Lunatism’s Statistical Selections (lss)

[lunatism]

Re: Lunatism’s Statistical Selections (lss)

As I said earlier' date=' splitting the results into leagues was likely to give too low a number in each for analysis purposes. The breakdown you have kindly provided tends to confirm that. However, that is not to say that the analysis is useless. You might well decide to use it to help you decide which selections to [b']omit (rather than which to include, as such) with a view to increasing your strike rate. So, for example, if you had a bit more data, and the trends remained the same, you might decide never to include matches from ENG03 as a 1-0. I always think it's prudent to look for what you should leave out

that's true, but i guess the perfect sample size still needs to be obtained before i make any premature cuts :eek

One word of caution though. The bookies will carry out much of this analysis themselves. If' date=' for example, a 1-1 scoreline comes up as regularly as clockwork in the Dutch Jupiler League, as your stats currently suggest, you can be pretty sure the bookies will know that and price their scores accordingly.

that's a little point nagging at the back of my mind, and yet it still puzzles me to find that i could still get decent odds for games that seem to look like they'll end in that fashion. for instance, check out the jupiler fixtures for this friday at bwin - all are at 7.00 for the 1-1.....but i guess this just confirms that not all bookies think along the same lines anyway. weird, huh?

[lunatism]

Re: Lunatism’s Statistical Selections (lss) not long to go now ladies and gents.....as i type this, excel is working it's magic, updating stats and what-not, within the next 20 mins or so i should be able to put up today's selection (yep...only one for today....but looks like a nice one ;)) and the over/under thing - once excel is freed up, i'll do that too, i'll just show overall and then from the list of leagues i displayed earlier in the thread

[lunatism]

[lunatism]

Re: Lunatism’s Statistical Selections (lss) don't fancy editing a post with a selection in it, so i'll make my correction here....it's only a small one anyway...... the AH is NOT from pinnacle, but rather from bet365.....i don't fancy waiting around for that one when 1.85 is good enough anyway

[F.]

Re: Lunatism’s Statistical Selections (lss)

One word of caution though. The bookies will carry out much of this analysis themselves. If' date=' for example, a 1-1 scoreline comes up as regularly as clockwork in the Dutch Jupiler League, as your stats currently suggest, you can be pretty sure the bookies will know that and price their scores accordingly.

that's a little point nagging at the back of my mind' date= and yet it still puzzles me to find that i could still get decent odds for games that seem to look like they'll end in that fashion. for instance, check out the jupiler fixtures for this friday at bwin - all are at 7.00 for the 1-1.....but i guess this just confirms that not all bookies think along the same lines anyway. weird, huh?

I think I know why. It comes down to bwin's business model. Let me explain: It just happens that bwin is also my favorite bookie (that's another story). I have been betting almost exclusively with them for almost two years and during that time I took note of some particular behavioural (odds pricing) patterns. The conclusion is basically that most of their oddsmakers devote their time to the most popular leagues and the most popular markets (1x2, O/U, and so on). Therefore, the odds for say, English Premiership's 1x2 are very accurate. However, bwin chose to offer a wide number of leagues and markets applying a "bruteforce" approach based on a large overround and each league's average for a market. For example, let's say that the H/D/A market for the Czech Gambrinus Liga has the following distribution: H:46.67% / D:25.70% / A:27.63%

and the respective fair odds are: H:2.14 / D:3.89 / A:3.62 Based on their business model the bookie then just decides to apply a 30% overround on the fair odds to come up with the following numbers: H:60.67% / D:33.41% / A:35.92% (130% total) ABUSIVE ODDS: H:1.65 / D:2.99 / A:2.78 The interesting part, though, is that they apply this odds fixing scheme to *all* of the matches of one league!! Therefore, there are potential value bets despite the fat overround. Disclaimer: Before you people come down on this post, let me make clear that: 1. bwin doesn't set odds in this fashion for *any* H/D/A market. They do it only in markets where they can get away with huge overrounds, (just like the correct score market). 2. The %s I posted are not accurate. They are just an example to illustrate my point. 3. The actual odds setting procedure might not be as simplistic as I just described. However, my point is that bwin's pricing is not as accurate with lower leagues and less popular markets as it is with mainstream betting. Therefore there are value bets to be found despite the fat overround. Anyway, a few months ago I posted a thread precisely about this scenario. Take a look at it if you want to explore this issue in greater detail. F.

[lunatism]

Re: Lunatism’s Statistical Selections (lss)

Anyway' date=' a few months ago I posted a thread precisely about this scenario. Take a look at it if you want to explore this issue in greater detail

WOW! definitely IS a lot of detail on that thread! Just read through it, and i must say i do agree with your points about certain bookies used "set odds" and generally being lazy. I guess that would these these particular bookies the best for the CS prediction market aswell...... i had noticed it before on bwin, but it never clicked as to why so many appeared similar, but i guess a lot of bookies do it, especially when it comes to the lesser known leagues and markets such as CS (i mean, what lucky sod would be consistent enough on that market anyway? :tongue2)

[F.]

Re: Lunatism’s Statistical Selections (lss)

WOW! definitely IS a lot of detail on that thread!

Well, you aren't the only one that goes over the top when it comes to stats-based research, you know? :tongue2 ;)

what lucky sod would be consistent enough on that market anyway? :tongue2)

You might be the first one !! :hope F.

[lunatism]

Re: Lunatism’s Statistical Selections (lss)

You might be the first one !! :hope

i'll drink to that! i've just gathered the stats for the under/over markets that The Hat Man was asking about, and one filtered down...and filtered again, it seems to show some good news - although if anyone could enhance my average under/over estimates i'd be happy enough ;)

[lunatism]

[lunatism]

Re: Lunatism’s Statistical Selections (lss) definitely know that the overs will need a lot more data before it could be classed as "good", but a 78% strike rate isn't half bad, plus it's always nice to know it passed the Chi Squard Test too (99.9999805830307%)......i guess something else to sit on! i'm still toying with the idea of starting a new thread for the CS, but now i wonder whether i should just carry on colecting data for the next month or so, popping up every week with an updated sheet of stats - or should i start posting the selections on here? what do you (anyone!) think?

[Punto Banco]

Re: Lunatism’s Statistical Selections (lss) Wow you guyz get alot done while I am asleep and at work!!!! Anyway personally I would just continue along in this thread... while making a separate thread for paper trailing would not be a terrible idea it may be a lot easier to just keep it all confined hear.... one thread is easier to manage than two and truly you probably don't quite have enough proof on some of this to :clap it yet therefore posting up bets etc could lead to being a big waste for yourself in a few cases... a page of stats with updated "overall results" I would think would be the better course of action.

[happygooner]

Re: Lunatism’s Statistical Selections (lss) Am I right in thinking you have really averaged out odds for under 2.5 market to 1.45??? I would be incredibly suprised if this were the case:eek Your strike rate for unders looks plenty good enough to profit averagely while your strike rate for overs (admittedly based on small numbers) is mightily impressive. A strike rate of anything around 60% should be golden in the long run. My own feelings on this would be to actually start posting selections. Nobody has to follow but it's good to see this thing working in practice. 1x2 market, over/unders, CS? Multitude of qualifying leagues.... Have you struck gold in them there hills:nana

[lunatism]

Re: Lunatism’s Statistical Selections (lss)

Am I right in thinking you have really averaged out odds for under 2.5 market to 1.45???

not at all - i didn't know what to use, so i just put that figure in (lower than the one that you gave me - so that at least it wouldn't be too high). Another reason i did this - when i was looking my data from the last time, not necessarily with the overa, but almost always with the unders, i would find that the bookies knew exactly which games were going to be low, and they would price the odds accordingly (not quite the same as they do with the CS). And so either the odds would be really low (not THAT low, but i was exaggerating), or they would only offer over/under 1.5 or 0.5 instead! what i'm after is a reasonable average for the under - but i know that (based on past data....which for some reason did slip my mind before) the odds have never been so pretty for the unders when my system was involved

My own feelings on this would be to actually start posting selections. Nobody has to follow but it's good to see this thing working in practice. 1x2 market, over/unders, CS? Multitude of qualifying leagues....

what do you mean by "multitude of qualifying leagues"? :) first, i'll get to seeing if there are any further ways to improve the strike rate by modifying the filters in place...and of course without reducing the sample sizes to nothing! and new ways of looking at the data, i'll immediately notify you guys

[lunatism]

Re: Lunatism’s Statistical Selections (lss)

Wow you guyz get alot done while I am asleep and at work!!!!

yep - i also get so much done while i should be at sleep and at work too!:tongue2

Anyway personally I would just continue along in this thread... while making a separate thread for paper trailing would not be a terrible idea it may be a lot easier to just keep it all confined hear.... one thread is easier to manage than two and truly you probably don't quite have enough proof on some of this to :clap it yet therefore posting up bets etc could lead to being a big waste for yourself in a few cases... a page of stats with updated "overall results" I would think would be the better course of action.

how about this for compromise (not like i even need one ;))? I won't start a new thread (that will probably just fraction input i'm getting, plus i'll end up forgetting to post in one of the other), i'll post my selections in here, i'll start posting on saturday (but EVERYONE MUST understand, these are NOT to bet on, they aren't even selections in the sense of things, and are only being posted to show not only that i didn't make up any data, but also as sample size fodder!). i'll split the posts into three tables - CS, 1X2 & Under/Over, the odds i post may not always be the best priced ones, but i'll be sticking to only one or two bookies for this, since in the long term there is no way i would (or recommend that you) open up 5/10 different account to up the yield by a fraction. but i'm sure there WILL be better odds for these selections (CS & over/under rather than 1x2) because it only looks at the list of leagues i put up earlier in the thread, rather than the "whole world". something i guess i will need some input on - which stats should i put up on here anyway? i know the 1-1/1-0 CS, but which stats for the Over/Under? or should i just go with all of them to be on the safe side?

[happygooner]

Re: Lunatism’s Statistical Selections (lss)

not at all - i didn't know what to use' date= so i just put that figure in (lower than the one that you gave me - so that at least it wouldn't be too high). Another reason i did this - when i was looking my data from the last time, not necessarily with the overa, but almost always with the unders, i would find that the bookies knew exactly which games were going to be low, and they would price the odds accordingly (not quite the same as they do with the CS). And so either the odds would be really low (not THAT low, but i was exaggerating), or they would only offer over/under 1.5 or 0.5 instead! what i'm after is a reasonable average for the under - but i know that (based on past data....which for some reason did slip my mind before) the odds have never been so pretty for the unders when my system was involved what do you mean by "multitude of qualifying leagues"? :) first, i'll get to seeing if there are any further ways to improve the strike rate by modifying the filters in place...and of course without reducing the sample sizes to nothing! and new ways of looking at the data, i'll immediately notify you guys

By multitude of leagues I just mean that you are tracking so many minor leagues as well as the big ones. Most over/unders fall lowest around 1.6 when a low scoring game is expected. Alot would fall into the 1.7-1.8 category and some around 1.9-evens hence the potential in a strike rate that hits at around 60% or over! Betfair always provide over/under 2.5 market if they are listing the total of a game at all. They list the total on most. Also bet365 have alternate totals alot of the time so you can buy down or up.

[lunatism]

[lunatism]

Re: Lunatism’s Statistical Selections (lss) over the next few days/weeks i'll decide what action to take over the CS, overs looks promising, but that definitely needs a larger sample, so i'll keep all my spreadsheets intact for that - but hold off on the posting of selections until i'm sure it's not a waste of time :ok the CS seems to be a tricky one - so i'm going to put some hard graft into fine combing for any irregularies/errors and anything like that before i start posting those "selections" (don't want to raise hopes too much). as soon as i can be as happy as possible at this, i will be posting those. which means (contrary to my earlier post) that i MAY not be able to post selections on saturday, but i really hope i can - there will be a LOT of fixtures for that day i'm guessing in the mean time, ANY more sugegstions of filters i could run on the cs or o/u data, i'd be welcome to try it. and any of you have any extra tests that you think i should run (to establish more confidence in the cs), then just say so....... .....haven't been to bed yet :loon so i'm off to catch up on that now :zzz

[happygooner]

Re: Lunatism’s Statistical Selections (lss) So this weekend are you going to start tracking the odds for these over/unders to see what you get? Probably get a much better idea of the averages you should be basing previous figures on too. Anyway have a well earned kip:ok

[The Hat Man]

Re: Lunatism’s Statistical Selections (lss) this, is a little more tricky - for starter, pinnacle don't do combo's, and because of the smaller number of minor leagues they provide, bet365 wouldn't be such a good idea either, leaving bwin. and the problem then, is that i cannot think of a suitable way to stick these together! if i picked the highest rated fixtures, and since i wouldn't (in most cases) be able to use AH on these, i might end up with a couple of 1.25's, combined to make a "meagre" 1.56, of course profit is profit at the end of the day, but when they combine for such low odds, i just wonder if there is any point. what would your take on that be? You're right this is more tricky. I was really hoping to look into the possibility of multiple AH bets but it looks as though this will never be possible unless the selections happen to be from more well known leagues which will be priced up by more companies. So it looks as though multiples may not be the way forward :cry.

[lunatism]

Re: Lunatism’s Statistical Selections (lss)

So this weekend are you going to start tracking the odds for these over/unders to see what you get? Probably get a much better idea of the averages you should be basing previous figures on too.

sure, but you need "tell me" which over/under stats i should be looking at first of all! which ones are the most appealing to you? because personally, i'd only look at the "over based on cs predictions of 4 goals or more", since that's the only one that shows a strong enough strike rate/yield

[lunatism]

Re: Lunatism’s Statistical Selections (lss)

You're right this is more tricky. I was really hoping to look into the possibility of multiple AH bets but it looks as though this will never be possible unless the selections happen to be from more well known leagues which will be priced up by more companies. So it looks as though multiples may not be the way forward :cry.

maybe not at the minute, but say this CS or over/unders thing works out ok, we could mix 1 1x2 with 1 cs and 1 under (or over), what dya think about that? could make combined odds of around 15.00!! it's a bummer my system seems to avoid the well known leagues, an interesting stat for you here - whenever a league from the list i posted further up the post has been a selection, the strike rate has been a "poor" 65%, average odds of 1.40 with a yield of -15%, so it's not exactly proving profitable to take the top leagues full stop for my 1x2 (so far :)) dropping the top leagues from my (original) system would see a 71% strike rate and a 3% yield, with average odds of 1.43.....compare this my original, and there is hardly any difference due to the better odds i guess..... so it wouldnt actually prove profitable to drop the top leagues just yet, but i do feel that there is more value to be made from the lesser knowns...basically due to their lesser known status! bookies know a lot less - just like we do ;)

[happygooner]

Re: Lunatism’s Statistical Selections (lss)

sure' date= but you need "tell me" which over/under stats i should be looking at first of all! which ones are the most appealing to you? because personally, i'd only look at the "over based on cs predictions of 4 goals or more", since that's the only one that shows a strong enough strike rate/yield

When you say top leagues which leagues does this include? Are we talking all UK leagues, top 2 Italian, Spanish, German, French & Dutch leagues? Sticking to the more well known leagues can you extract data to show separately how 0-0's have fared, 1 goal predictions & also 1-1 predictions? Logic would suggest 0-0's fare the best but it would be interesting to find out if that's the case in practice. Over how long a period are the numbers tracked here? Hundreds of samples for under and only very small for overs? I'm assuming the 10 bets for overs in top leagues are probably no more than a 2-3 weeks data? So the 405 under bets, over how long a period is this data taken?

[lunatism]

[lunatism]

Re: Lunatism’s Statistical Selections (lss) damn it - i too seem to have trouble of skim-reading :lol you did mention "separately"...anyhow, at least the stats are there now anyway

[happygooner]

Re: Lunatism’s Statistical Selections (lss) Don't you just hate it when people are too fcukin lazy to go back and read:$;) Well for what it's worth I reckon it's definitely worth tracking the 0-0 predictions as well as the 4+ goals. The minor leagues, if found, would produce excellent profit while it's too early to dismiss the top leagues imo. The 4+ results speak for themselves...again it's probably a good idea to track them both but do it separately as top leagues may end up being only feasible option. Now could you split the 0-0's into games where either the main striker or main defender of either team earned a red card:tongue2

[lunatism]

Re: Lunatism’s Statistical Selections (lss)

Don't you just hate it when people are too fcukin lazy to go back and read:$;)

haha....we all have our faults ;)

Well for what it's worth I reckon it's definitely worth tracking the 0-0 predictions as well as the 4+ goals. The minor leagues, if found, would produce excellent profit while it's too early to dismiss the top leagues imo. The 4+ results speak for themselves...again it's probably a good idea to track them both but do it separately as top leagues may end up being only feasible option.

ok, here's what i'll do then - i'll show the overall / top league stats for under from 0-0 predictions and overs from 4 goals plus....that should be fine, right? any more will be "available on request" ;)

Now could you split the 0-0's into games where either the main striker or main defender of either team earned a red card:tongue2

sure i can do that - but first i'll need a runner - running from country to country, (of course fluent in ALL world languages) who can check up and make tables on that info he/she gathered. upon his/her return, there would be more statistics, a job worthwhile, don't you think? :D fancy it?!!

[happygooner]

Re: Lunatism’s Statistical Selections (lss)

haha....we all have our faults ;) ok, here's what i'll do then - i'll show the overall / top league stats for under from 0-0 predictions and overs from 4 goals plus....that should be fine, right? any more will be "available on request" ;) sure i can do that - but first i'll need a runner - running from country to country, (of course fluent in ALL world languages) who can check up and make tables on that info he/she gathered. upon his/her return, there would be more statistics, a job worthwhile, don't you think? :D fancy it?!!

You know what, if you're paying travel expenses...i'm in:rollin

[lunatism]

[lunatism]

[lunatism]