2008
|
| Posted:
Mar 24, 2015 - 10:45 |
|
I think try to explain what we are doing is more difficult than playing with the numbers/formulae but thanks for playing along with me |
_________________
|
|
jamesfarrell129
Joined: Dec 23, 2009
|
| Posted:
Mar 24, 2015 - 12:43 |
|
Now I'm just worried that your idea is going to inspire me to create another spreadsheet... I think I've got about 5 to track various different stats now, all linked to each other.
Yeah, I'm a geek, and I get bored at work sometimes...  |
_________________
NAF name: sandwich |
|
|
|
Conceding v Goblins/...
Search
