Footbe

Footbe Forum

Welcome to our community forums. Your feedback is greatly appreciated.

Please note that your Footbe app account details cannot be used on our website. You'll need to sign up
to use the forums separately. However, it is advised to use the same email address with your app account.

Register Login


Avatar

Seasonal Results Review: Personal Accessment

Started by spirit - July 14, 2015


Please login to reply


#1 spirit


13 Posts

2015-07-14 19:56

I have been going through the results of the 2013/2014 and 2014/15 leagues seasons respectively and have noticed some consistency of results patterns and the very vivid example shows the month of september seems to be very fruitful whiles march seems to be the direct oposite for both two seasons. From the way I see things, it wont be any surprise to see history repeating itself unless something is done about how each of these turn of events contributes to the goal of the footbe app, and my first question is. can we be rest assured this coming leagues season, to expect a positive turn around with the results of the months that seems to be unsuccessful in terms of accuracy. Because the logical thing anybody would want to do to scale up profits will be to read the trend, wait for the exact set of circumstances require to all come together and then take advantage. Simply put, stake when the lights are green and stay put when its red.

With regards to my second question, I have been very keen with the ticketing wizard and pleased with the kind of results it yields. Going through some of the results of the last season I noticed four of the predictions given on my ticketing wizard won but looking through the results on the blog with the same matches the results lost as there was a different prediction, I have read many times on the forum about prediction changing when starting lineup is announced but my question is, Does all the results of the predictions we see on the blog follow the annoucing of starting lineups?

 


This post was edited by spirit (2015-07-15 09:23, 4 years ago)




#2 Eugene


470 Posts

2015-07-15 05:59

1. We're updating our statistical models every season in order to improve overall efficiency, but still didn't success in finding the reason of our performance drop during Autumn and Spring weeks. We hope this season will be more stable, but if we have the same pattern 3rd season in a row that's very important information for us as well, since we will know for sure what periods we should avoid or work on in order to increase overall seasonal performance.

2. Yes, our History and blog results based on confirmed lineups only.




#3 spirit


13 Posts

2015-07-15 07:43

Thank you for getting back to me mr eugene. With regards to your answer to my question about stability of results, i think I agree we should hope for the best this season and prepare for the worse should history repeat itself. 

To your answer to my second question. I have read most times on the forum that results are mostly positive with confirmed lineup and I want to know if you have conducted any detail analyses to back that claim? We could be doing better maintaining the pre determined predictions of the software before lineup is confirmed but if going over results proof otherwise,then our best chance is with the confirmed lineups. My reasoning for asking if an analysis has been done to determine if we are doing better with confirmed lineup or not is because when you look at the screenshots below:

RESULTS WITH PRESUMED UNCONFIRMED LINEUP USING THE TICKETING WIZARD OF FOOTBE

RESULTS WITH CONFIRMED LINEUP AS SEEN ON THE BLOG

You would see that the match between Rostov vrs CSKA Moscow ended in a draw as predicted by the ticketing wizard which I assume was with unconfirmed lineup. And the same match was updated with a new prediction from confirmed lineup which ended with a lost as seen from the results on the blog for week 41 of the 2014/15 season.

You can also see the match involving Atalanta vrs AC Milan was successful with the prediction of the ticketing wizard with assumed unconfirmed lineup and lost with the confirmed lineup prediction results on the blog. Going further, the match between Cagliari vrs Udinese was successful with ticketing wizard and unsuccessful with confirmed lineup as shown with the results on the blog. Moving on, we see the match between Napoli vrs Lazio was also successful with presumed unconfirmed lineup with ticketing wizard prediction and lost with prediction with confirmed lineup. Well maybe am missing something here but you can tell the confirmed lineup prediction lost a total profits of +9.31 odds which could have turn around overall results of that week into a profitable one.

My general view is the ticketing wizard is great but I want to be sure if you have any analysis that could proof confirmed lineup results are successful as against unconfirmed ones. With the upcoming season ahead of us, we need to be looking at the right places to increase our success rate with this incredible app.




#4 Radimson


46 Posts

2015-07-15 08:51

Hi Spirit,

this could be caused simply by variance because you don't have enough data.

But what I would like to know from Eugene is if you have results with unconfirmed lineups for the whole season. Unfortunatelly I think you don't which is a shame for us. Because if these results were 1 or 2 percent worse than those with confirmed lineups I would happily take it and placed all bets every morning for the whole day.




#5 spirit


13 Posts

2015-07-15 09:20

#1120 Radimson:

Hi Spirit,

this could be caused simply by variance because you don't have enough data.

But what I would like to know from Eugene is if you have results with unconfirmed lineups for the whole season. Unfortunatelly I think you don't which is a shame for us. Because if these results were 1 or 2 percent worse than those with confirmed lineups I would happily take it and placed all bets every morning for the whole day.

Yes Radimson, you could be right and I stand to be corrected if there is enough data I dont currently have access to prove otherwise, besides we are trying to make the best out of the information we have from previous results going forward but not to delibrate on our shortcomings. I can fully make a suggestion based on the response from Eugene with regards to the data I could be wrong not to have access to in order to accept the claim of confirmed lineup generating good results as against unconfirmed ones.

Added 36 minutes later:

I dont like the timing of my question especially today that predictions are expected to be coming in and I strongly believe the time to come out with data proving confirmed lineup as against unconfirmed lineup is a luxury Eugene cant afford now, so a simple answer with a yes or no is ok with me for now. Will want to see prove to that later when there is time but its very important this unanswered questions in every bodys mind is dealt with once and for all to prevent any future curiouse minds from demanding answers. On this note I think we could be at peace with issues of this nature if we have a feature like the lineup wizard which I have suggested on the feature request page of this forum.


This post was edited by spirit (2015-07-15 10:18, 4 years ago)




#6 Eugene


470 Posts

2015-07-16 07:52

The most important thing in statistical modelling is consistency. When you create statistical model you make a lot of decisions in terms of data processing, choice of time period, variables selected, modelling approach etc. In order to achieve satisfactory results on future data using created statistical dependence, you have to follow exactly the same approach. Any deviation from the approach used during model development leads to distortion of the forecast, and thus deterioration of the results in terms of accuracy and profitability.

E.g. you can't apply statistical model developed using English Premier League data for Eredivisie games. You have to be 100% consistent. Our models developed using historical ACTUAL lineups data. So if you don't know starting lineups of future game for 100%, you can't expect same performance of the model achieved during it's development.

If you can guess lineups for 100% - you will receive the same figures, but any mistake in starting lineups inflicts model's accuracy - it's a rule. Of course, sometimes opposite trends can be observed on short periods of time, like in your example. But you can't make any assessments until suffucuent data is collected. 7 games won in a row mean nothing from statistical point of view.

We're doing our best to predict the most accurate starting lineups in order to achieve reasonable profiatbility on preliminary lineups as well. We understand that most of the people not able to track our predictions online. But still it's impossible to be 100% accurate in terms of starting lineups, so performance on preliminary lineups is lower. It provides with some reasonable profitability, but still lower than our reported results.

We've been tracking performance based on preliminary lineups last year, but it was too time consuming since this process was manually managed, so we don't have statistics for entire season. Anyway, we tracked it for quite a significant period of time (over half a year, between weeks 7-35), and collected 1,850 events in total which is enough to make conclusions. Below you can find comparison of balanced ticket wizard strategy based on the preliminary lineups to our reported results based on confirmed lineups during same time periods.

 

 

Of course you can observe better results for some weeks (especially #8), it's expected for long series of observations. But in terms of the long-term performance confirmed lineups provided with 2.5 times higher profit on the period analysed. Still you can see that accumulated profit lines are quite similar until March, but when we had extremely unsuccessful series during Spring, confirmed lineups performed much better. And in the end confirmed lineups accumulated +83.5 points while preliminary just +34.4 (see file attached).

Attachments

#1 zip

preliminary.zip

613 kB (Downloads: 447)


This post was edited by Eugene (2015-07-16 09:08, 4 years ago)




#7 spirit


13 Posts

2015-07-16 11:26

Hello Eugene gone through the informations you provided including the attached file and I must say even though I didnt digest the data you provided deeply enough I want to agree with your sentiments that you have enough data to conclude confirmed lineup provides higher profit than preliminary lineup. What I want to know is do you believe the greater majority of us (footbe users) get the previlage to apply the predictions with confirmed lineups? considering the timing of  updating lineup inside the app. And since the entire history and blog results are based on confirmed lineups only, do you believe users are able to make the profit margins (ROI) stipulated at the end of an entire season?

If your answer is yes, I wonder why you made this statement 'we understand that most of the people not able to track our predictions online" a statement I consider to be true based on the fact that there are lots of complaints on the forum regarding getting notified about updated lineups. Which simple means only few people manage to catch this updates and make higher profits provided by confirmed lineups.

If your answer is no, then I want to know what you are planing to do to make sure we are all on the same page and making good use of the updates of lineups. I would not have believed, if there was no records showing results on the blog that at some point the predictions were changed. When I saw the results I was alarmed and asked what if I got it wrong with the preliminary lineup prediction from the ticket wizard. And now that we accept that confirmed lineup provides higher profits, what measures are you putting in place to make sure we all catch this important updates before kickoff to matches.  




#8 Eugene


470 Posts

2015-07-17 05:02

I believe it depends on what exactly the betting for you. If it's just for fun, like a hobby, you can make bets on preliminary lineups without spending much time, and having lower results in terms profit.

If you really intend to make money on it - you should devote some time. And I know few people who live by betting, so it's work for them. And spending few hours during weekend worth 200% of yearly profit. I don't know any riskless and effortless way to make money.

Added 1 minute later:

And yes, I'm sure users are able to get profits reported in the blog following our recommendations online. I don't understand why you say that only few people can "catch" these updates. Indeed, we had some problems a couple of days due to technical issues, but it affected maybe 0.5% of games in total, it doesn't really affect long-term profit. There are over 40 weeks during a year. The most of games updated in the period 15-45 before game begins, it's not a problem to place a bet (except of the days we experienced problems).

We're still in beta and work hard to avoid problems like this in the future.


This post was edited by Eugene (2015-07-17 05:17, 4 years ago)




#9 spirit


13 Posts

2015-07-18 08:14

#1131 Eugene:

I believe it depends on what exactly the betting for you. If it's just for fun, like a hobby, you can make bets on preliminary lineups without spending much time, and having lower results in terms profit.

If you really intend to make money on it - you should devote some time. And I know few people who live by betting, so it's work for them. And spending few hours during weekend worth 200% of yearly profit. I don't know any riskless and effortless way to make money.

Added 1 minute later:

And yes, I'm sure users are able to get profits reported in the blog following our recommendations online. I don't understand why you say that only few people can "catch" these updates. Indeed, we had some problems a couple of days due to technical issues, but it affected maybe 0.5% of games in total, it doesn't really affect long-term profit. There are over 40 weeks during a year. The most of games updated in the period 15-45 before game begins, it's not a problem to place a bet (except of the days we experienced problems).

We're still in beta and work hard to avoid problems like this in the future.

Well said, I am always amased to see the kind of work and effort you and your team put in to keep a project like these running. Making the best out of it is my concern so I hope you dont get burdened  by the approach of my questions. There is a wise saying that goes like "better ask thrice than go astray once". With the strongest convictions expressed in your feedbacks I am at peace to make time and follow the confirmed lineup predictions but would want to know how  one can spot a confirmed lineup prediction within the software from a preliminary one. Are there any display that shows this prediction is a confirmed lineup prediction. I would be pleased if you can answer that question with some illustrative screenshots. Thank you.




#10 Eugene


470 Posts

2015-07-18 08:39

Matches with confirmed lineups have a specific icons on fixture screen:

Once new game with starting lineups are confirmed, you will see a notification bubble in the taskbar icon with the number of matches Footbe has a final prediction for. These matches will be highlighted on the "Fixtures" screen with blue tick icons next to the formation sketch.




#11 spirit


13 Posts

2015-07-18 09:05

Oh ok, this really makes alot of sense now. So that means this automatically updates the prediction of the confirmed lineup fixture on both the prediction tap and on the top bets of the week area of the software right?




#12 Eugene


470 Posts

2015-07-18 13:15

Yes, every screen where probabilities involved.




#13 spirit


13 Posts

2015-07-18 13:53

Ok thanks Mr Eugene, The extra feature I was requesting for was already in the app I just didnt notice it. Just saw the live update of confirmed lineup a few minutes ago. Well cheers to another great season of making profits with footbe. Thank you




#14 gabrijel


57 Posts

2015-10-05 09:45

Hi Eugene !

Just few questions ? What is your plane for upcoming week or maybe months about app ? 

Having played say 20-25 % of total matches for this season and more then 500 recommendations, I think we can make some conclusion in which directions we are going this season ?!  Minus is not big , in my case is 8-9 units , espacially when you can not catch odds of Marathonbet and I think vast majority of us can not catch odds from Marathonbet ( limitation problem )  and not only Marathonbet but because it is most frequent bookie ( but ok this is for some another topic ) . I know that this is game for strong nerves but this season app has much higher amplitude and much larger series of false predictions which further destroys the hope that we will get out from red ( I hope I am wrong ) . So I wonder if you have some guidelines for the rest of season ,maybe some changes, becuse now is short break and maybe you have time to do some analysis of previously made ? It is not time for alarm but I think that somethnig needs to change !




#15 Eugene


470 Posts

2015-10-07 07:49

Hi Gabriel,

We invested a lot of time in statistical models update, especially in top leagues where we changed approach completely. Now it's much more complicated and takes into account much more factors, it's applied to top-5 league + Champions League. But based on the current result we can not say we're satisfied with the new approach performance. After 11 weeks in these 6 leagues we had 281 recommendations, which generated a loss of -46.3 points, or average ROI is -16.5%. Awful result, but we're going to keep it testing for a couple of months more, to understand how it will perform on our 'black' period during November, where we have losses traditionally. We have some expectation that this advanced approach can perform better on this period than those we used during last 2 seasons.

Still we're qutie optimistic about our future, since in other 6 leagues where we use same approach, after 330 recommendations we have +41.2 points earned, or +12.5% average ROI, which is qutie close to our previous seasons results.

So if performance of new approach remains the same, we will switch top leagues to previously used.


This post was edited by Eugene (2015-10-07 08:40, 4 years ago)




 


Terms of UsePrivacy PolicyBlogForumContact
© Copyright 2017 footbe