*****New Football Thread*****

Arran Shackell

This thread has been very quiet recently. Has everyone had a good weekend?

Greg Mitchell

backed Ten Haag to be next prem manager to leave at HT he was odds of 4, traded out at FT for nice profit odds of 1.8

Martin

@Richard-Latimer I mean they are usually very good on NI games

Richard Latimer

@Martin Random s**t that makes no sense is very like them.

I'm not spreading my money around but I wanted to lay Coleraine.

Martin

@Richard-Latimer not like them is it!

Richard Latimer

So only 1 Northern Ireland game on betfair today. Bit annoying

Simon Bates

@Shannon-Townsend cashed out at half time, too early
![alt text](image url )

Shannon Townsend

Anyone else's eyes light up when Liverpool went 1 nil down tonight... odds went to 2.2 for Liverpool win and was green and on a free bet within 4 mins...took the profit at 2-1... if Carlsberg did in play trades...

Matt Wood

Thanks for all the replies. Loads of stuff to think about there.

@Richard-Latimer Totally agree with you about starting with the stats before leagues. I only had a look at it the other way around to shut the little voice in my head up. Whilst it did produce some interesting results, it doesn't make sense.

@Alex-Rendell Averaging around 200 selections per month at the moment. The initial set was very broad and was just an idea I had that I did not really expect much from. Funny how things go when you look at the numbers though. Rolling seasons seems like a good idea, I'll steal that too.

@Stuart-Capstick Evening Stuart, Ive always had in mind that if I can get a solid LTD out of this then SHG trades would be possible if the games are drawing at HT. I am mindful, like you say, that this is only 1 seasons worth of data and would like to have another to back it up and see if the trends follow year to year. Interestingly I had a look at the losing selections and it seems like a BTD could be possible from this data too.

Stuart Capstick

@Matt-Wood

Hi Matt,

Well done on finding something with LTD. It really isn't dead; the draw is the hardest to predict, so prices are often wrong, which is lovely for those who trade the draw.

Maybe:

Each league needs its own strategy. The Irish league is very different to Brazil Serie A. It wouldn't be so outlandish if the draw was more likely under certain circumstances in one league as opposed to others.
Have a look at other outcomes- FHG, SHG, O2.5 etc
So many times for me a " failed" LTD turned out to be fruitful for SHG or laying the home team.
Be mindful that you have only one season's data. Sometimes leagues have blips.
BUT- FHG and SHG tend to resist those seasonal variations.

Alex Rendell

@Matt-Wood if you want to send me your spreadsheet or results etc to see if I can see things from another angle then feel free!

Alex Rendell

@Matt-Wood 1500 is decent! I'd say once you get in excess of 5000 then you can really start to drill into the individual leagues and see what works and what doesn't. Until then, perhaps a more broad approach would be best? I'd definitely look to filter odds ranges at the top and bottom ends though as a starting point, as opposed to leaving out say odds of 3.4 because they're showing a loss, whereas 3.35 and 3.45 both show profit, so they don't get removed. It just doesn't make sense.

How many selections are you averaging per month? If it's a fair amount then surely just aim for maximum profit over ROI? Tough call that one though.

I think 3-5 seasons' worth of data is good, anything older just doesn't feel relevant anymore. I only look at the past 3 seasons data personally, and keep that rolling so it's never more than 3 seasons old, if that makes sense.

Richard Latimer

@Matt-Wood said in *****New Football Thread*****:

Afternoon all.

Just wondering if any of you suffer from 'analysis paralysis' at all and if yes how and what do you do to move past it?

I ask since I am struggling with it now. I had an idea for a LTD (even though the internet says its dead ) for the summer leagues since I don't do much in the summer.

I have collected a fair amount of data now and just betting every game blind it has made 50pts.
I know with a couple of sensible tweaks to the data that return can double. The problem I am stuck with now is that I have become obsessed with maximising the return and therefor won't commit to making the change.

For example if I apply a couple of filters to the data Ireland is in good profit for the summer but with a couple of different filters it is in a slight loss. This is reversed for other leagues. Repeat this over and over across multiple attempts and I have become stuck with trying to find a way to extract the maximum winning leagues in one go, even though I know there is probably not a 'best' way.

Just curious as to how others find their way through this sort of thing.

Cheers in advance.

I know others have said tailor your approach to each league independently. I'm sure that works but for me it's way too much and I would find it overcomplication on mine.

When I look to optimise any filter I have I never look at what this does to a particular league. If the overall improves and the league is looking sub standard then it's the league which will hold my concerns.

Richard Latimer

In terms of what I filter first it's always the stats. Stats is logical. When I can't find any reasonable change to the stats and I'm still not happy that's when I'll look at removing poor leagues.

Richard Latimer

@Matt-Wood said in *****New Football Thread*****:

@Richard-Latimer Thanks Richard,

I think I was falling into a similar trap you describe and have taken kinda the same path you have in keeping the filtering to a minimum and making it sensible as Alex suggested.

I know you keep massive spreadies, do you keep traking a poor league and add it in if it turns good later down the line? Also do you filter first or remove bad leagues first? I have found doing these two things in different orders can ave a massive effect on my results. Not sure if thee is a 'best' option though.

I started off with all leagues which are 30+ on the global corruption index. That rules out the real bottom feeders and I didn't even track them.

After that, even the ones failing I will keep tracking for later with a few exceptions.

I've recently binned Estonia completely from my home lay, that and Bosnia. Sometimes it's liquidity, sometimes it's a feel all is not quite right.

South Korea I have found to be crap though and I'm still tracking them.

On my 3 most promising filters at present I've got between 12-16 filters, most looking very broadly. This gives me plenty of scope to change things later and really see

Matt Wood

@Martin I know! stupid right. All the more for us though

Matt Wood

@Richard-Latimer Thanks Richard,

I think I was falling into a similar trap you describe and have taken kinda the same path you have in keeping the filtering to a minimum and making it sensible as Alex suggested.

I know you keep massive spreadies, do you keep traking a poor league and add it in if it turns good later down the line? Also do you filter first or remove bad leagues first? I have found doing these two things in different orders can ave a massive effect on my results. Not sure if thee is a 'best' option though.

Richard Latimer

In my current home lay I have combined a max 2 stats although going forward I won't even do that. It's just much easier to analyse when you're looking at each stat in turn.

Richard Latimer

@Matt-Wood said in *****New Football Thread*****:

@Alex-Rendell Thanks for the reply Alex.

I have 1500 games total so not a massive amount of data but a fair bit.
I am trying to be very careful about the danger of back fitting the data with over tweaking, as you say this would in effect ruin the strategy. I have limited myself to making 2 or 3 sensible tweaks like you have suggested, and those are usually around an odds range or goals scored.

I like your idea of each league having its own rules and I might adopt this long term. Dont think I have enough data on most of the leagues to properly look into that yet. One for the future though certainly.

What I have been doing is trying 2 or 3 sensible tweaks recording the results, then resetting the data and trying some other sensible tweaks. It would seem that I am stuck trying to find a set of rules that gathers all of the profits from all of the leagues in one go. I honestly dont think this is possible so I have to make a choice on what 'best' is. Max profits or max ROI for example.

I think I will take your idea of each league to its own rules going forward. How many years do you take data from each league as being 'relevant'? More than 5 years back for example?

For my tuppence worth I found it to be really back fitting when I was combining stats, removing random things/numbers.

What I have made myself do this time is only remove numbers at the top or bottom end of each scale. Sometimes I think numbers are too high and sometimes too low. Even if I remove I mostly still collect the data so I can add it back much later.

Keeping your filter broad and tweaking it before placing trades is the way to go. Easier to see what's going on. I also use the excel smart filter as this helps me massively zoom in on potential issues.

Trick is seeing what is variance and what is just a case of where general improvement can be found.

Martin

@Matt-Wood the internet says LTD is dead? Try telling that to Ryan!