Forums

We want to share an important update with our Betfair Community users.

Over the coming months, the Betfair Community platform will be winding down. The final active day for the platform will be December 21st. From December 22nd, users will no longer be able to access the platform.

Your discussions, insights, engagement, and, most importantly, feedback have helped shape the platform and guided innovation. Your feedback continues to be important to us, and we will continue to listen via our dedicated feedback channels, as well as through product testing and innovations.

To share your feedback and experience, visit: https://support.betfair.com/app/home

General Betting

There is currently 1 person viewing this thread.
FAlonso
04 Jul 10 16:31
Joined:
Date Joined: 24 Sep 07
| Topic/replies: 7 | Blogger: FAlonso's blog
I've been backtesting something over a sample size of about 1700 games. Haven't done the ROI calc, but it is over 20% based on the strike rate and the odds that I've been using to do the backtesting.
Furthermore, as of now the sample size includes games from the majority of leagues that one can name, bar most that run interyearly. Backtesting for some countries goes all the way to the 05/06 season. If I recall correctly, so far I have a total of three losing seasons in the countries that have been tested: Belarus 07; Belarus 08, Cyprus 09/10.

I've been trying to figure out what I am missing, and why this won't work, and the only reason I come up with is that it is backtested.

So, pls take me out of my misery and help me fathom why this will end up costing me.

Post your reply

Text Format: Table: Smilies:
Forum does not support HTML
Insert Photo
Cancel
sort by:
Show
per page
Replies: 13
By:
sweetchildofmine
When: 04 Jul 10 17:35
no offence but i cant be the only one flummoxed by your post
By:
raspberrybottom
When: 04 Jul 10 18:28
If you've been honest with yourself and honest with your backtesting (not too much backfitting) then give it a go.

Good luck with it.... :)
By:
sweetchildofmine
When: 04 Jul 10 18:33
thats the trouble with back testing...its so tempting when a stat threatens to ruin the dream, to just quickly overlook it
By:
kef4o
When: 04 Jul 10 18:54
Everything is kept absolutely consistent. The only mistakes in my data research would be because of me overseeing something in a game or two due to going over numbers too quickly, i.e. it would definitely not affect my current sample size.

The negatives that spring to mind all revolve around the fact that it's only stats. No team news, no personal opinion, no gam schedule checks, no "don't play the first/last rounds" etc. Yet, it is profitable despite all that.

Moreover, I have covered all types of leagues, from the EPL to the Conference, even countries that I think are unavailable here, like the Albanian Super League. So far, this is profitable throughout every league. Even Belarus is profitable over the last five seasons, including the current one, despite having two losing ones and being 0/2 for this one.
By:
Northbouy
When: 05 Jul 10 03:43
email me your system and I will tell you why it wont work [;)]
By:
Lori
When: 05 Jul 10 10:46
If you've done what you say then you'll win.

My suspicion is you've made a fundamental mistake/assumption though. I'm sure you're not an idiot as you've bothered to backtest, so excuse if some of these seem to be aimed at morons... I just find it hard to believe there's a 20% ROI system out there (note, I don't think it's impossible, just bloody unlikely)

Possible brainfarts:

1: You've taken all the odds and added them up then averaged. Something like 1.4+1.5+1.6 (1700 times) averaging out to 1.5, something like that

2: um, i'll get back to you... 1) is such a common error!
By:
kef4o
When: 05 Jul 10 14:34
The way I've estimated the ROI to be at 20% is 'strike rate' x 'MINIMUM odds' - 1. The average odds in the sample size should be at LEAST 15% better than the minimum ones.

There is a theoretical possibility that the odds I've used are wrong. It looks like the odds quotes that I use are running on the same "engine"(?) as the ones of betexplorer. Are betexplorer's quotes to be trusted?
By:
SHAPESHIFTER
When: 05 Jul 10 14:47
If it involves laying, then the 20% can easily disappear. 

From someone that uses back testing to find "potential" markets, it also comes down to being "selective" on the games (someone mentioned team news, etc.  This would factor into your decision or your wagering unless you are simply doing "click and win").
By:
kef4o
When: 05 Jul 10 15:41
It's not that either, though a very good shout. I'm backing.

Also, I am basically testing exactly that, "click and win". I'm only using stats, or so I believe at least. I'm testing every league that I possibly can in order to increase the sample size, test for changes in the effectiveness based on time, geographical location and league competitiveness. I might need a team of twenty people to keep up with the news and stuff.

And this is where my problem is. I've obv done a lot of this stuff in the past. And every time this boils down to about break-even, so neither laying, nor backing would be profitable. It makes perfect sense for it to end like that. Yet, this is not the case here. What is even scarier, I've used the very same data previously and actually thought of it as being pretty much worthless. All I did now is add one condition, and a price band.

I do experience normal variance stuff, like losing seasons, inconsistent ROI from season to season and from country to country. Some countries have proven themselves more profitable than others over a five seasons sample and that is to be expected. I've also seen long losing streaks to remind me that I need a very good staking system. I've also had the end of the stick, 100% seasons, albeit for a small number of bets. But I'm yet to really figure out why this won't work, and it can't work b/c by definition 20%-25% ROI is next to impossible, especially given the fact that I only use stats.
By:
Soccergeez.
When: 05 Jul 10 16:37
If it deffo cant work than whack it up here and we'll work out your mistake Devil
By:
Chilly the Dog
When: 05 Jul 10 17:23
Can you simulate the actual P&L of all the theoretical bets placed, taking into account x% (where x is whatever your rate is) comm? Would this push you to the PC (and would you then be running apointless system?).

Have you compared it to a random strategy for a baseline?
By:
kef4o
When: 05 Jul 10 17:42
I have a long way to go to be PCd, and if I understand the terms correctly, I should be just about pushing the envelop and in time every week might turn into a sweat over it. Pinnacle and some asian bookies might be a viable alternative. Moreover, some leagues almost certainly will have to be bet there.

"Have you compared it to a random strategy for a baseline?"

Provided I understand you correctly, then yes. As stated previously, I've done system backtesting like that many times. You might as well call the factors that I've used random but tailored after some basic assumption.
By:
kef4o
When: 05 Jul 10 17:43
*to be bet ONLY there
sort by:
Show
per page

Post your reply

Text Format: Table: Smilies:
Forum does not support HTML
Insert Photo
Cancel
‹ back to topics
www.betfair.com