Thursday, 27 August 2015

Is there a blind obsession with statistics in the betting world?

Third trading post and this one is a direct consequence of my incursion in the mighty world of Twitter. I already knew about betting advice based on statistical data, what I wasn’t expecting was the magnitude and the reach of it. Bear in mind I’m probably biased when it comes to this issue due to my research background, which involved statistics, data and computational modeling. So, let’s begin…

Would a person who never cooked be able to make a top dish, if given a large amount of ingredients, of variable quality, to choose from?

When it comes to cooking, there are at least two important parts of the equation: technique and ingredients quality. If you’re like me and don’t know a lot of cooking techniques, the only way you’ll succeed is by choosing good ingredients and somehow don’t spoil them in the process. So, what does this have to do with the use of statistics in the betting world? Probably there are better analogies, but my point is: we live in a world where the amount of data is huge and readily available for everyone. Adding to that, nowadays you don’t need to have a Math degree to somehow be able to extract some odds out of that data. That being said, I can assure you that if you don’t have an edge statistically-wise and you don’t know how to select quality data from the whole bunch, the odds you are cooking will suck!

Enough of cooking, let’s get down to business!

Fair enough, let’s start with the “problem” of the amount of data available. Manchester United are playing against Swansea on Sunday and with some tweaking here and there, you can select two sets of data: one that supports Swansea to win and other in favor of Man Utd. Furthermore, you can also find evidence supporting the over and the under for the same match. Politicians and economists with some agenda do this all the time, it’s all about cooking it the right way.

When it comes to the punting world, I understand the commercial side of it: you’re selling something and a bit of data can validate your tips. If you’re right, you’re the man, if you’re wrong blame the variance! 

This leads us to the other part of the problem, the selection of meaningful data. In science, if you want to obtain credible results, from an experience, the number of observations is an important issue. Furthermore, each observation must be obtained in the exact same circumstance as the others. So, transposing this to football databases, how can you reach any conclusion about Van Gaal’s Man Utd, based on historical data from Ferguson or Moyes tenures? Even the first Van Gaal matches are not meaningful in the present, as he experimented a lot in the beginning; ex: what does a 3-man defense with the likes of Tyler Blackett playing as to do with the current setup?

Football: the sport where results lie!

One of the reasons behind football popularity is the frequent unfair nature of the final result; in some sense, it’s a bit like trading sometimes: a team can take more EV+ decisions than the opponent and still lose or draw the match. Final results lie in football! 

As an example let’s take two 0-0 results from last week: Man Utd vs Newcastle and Arsenal vs Liverpool. In the future, these results will be used to support some Under bet or to model the odds of the Under/Over market. While in Man Utd game, the 0-0 can be considered a fairly true result, anyone who saw Arsenal vs Liverpool can’t possibly say that the 0-0 reflect what happened. Both GK had amazing performances, plus there was a goal unfairly ruled out for Arsenal and 3 shots off the woodwork. My point being: don’t follow or use stats blindly, they don’t tell the whole story!

Final remarks – Take advantage of stats and not the other way round!

As I stated in the introduction, my professional background probably introduces some bias in my views, as I know it’s relatively easy to drown in the sea of statistics available nowadays. 

Nonetheless, I’d like to state I truly believe it’s possible to take advantage of stats in order to score some profits out of the markets. What I don’t believe is that anybody can do it, just because they are available. As always, you have to have an edge: it can be your data selection, your statistical knowledge, your ability to construct customized data from the databases available or others I don’t even know about. 

As a final remark, you may ask me: “so, if you work with stats and data, why do you approach the markets with an intuitive-based approach?”. Fair question, but I guess I choose to follow the old proverb: “In the house of a blacksmith, the ornaments are made of wood”.


  1. Really interesting post Unknown,

    Thought I would be the first to comment as no one else has decided to take the plunge!

    "When it comes to the punting world, I understand the commercial side of it: you’re selling something and a bit of data can validate your tips. If you’re right, you’re the man, if you’re wrong blame the variance! "

    As this probably relates somewhat to me, considering I run a service on blogabet, this is true in some instances.

    For me however, I am completely aware that a bad run or being wrong is most likely variance, however as you touched on with historical results you always have to be aware, that what happened in the past won't automatically transfer to the future.

    As a result, it is why I pry over my statistics/data that I record on a, monthly, quarterly, yearly and 3 year basis. Yes that may seem quite anal, however I am well aware just because I am profitable now, doesn't necessarily mean I will be profitable in a year to come. Thus I need to be aware if results are deviating more than what I would class as the maximum/minimum amount that they should.

    However, reviewing my data regularly put's me in a much better position to be able to see if results are starting to trend differently over periods of time. Of course you have to be careful (not just me but in general) to make sure that the data you are using is still relevant.

    For example, I could look over the my results which I record and see how they have changed over the different periods. However I have to keep a watchful eye on the data that makes up my selections, to make sure that this is still relevant. If it isn't, then no matter how much I look over my recorded results, my edge may have been eroded as a particularly imporant piece of data that make's up my model, has changed.

    I'm probably babbling on a bit, hopefully it make's a little bit of sense.

    Whilst on the subject of statistical / historical data, I am currently contacting fracsoft to see if I can gather some additional tennis information (In-Play odds).

    This is because I think I have found an edge for straight betting in the in-play markets at certain points in a match, so hopefully with the historical odds data, I can atleast look for the past 2 years (about the same length as my results I have recorded for my tipping), to see if that edge does indeed exist, rather than just in my head.

    The reason I say 'I think' I have found an edge rather than have, is because my current recorded results for my tipping and all the other results I have recorded, have shown a particular trend which I believe I can take advantage of.

    One last thing however is what you wrote at the very end of your post: "so, if you work with stats and data, why do you approach the markets with an intuitive-based approach?”.

    I do find it interesting that you approach the markets with an intuitive-based approach. I also used to do a similar approach, however the more I get into statistical data, the more in my opinion, I find it outweights what I thought were advantages in intuitive-based trading.

    I wonder if that's just because I'm understanding statistical based betting/trading more then I ever did, or bias is starting to set in?

  2. First of all, thanks a lot for your feedback which I really appreciate!

    When it comes to my sports knowledge on the betting/trading world I’m more or less one-dimensional: football! So, some things I wrote may not apply so markedly to tennis, although there are surely common aspects.

    That sentence was not aimed at your kind of tips service, but for those who write several paragraphs with lots of historical data which is not relevant to the present situation. Also, and I’m sorry if my tennis knowledge lets me down here, tennis and football present different challenges when it comes to selecting valuable historical data. Football it’s a team sport where the manager plays a huge role; so, in a football match you have 22 players, 2 managers and to help increasing the variability, you don’t have a lot of technology to help the referees. As managers and players change a lot between seasons, it is very hard to find standardized data. Of course, in tennis you have different surfaces, the coaches also change but I think you can get a higher number of meaningful observations to the present situation.

    Furthermore, I don’t consider you to be anal about your results but pretty conscious and wise about the challenge you’re facing. It’s a tough world, where you have to perform a consistent auto-evaluation because results can lie in the short-term. Like I said in other post, you can lose with EV+ decisions and the other way round. So, I’d say you’re on the right path, constantly evaluating what makes your data relevant.

    About intuition vs statistics, my point is not advocating intuition over statistics but to offer some food for thought about it. Also, when I say intuition it’s not some psychic power but an acquired ability from thousands of hours in the markets. On a lighter note, it’s a bit like going to the market for some groceries. In the beginning, you’ll accept the prices they are offering as you don’t have a lot of price knowledge. After a lot of trips to the market, you start to notice the bargains and the overpriced items and your wallet becomes heavier =)

    To answer your last question, I would say bias can certainly play a role but if you’re being successful, I’d continue exploring the statistics world. Trading is all about finding a comfortable setup where you minimize your pshycological shortcomings, whether this setup is more statistical or intuitive is not that relevant.

    Hope I was able to give you more insight on what I meant with this post!

  3. "Hope I was able to give you more insight on what I meant with this post!"

    Yes you did and a great post it was in my opinion, it's a shame I don't have the writing ability of yourself, to write such posts.

    Hopefully we'll continue to see such interesting posts in the future.

    1. Thanks again! Also, if you have some suggestions or topics you would like me to write about, let me know.