Frequently Asked Questions
 |
Questions strictly
related to LOTOFOOT are indicated by this green arrow, and if you're nor
interested by this game you can skip them.
Others are relevant for any kind of bets |
|
|
|
WARNING : most of the illustrations and examples below are relative to "13 games LOTOFOOT grid", as it was the rule until september 1st 2004. Today, there are 2 kinds of grids : with 7 and 15 (sometimes 14) games to predict. The site is mostly devoted to "7 games grids", |
|
|
|
WHY LIMIT MULTIPLE BETS TO 9 (on 15/14 LOTOFOOT) ?
The main reason is that succes probabilities quickly reach a limit while betting amounts exponentially rise (9 doubles=512 grids, 10 doubles = 1024 grids...).
To illustrate that, we have plotted below the expected mean result (dark blue curve) as a function of the number of multiple bets for a 13 games LOTOFOOT grid (FdJ published this kind of grid before september 1st 2004, and the results are similar for 14 or 15 games grids).
These are compared to the effective results of METEOFOOT picks (purple curve).
9 or 10 doubles (about 2500 F to 5000 F when base bet was 5F/grid, which would give approximately 500 to 1000€ with present rules) seems to be a not to cross limit (at least for individuals, groups of players can eventually afford higher bets).
"FOOT" stands for "SOCCER" in french.
"METEO" stands for "Weather forecast"
The "FOOT" part of METEOFOOT seems straightforward as the site is devoted to soccer and soccer punting.
We were looking for a name which would suggest the statistics, picks and predictions part of the site. We thus thought to "pronofoot" (litteraly "soccerpicks") but the name was already taken by a site we advise to pay a visit.
"statifoot" (we could translate as "soccerstats") could have worked, but sounded unpleasant to our french ears. Plus, we were afraid to deter people who are not fond of stats and mathematics from visiting us...
... or on the other hand to attract pure statisticians that may not be soccer fans (if some exist, please let us know).
At the end of the brainstorming session, meteofoot was considered as an acceptable compromise :
with "météo" comes the idea of prediction ,
the name is easy to remember (at least frof french speaking people)
soccer matches predictions are as risky as weather prediction !!!
guarantee to be the only one to claim for the ownership of the domain ( who else the hell can have such a stupid idea to mate so unrelated words !!!)
by the way météo is a word rather frequently seeked by webcrawlers (but much less rude than some others)
1°) example for 13 games to predict
This index measures two different elements (according to different point of views) :
First, the quality of the statistical prediction of 1 X 2 outcomes for the LOTOFOOT (assuming the effective results of one championship day are realistic"). In that case, the closer the index to 13, the better the estimation of matches results probabilities.
Second (considering METEOFOOT probabilities are correct) the surprises level of the day (which means such matches ending as the 18th team defeating the first ranking at home etc...).
In that case, the further the index from 13, the more surprises we got. And the earnings of LOTOFOOT will be high, but the winners will be scarce (and you most probably will not be one of them
As an indication, the nominal rating of random picks is 4.33
If we look at the curves of 2001/2002 season below, the mean rating of METEOFOOT picks is 1.2 point above "au hasard" ("random") picks. This last one is above its theoretical 4.33 value because of some days where delayed games were considered "good pick whatever pick". This mostly occurs during winter days as some stadium froze and cannot guarantee safe playing.
2°) example for 7, 14 or 15 games to predict
All preceding explanations still hold, except the target values for the index (which obviously becomes 7, 14 or 15 instead of 13) and the theoretical value of random picks which becomes 2.33 (LOTOFOOT 7), 4.66 (LOTOFOOT 14) or 5.00 (LOTOFOOT 15)
HOW SHOULD I READ THE "CLASSEMENT" CHARTS
The red curves is the "classical" ranking which appears in all publications and radio or TV programs at the end of each championship day.
The green curve uses the stats computed here to make a ranking extrapolation to the end of the championship. The main interest of this estimation, excepting an answer to the usual "where will my team finish" question, is to improve the readability of the championship by discarding all random events that can impair it, for example :
a particularly favorable calendar which makes one team receive all the weakiest ones, while stronger teams meet eachother (or on the contrary a total lack of chance which makes one team go play with the top teams for its first matches of the championship)
The weather hazards which sometimes oblige to delay some matches and make consequently the classical ranking non representative
etc...
To illustrate, the chart hereunder shows the 2001/2002 history of NANCY (french Ligue 2). After the 26th day (february 2nd 2002), the classical ranking finds NANCY at a concerning 15th place, which can be explained by NANCY having played 3 matches less than other teams.
The green curve shows however that METEOFOOT picks estimate NANCY to be worth 8th to 10th at the same period of time. Which will be confirmed at the end, NANCY finishing 9th.
HOW SHOULD I READ THE "POINTS" CHARTS ?
The idea is roughly the same as for the "CLASSEMENT" charts. The red curves however slightly differ. They are not exactly the points at a given day of the championship, but a crude estimation of the final points, assuming the average points earned to that day will be constant throughout the championship (and thus accounting for the effective number of matches played so far).
HOW SHOULD I READ THE "PICKS FOR THE NEXT DAY" ?
These charts show the evolution of the "home win" probability for each given match.
At the end of each championship game, the relative "skill level" of the teams is updated. For a given match of a given day (hereunder a chart extracted from the french Ligue1 34th day of 2001/2002 season), it is possible to build "a posteriori" the home win probability history, had these teams met one of the championship days before.
According to the chart, SOCHAUX defeating MONACO would have been a high probability event if both teams had played against eachother at the beginning of the championship (mainly due to MONACO terrific start during this season rather than SOCHAUX good performances).
A few months later, when they effectively met, the situation had changed as our estimations of SOCHAUX odds sensibly decreased. The final résult (3-0) was nevertheless compliant with the odds (60% for SOCHAUX home win).
NEW SINCE 2004/005 season:
A red curve has been added to the green curve on these charts. This new curve intends to give a more dynamic illustration of the teams' relative level. (see "Le Mans - Lens" chart underneath).
The green curve takes in fact ALL the games played so far into account to estimate the two teams realtive performances. That means that at the end of a season, the games played in summer and autumn, or the games played lately, have an equal importance. This may induce a non relevant estimation of the late teams' performances (which can be much better or worse than at the championship start).
The red curve on the contrary is based on the last TEN days before the day for which the picks are computed, and should reflect more consistently the effective level of performance of the teams.
Exemple below (predictions for the last day of 2003/2004 french Ligue 1 season) : it can be observed that Le Mans (freshly promoted from Ligue 2) had two good periods, during which it could have succesfully compete with such a good team as Lens (acustomed to this level of competition for a long time). The second period occurred at the very ending of the season, when Le Mans was closed to save its place in Ligue 1. And this last game was no disappointment for Le Mans as they won 3-0, confirming a playing level much higher than the season average. This was unhappily not sufficient, as Le Mans went down to Ligue 2. We are waiting for you again "les Manceaux"
Last minute : Le Mans is back again for 2005/2006 season !
HOW SHOULD I READ "LOTO FOOT® PREDICTIONS"?
These predictions (computed by the code developped here) give the games outcome probabilities for LOTOFOOT lottery AND all other games in published championships (which in 2005 are France Ligue1 & Ligue 2 plus England Premiership)
On the chart below (2001/2002 french Ligue 1championship again), there is an estimation of 47% probability in favor of a LYON home win, about 14% for LENS away win , and 39% for a draw.
Just for your information, the match ended 3 - 1 (34th day of 2001/2002 season and first champion title for LYON)
WHAT IS THE MEANING OF "PICKS PROBABILITIES"
1°) example for 13 games to predict
For each championship day, 3 LOTOFOOT picks with 9 doubles are proposed in this site (which means 512 single LOTOFOOT picks for each of the 3 multiple LOTOFOOT picks). These 3 grids are computed according to the predictions of game results probabilities as the one depicted above.
With these predictions, an estimate of the expected results (from 0 to 13) probabilities of each 3 grids is computed. These probabilities can also be seen as the probability for each result to be the maximum among all 512 grids.
In the chart below, it can be seen that 7 (good results among 13) is the smaller significantly probable result. From which we conclude that with our picks, there is absolutlely no chance to have less than 7 good results.
If we continue the analysis of the chart, and add all probabilities between 0 and 8, we get something about 3%. This can be translated to "there is 97% chances that the best result among 512 single grids will be between 9 and 13". This information is important as you can only win if you have at least 10 good results.
In reality (2002/2003 Ligue 1 season, 34th day), the proposed pick gave 11 good results (the probability to have 11 good results or more was 68.2% : not sure, but very probable !).
QUESTION : Are there some times when this probabilistic analysis fails ?
ANSWER : sometimes effectively, but did you really think I would have chosen these failures on purpose to illustrate the concept ! So what !
2°) example for 7, 14 or 15 games to predict
Everything is similar to the preceding explanation with LOTOFOOT 13, except that the results are to be taken within the (0;7), (0;14) or (0;15) intervals
WHAT IS THE MEANING OF RESULTS REPARTITION CHART ?
1°) example for 13 games to predict
When the championship day is over, and the LOTOFOOT picks results are known, this kind of chart gives the repartition of the 512 grids results.
On the graph below, it can be seen that there were1 grid with 11 good results, 9 grids with 10 good results, aso..
2°) example for 7, 14 or 15 games to predict
Everything is similar to the preceding explanation with LOTOFOOT 13, except that the results are to be taken within the (0;7), (0;14) or (0;15) intervals
I CANNOT FIND NEXT DAY LOTOFOOT PICKS
If this is the case, four possibilities may be considered :
| 1 |
The LOTOFOOT lottery has selected cup games (this the most frequent issue). METEOFOOT does not wish to make predictions on this type of match, because estimation of teams relative force is most of the time difficult to infer as they often do not play in the same league |
||
| 2 |
The webmaster is off for holidays (it
sometimes happens |
 |
|
| 3 |
the championship has started, but less than 4 rounds have been played. METEOFOOT does not want to make picks before enough games have been played to make teams relative force estimation relevant |
||
| 4 |
or more simply, the championship is
over |
||
HOW CAN BOOKMAKERS ALWAYS WIN AGAINST PLAYERS ?
The explanation is not complicated in itself, but the demonstration requires some statistics developments.
To stay understandable : for each game, bookmakers publish three numbers (x1,x2,x3) that permit to compute the player's earnings according to the final result of the game (1,X,2). So if I bet 1€, and the result is a home win, I will geti x1 €, and my net earning will be 1-x1 €. (nothing new until here).
If the bets were fair, xi and results probabilities pi values should be linked by the following simple relation ::
For example, if bookmaker's results probabilities were estimated to be 50% for 1, 30% for X, and 20% for 2, the odds should be x1=2, x2=3.33 and x3=5.
Generaly speaking, we should always have :
In practice, this never occurs, because bookmakers reduce the xi (to give less back to
punters, and 
is always greater than 1.
Examples extracted from real bets :
online bookmaker : 2.7, 2.9, 2.55 which gives 1/x1 + 1/x2 + 1/x3 = 1.107
La Française des Jeux for the same match : 2.7,2.65,2 gives 1/x1 + 1/x2 + 1/x3 = 1.25
info : Française des Jeux is the only one authorized bookmaker in France, and organizes a game called "Cote & Match" which is very similar to bookmakersOn some websites wich explain these kinds of odds manipulations, the bookmaker margin is said to be 10.7%, and FdJ margin 25%.
THIS NOT EXACTLY TRUE
In fact, real margins are 9.7% (not very different, but 1% of millions € is not negligible) and 19.9% (still comfortable but less than previous figure), because the true margin expression is not (1/x1+1/x2+1/x3) - 1, but looks rather like :
In order to explain the difference, a statistical demonstration can be found in the pdf file hereunder
.
A topic on this subject is opened on the "english forum" , "bookmakers" thread : do not hesitate to post your comments
), 