Arbusers

Dropping odds... What a magical thing for those, who know what to seek and why...

So, here I'll write some thoughts in regard to that theme. I hope those will be useful to you.

Usually, most people are seeking the dropping odds using third parties services, but another possible approach is existing.

Just keep reading this article at www.arbusers.com.

Are you familiar with Math? All betting businesses really lie in this area. Many mathematical and statistical approaches are involved here. But among the initial steps here is a little jewel. I mean something called Probability Theory. For example, are you sure, that Messi or Haaland will score in the next soccer match? Yes/No is a very probable thing. Imagine, that the question is slightly modified now - Are you sure, that Messi/Haaland will score in the next three matches? Do you see the difference now? What about the question "Are you sure, that they will score in the next ten matches and how many"? And so on. The basic idea is that the next event won't be "predicted" with certainty, but when it comes to a given group of similar events, Probability Theory does the "predictive" magic here.

In other words, a lot of things in our lives, such as Weather forecasting, Economics and budgeting of a country projecting, and even a small business development projecting by the owner of that business in some future period, are relying on the magic things, called Maths and Logic. Especially they rely on Probability data string processing.

Now I want to turn your attention to Financial Science. What a surprise, it is based on this strange but useful Math... Have you ever seen a lot of graphs on some screens regarding Forex or Stock Exchanges, supported by Math? Have you heard of a lot of non-graphical kinds of analyses, backed by reasonable Logic?

Although various utilities provide your attention the dropping odds when they have already happened, my point here is that you are more likely to benefit through earlier projections to be faster than those services, through Math/Stats and Logic.

There are two things in a financial analysis called technical and fundamental analysis. Let's extrapolate some ideas to help our activity.


A basic analogy with technical analysis will be introduced later.

You've probably seen the simplest form of Statistics posted on the Internet. Let's see the football matches. Each team's possession percentages, number of shots, number of attacks, passes, tackles, number of corners, offsides, yellow cards, red cards, last five matches, etc. Are they useful? Yup and Nope at the same time.

Let's look at some of the popular myths:

Percentage of possession. Well, the basic idea is that a higher possession guarantees a higher chance of winning.

The number of shots. It's the same here. The popular belief is that the more hits, the more chances to win or score. As well as the number of corners.

The number of red cards. Significantly reduces the likelihood of a good ending.

If a team has an excellent streak of five consecutive victories, it is likely to be the winner or favorite in the next game.

And so on.

I will give you another perspective on these cases a bit further on in this helpful article on www.Arbusers.com…

Now let's recall technical analysis in financial science again. There are very well-chosen variables there that are handled appropriately and supported by mathematical/statistical algorithms. Therefore, if we apply this analogy here, we may increase our chance of winning. By this I mean the probability that falling (or rising) odds will be predicted before other people, bookmakers and services react.

Here I mean predicting with some percentage of accuracy. In reality, 100 percent accuracy is about a few things, everything else is uncertain. As Benjamin Franklin once said, "Nothing is certain in this world but death and taxes." Or if you prefer, this phrase was heard in the movie "Meet Joe Black".

Surely many people have tried to deal with these pretty elementary variables in some way.
The odds-makers who create odds and sell them to bookmakers have their own methods.

Now suppose we add a few more variables. Are the results better? I'm not sure. Some things are better and some things are worse. This is mainly due to two things: the noise in the raw data and the imperfect processing of that raw data. Remember that the odds-makers are hard at work non-stop. Remember, the published odds by design are stacked against you…

Therefore, before we add new variables to our model, let's make some improvements. How about slightly different data processing aimed at mitigating the noise? And if we try some improvements aimed at better processing only the valuable signal, if the noise is already mitigated to some extent?

I'm afraid you might introduce an improved signal-to-noise ratio into your model, mainly by carefully observing which variables work well and which don't. Let's just start processing them your way for a long time, shall we? Just start this. Trust me, then you will gain valuable experience.

But what is the connection with the title of this article? My idea is that all processed data at any given time can generate a probability of dropping odds. In fact, these can generate and predict the entire range of varying odds.

Now let's go back to the aforementioned "popular beliefs".
The number of shots. "More shots in our Stats, more chances for win or goals". Now imagine relying on total shots/shots on target in your model. Well, Man City regularly has plenty of those against the opposition team. Can we try to process some elementary raw data and generate some simple probabilities? Surely.

A hypothetical example.
Let's assume that before the visit of Manchester City against the opponent, there is no clear favorite. In the first 10 minutes, they have the upper hand with 5 shots on goal vs no shots for the opponent, the current result is still 0:0. Is it possible through probability calculations in some markets to anticipate some dropping odds before they occur? Surely. But what about the odds-makers? They don't sleep at this time.
But imagine, the shots ratio at that moment is 2:2 or even 0:1? What if Man City hosted here? What if Man City is a very strong favorite? How about some odds moves on the markets in these cases such as "Both Teams to Score", "Correct Score", "1 X 2 (home win, draw, away win)", "Asian Handicap", "A Given Player to Score at any time", "Over/Under" "Red/Yellow cards markets", "Corners Markets" and many more? Some of these markets might be affected within some seconds or minutes. Why?
How about your "predictions", based on your model processing this raw data at this point? Could you "predict" the probability of dropping odds in some markets, 30 seconds or 5 minutes before they occur? Why? Based on the model you are working with, can you "predict" the probability of a given range of dropping odds in some markets, 30 seconds or 5 minutes before they occur? Why?

And what about the reliability of these calculations based solely on that variable? Now, let's introduce now more stats to help our activity.
How about considering handling additional variables like Corners? What is their influence? Possession of the ball? If there are three red cards given around the 70th minute at a given score? Have the potential consequences associated with the previously predicted state changed and exactly how? Could you consider some implementations of Bayes' Theorem?

Several Real Examples.
Often an appropriate combination of some elementary variables may give a predictable result in typical widely-spreading pseudo-modeling of matches such as Man City vs Nottingham (Round 5th, held on 31st of August, 2022). But do you react faster than odds-makers to predict dropping odds that way?
Now see the possible inputted raw data, mentioned above, and the final results that happened in the matches like Everton vs Liverpool (3rd of September, 2022), Liverpool vs Brighton (1st of October, 2022), and Liverpool vs Man City (16th of October, 2022). Not to mention World Cup 2022 matches like Germany vs Japan and the notable clash between Argentine vs Saudi Arabia... What improvements to our model are needed to be set up here?

Of course, these are a pretty elementary technical combination of variables of 10 or more variables, generously published by bookmakers on their platforms and elsewhere on the Internet. Let's consider adding more variables. Let us now make some observations. Is the quality of the prediction model different? Later we add new metrics to our model, even custom-built ones. Naturally, the quality factor of each variable and its weight factor is different. And so on. Just experiment and observe, change some things and improve. Go ahead.
Did you know that the science of sports metrics and data is well-developed in US professional teams? How about 300 variables?

Well, due to several factors, Time Series Processing is well-suited here. All mathematical forecasting methods are OK. I advise you to read literature by Prof. Spyros Makridakis, as well as books by Prof. Nassim Taleb.

Consider factors such as the weight of money, the psychology of market participants, etc.

Regularly improve your model through long-term observations. As a result, one day you can regularly benefit from your "caught" value, worth the hundredths or tenths after the decimal point of given dropping odds you've calculated before they happen! Why not even more, from time to time?

Now consider going back to the Stock markets/Forex analogy.
Are your support and resistance curves well constructed? How do you build your probabilistic model to calculate the breakout moment of these lines? Are you fast at this activity? Test and backtest them patiently.
Are the exponential moving averages useful here? What if some of them cross each other? And so on...
Or in some cases simple moving averages are best and why?

Do you feel the creative power of the ancient Socratic approach to questioning as you seek to improve your model?

Modeling your activity "on steroids".
It's a big deal to significantly improve your business through something called fundamental analysis. For a further reference, research Forex/Stock Exchanges.

Everything is a consequence of something else. So the aforementioned enumerable raw data is only a consequence of the tasks assigned by the coaching staffs and performed by the available players.

Were you amazed on those occasions when a Goliath-like powerhouse like Manchester City was sometimes unable to beat a David-power-similar team? Please, observe the brilliant tactical battles between Guardiola and Hodgson in their meetings, especially the Hodgson approach on the 22nd of December, 2018, 2:3 at the end. Somewhere about 20 percent of ball possession for Crystal Palace away... Well, given their managerial decisions, when exactly did the dropping odds in some markets appear?

The same for Sevilla under Lopetegui's management before the 2022/23 season, when plenty of their winning games were dominated by an astonishingly low number of shots on target. The shots-on-target/total shots Ratio was even more shockingly low for a top-flight team, but ball possession far outstripped that of the opposition. And now, this pretty elementary technical combination of variables, combined with specific tasks set by Lopetegui, was an indication of huge probabilities of dropping odds in certain markets.

When you spot some fundamental things, you may react before some bookies regarding the dropping odds. Remember the usual set of tactical approaches of Marcelo Bielsa, exploding the Asian Handicap markets in early pre-match markets as well as during in-play markets. Another example of the early pre-match markets is Fabio Capello's particular firm tactical approach, whose logically delayed breakout at the middle and at the end of the Spanish championship provided observers with additional late hedge positions for a sure profit, via well-calculated probability for dropping odds. Especially his second stay there in Madrid.
And so on.

When it comes to in-play predictions on dropping odds probabilities, the game for us starts even before the match through in-depth pre-match analysis and accelerates as soon as the starting lineups are announced.

Therefore, combining technical and fundamental analysis in our model can ultimately help us take advantage of the dropping odds much better than before.
For example, the Greek soccer team Volos at the start of the 2021/22 and 2022/23 seasons, through proper fundamental analysis plus some curves built on technical analysis, provided some observers a significant value through well-predicted dropping odds in almost every major market, such as 1X2, Over/Under, Asian Handicap, Both Teams to Score, etc. Here I mean both early markets and in-running markets. Of course, all these markets for Volos happened slightly stretched in time.
This case was made possible by a unique combination of rare factors. This was underestimated by the bookies at this time, offering very generous odds in the markets.

What will the future bring us? To some extent, it depends on us. Mathematics and Logic can help with this.

I always find great value in your posts and I must thank you. I also find your posting style very balanced because they tease the reader without exposing to dangers. This is something I must work on because my posts are exposing to dangers in various ways.
There is a lot of work and hard earned experience behind your posts and you bring them to a position that inspires others without revealing your own niche. It would be naive to ask you about your niche so I will not do it. But reading with full attention these 3 posts creates a question and a comment at every line. I will start commenting here and I hope others with knowledge will follow. But thank you really.

campeones wrote: ↑

Tue Nov 29, 2022 10:03 am

I always find great value in your posts and I must thank you. I also find your posting style very balanced because they tease the reader without exposing to dangers. This is something I must work on because my posts are exposing to dangers in various ways.
There is a lot of work and hard earned experience behind your posts and you bring them to a position that inspires others without revealing your own niche. It would be naive to ask you about your niche so I will not do it. But reading with full attention these 3 posts creates a question and a comment at every line. I will start commenting here and I hope others with knowledge will follow. But thank you really.

Thanks for your words, Campeones.

I would be glad if I could give some inspiration back to the forum members. I learned a lot here on this website some time ago, processing some "sparkling" ideas here. But you are absolutely right, the algorithms are made by me. Regretfully the model is in perpetual Beta testing.

Assure you, that if you work with instrumentation, not made by you, your results won't be the same. If I work via your tools in your business in your way, I won't achieve the result exactly same as your ones. It's impossible. Yes, better or worse results are probable, but not the same. Imagine if Person X and Person Y are testing the same Lamborghini Miura on the Spa-Francorchamps. Even if they both are the same level of talented people, someone will be milliseconds away.
What if person X writes some ideas for the Miura tuning, and person Y counting her/his own driving style, improves some ideas for her/his Isdera Imperator tuning?

My recommendations.
Read both Authors mentioned in my previous posting in this thread.
Learn a lot about sporting metrics science in US universities. Every point of progress is only available via Science. Master your observation of every market. Improve your model. Test and backtest. Improve some things and do further observation. Test again, and so on. What happens when Science (a given toolkit) becomes Art (a vision for working with that toolkit)?

The endless questions in my previous posts in this thread are not mine. You have to ask yourself through the Socratic approach during your own creative process.
What about Real Madrid's scoring model under Leo Beenhakker? The defending model for that team under Capello? The differences in detail, when Radomir Antić coached Atlético Madrid, Real Madrid, and Barça? Is Barça the same under Cruyff, Rexach or Serra-Ferrer? Why? Data and Stats?
On-the-pitch approaches and Stats for various goalkeeper generations like Preud'homme-Byuo-Grobbelaar vs Toldo-Friedel-Meola?

I probably won't have any more educational comments in this thread. I wrote some introductory thoughts. Only time will tell if they are right or wrong. Hope these will help everyone improve their own model.
I replied in that thread due to my respect for you.

arb12 wrote: ↑

Mon Nov 28, 2022 10:31 am

Percentage of possession. Well, the basic idea is that a higher possession guarantees a higher chance of winning.

With just 17.7% of ball possession, Japan have won a World Cup game against megatherium Spain with the lowest possession figure in recorded WC history. For the history, Spain's ball possession was 74%.

Well done Arb12.

Before anything else, thank you Arb12 for the excellent piece of work. I feel the need I must comment to almost every paragraph you wrote, and I will certainly do for as long as I have some time.

arb12 wrote: ↑

Mon Nov 28, 2022 10:32 am

I'm afraid you might introduce an improved signal-to-noise ratio into your model, mainly by carefully observing which variables work well and which don't. Let's just start processing them your way for a long time, shall we?

I will pinpoint a similar approach called ''Via Negativa''. It is widely highlighted in Nassim Nicholas Taleb's work, but it is also mentioned several times in Charlie Munger's ''Wit and Wisdom''. Well, if they use it, it can't be wrong.

Via Negativa is essentially the study of what not to do. So instead of searching for what to do, it is easier and more perspicuous to search for what not to do. Like this, you take the noise out and you are only left with a handful of universal trues that actually works. In the Army, they usually say, the simple the plan, the better it is.

Over the years I studied 100s of projects. Those that work better are the plans that you can explain with a couple of sentences with the easiest mathematical calculations that could be done in a half page of paper.

Thanks for your good words, Mr. Arbusers.

I fully agree with what you've said. You're absolutely right that the KISS principle ("Keep It Short & Simple") is among the best possible approaches.

But there is a big trap for those newcomers who have never designed or built anything (technical systems, business systems, health and wellness projects, etc.). Albert Einstein once said: "Make everything as simple as possible, but not simpler". Any possible system has a "degree of simplification" limit, and if the inexperienced system creator applies the "Keep it short and simple" principle TOO LITERALLY, failures are possible later in the timeline.

The Creator/Operator of any high-performance system must consider at least three possible cases:
- Economic pressures for cheapening/cost reduction;
- "Black Swans" types of occurrences (according to Prof. Nassim Taleb's Theory);
- The underestimation of the importance of very good quality.
I think these three conditions are deeply related to each other.

For example, zero failures are tolerated when it comes to Nuclear Power Plants or critical and life-saving infrastructure of any kind. The Fukushima Daiichi Nuclear Power Plant had almost all possible protection systems (including against earthquakes), except for the extremely large waves caused by the tsunami.

An inverse example is also possible: sometimes the market doesn't appreciate the most valuable quality of the products and the projects disappeared later. I mean those systems/products/projects that are ahead of their time. Do you remember SAAB cars under Swedish ownership/management and their unique safety systems? Those systems are standard today, but SAAB Automobile AB is gone.

Or, if you prefer an earlier stage in History, Master Kolyo Ficheto (actually Nikola Fichev) accepts to be paid for the bridges after demonstrating to the authorities the full quality of his products. A story from his usual practice - the Ottoman army squads moved across the new bridge, while the creator remained under the bridge during the tests. I mean products created by a self-taught Craftsman, created in a much shorter time, cost much less compared to the technology of that era. In the end, they turn out to be extremely high-quality.

This story teaches us that the three problems mentioned above, which must be considered by the experienced system designer, could be mitigated simultaneously to some extent.

Now let's get back to the core topic.
My recommendations regarding the mentioned three problems for the system Creator:

- When building and setting up your system, don't cut costs. Here I mean mostly intellectual efforts for the development of your things and time, a lot of time for testing and subsequent adjustments.
I don't think expensive software and hardware are a must for your business at all unless you adopt the HFT-like method or some exceptional one.

- When it comes to the "Black Swans"-like events, only your experience could teach you.
I personally may have some clarity on what caused the outcome of Zalaegerszegi TE vs Manchester United 1:0 in the Champions League in terms of soccer, or even why Olivier Panis and the Ligier team won the F1 Monaco Grand Prix.
My approach to handling these cases is similar to the Japanese approach, recalling how they won Le Mans with a Mazda 787B fitted with a custom Wankel rotary engine.

- When you remove the modern cost-cutting memes widely spread across the businesses, you automatically increase the quality of your model in the Alpha stage. But turning it to Beta is not automatic. Keep in mind that most future efforts do not equate to the same progress in terms of quality improvement. Sort of like an exponential curve. Don't give up here at all.
Invest in data collection and processing every day. Go and watch some live tournaments. The things you will see and feel in live tournaments could replace the data from 15 monitors working simultaneously. Live tournament experience is invaluable to your model.

@ Arbusers,

I'll be very grateful if your trained eyesight and experienced point of view give me advice or some constructive criticism regarding the following cases of my practice.

Three of the players I watched live in some tournaments were Stan Wawrinka, Cristina-Andreea Mitu, and Ana Bogdan as well.

Stan Wawrinka was in an ATP tournament. It took me a few minutes of observation to realize that something strange was going on with him here, regardless of his good play as a top-heavy favorite here. He later lost. The minor injury he was playing with was reported much later.
My question here is was I wrong to have in mind some initial unproven short-time impressions for some trading? The dropping odds for the pre-match heavy underdog were sparkling here, my rationale for this "prediction" was a noted deviation from the usual premium Wawrinka's playing behavior. At that time, I wasn't a rookie in that business at all, that's why I followed my fundamental analysis.

Visiting some of the ITF tournaments, I spotted the potential of very low-ranked at that time Cristina-Andreea Mitu (far, far before her motherhood and her non-declared semi-retirement from tennis) and Ana Bogdan as well. I had no idea that one day they would be in the Top 70 and Top 50 respectively of the WTA rankings. I simply noticed their talent and some things in the model were ready when soon the victories over the Top 20 players came.
I'm referring to their future potential here, but the exact moment they explode up the WTA ranks in terms of dropping odds is the thing, I should have guessed a lot more accurately. But in general, I was OK here in trading, about Mitu and Bogdan's cases.

When I talk about Mitu's unannounced retirement, I remember Vera Zvonareva and Kimiko Date-Krumm's cases after their semi-retirement. That's why participated in the markets in the București Open doubles, I wasn't surprised at all of her success.

Naturally, most of the matches I watch on live streams or analyze via Live data, are decomposed by my own-made way of Stats. At that time, the monitors are irreplaceable.
I love to see the potential of future tennis stars. I'll never forget the way the rookie Rublev won the qualifying and later the Challenger somewhere in France, far far away from the Top 100. But the Exchanges went bonkers, including in terms of dropping odds.
Somewhere at that time (plus - minus a year or so, I don't remember in detail), I was impressed by the way, the future young star Tsitsipas refused to give up in the qualifying stages of some tournaments. Again, at that stage, I had no idea if he will be one day No 1 or no 50, but the eyes of the observer can spot some things in advance.
And so on and so forth. Not to mention that doubles tennis is extremely difficult to analyze. For example, Dallas Challenger Winners, the pre-tournament underdogs Nicolas Meister/Eric Quigley. Or some Harri Heliövaara/Henri Laaksonen appearances at some Challengers Tournaments and so on.

I forgot to mention the good return of Paula Kania-Choduń in some tournaments about a year or two ago after a long pause in mediocrity for a long time.

Only Wawrinka's close circle would new the injury he suffered. If you dont belong to it, then you should face the fact as a ''sh.t happens'', and add an extra margin in your calculations when ''sh.t happens''.

I also have in mind of some talents in various sports. The thing is that you can only take advantage of them only once or twice when a major upset happens. After these 2 times all planet will know something good is going on with them. The train is gone.

Tottenham Hotspurs vs. Man United finished minutes ago and eventually received 9 extra stoppage minutes in the end. Here don't count the first half's stoppage time...

Is it time for various additional in-depth Stats testings aimed at how exactly to re-adjust the probability percentage of very late goals, corners taken, some adjustments to various Asian Handicap lines, Both Teams To Score market (if they haven't scored by 90th minute), and other markets? Here I mean the point of no return in terms of increased additional stoppage times in the English Premier League during the initial matches this season 2023/24, more than 10 percent increased cumulative game play for sure (additional stoppage time at the end of the first half plus ST at the end of the match).

I've remembered some of the SS Lazio's unique ability from time to time to score later one or two goals after the 90th minute (under Inzaghi's management as far as I remember), for a turnaround several seasons ago.

Due to the fact that I have some activity when it comes to a vast range of soccer games in the Balkan Peninsula, such as Greece, Serbia, Romania, and Türkiye, additional projects through pure analogues can be opened elsewhere when soccer is guided by a set of similar sporting and psychological characteristics.

Minutes ago, my model alerted me that the zenith of a very complex short-term experimental project in the championship of Bulgaria had been reached and changes were likely from here on.
In that experiment, I modeled two teams from this country, named Slavia Sofia and Lokomotiv 1929.

Slavia Sofia is an extremely difficult team to model even in the short-term project due to many irregularities according to pre-selected data. They are a pretty suspectful team to me. So, I generally avoided them, but when they announced José María Bakero (yes, José María Bakero!) as the new coach of the team with the old mentality, and when odds compilers tend to publish screaming out-of-reality odds, this team went back to the matrix.
In addition, Lokomotiv1929 was put in that group for that short-term experimental project. Here odds compilers again overestimated various things, such as the capability of the defensive department as a whole, probably due to a good season in 2022/23 for that team under the previous coach, who knows...

In the Asian Handicap, Over/Under, Both Teams to Score, 1X2 and other markets, some of the market odds were screaming to be attacked for this group of two teams, provided that different families of data-driven curves, that describe their performance, after an initial good start, has tended to steadily deteriorate over the past few weeks, while published odds are far behind this trend…

Next matchday both teams are meeting at the ground.
Today I stopped that two-side-group experiment, awaiting a couple of sporting and tactical changes, probably in both coaching staffs and some players. The zenith of possibilities of that model was reached here.

"Modelling" things doesnt matter anymore, there isnt such thing as "modelling" teams if we talk about lower division teams. One important team news and odds will crash in no time or skyrocket so you wont be even able to catch decent odds considering low pre-ko limits these days. Doesnt matter what you think how particular team play, if we talk about non mainstream football any kind of analysing is absolutely pointless considering crucial odds impact due to team news.