Monday, 2 August 2021

Buying the Farm

On May 5th 1978, a young Warren Buffet (age 47) skipped the annual celebration of Cinco de Mayo to attend a seminar at Stanford Business School which was held as part of an Investment Management course being taught by Professor Jack McDonald. As of 2016, McDonald was still teaching the class.

I was reading through some notes from that day, and some of young Buffett's thoughts from that event seem to me to apply equally well to sports betting markets as they do to the financial markets that were being discussed that day. For example, his opening statement:

The game of making money in the stock market is deceptively simple. It is one of the few businesses where one makes offensive decisions and is not forced into making defensive ones. You play the game only when and where you wish to. You need only swing at the fat pitches which are over the plate and belly-button height.
In other words, be patient. Wait for the market to offer value and only play when it does. As punters or investors, we have the huge advantage of not having to bet in every event. Buffett gives an example of a farmer, who...
only sells or buys a farm once. Therefore, he must be right when he does. In contrast, the stock market gives tremendous choices. Most people turn this vast array of choice into a disadvantage rather than an advantage. Be patient and wait for the right buy.
In a theme which has recently been well documented by Morgan Housel in his excellent book The Psychology of Money which I mentioned in my last post, Buffett touches on the fact that when it comes to investing:
There is absolutely no correlation between hours worked or intelligence and money management success, but that there is a tremendous correlation between approach and temperament, and investment success.
As Housel puts it:

In what other industry [finance] does someone with no college degree, no training, no background, no formal experience and no connections massively outperform someone with the best education, the best training, and the best connections? I struggle to think of any.

Buffett lists four basic rules: 

1) resist the temptation to be in every game all the time 

2) maintain an even temperament 

3) have a reasonable knowledge of the subject and interest in it and 

4) be disciplined 

There are other nuggets hidden away in the notes, including the advice to "learn what you can’t value" and that he "would absolutely never invest in any of the largest 50 companies in the U.S. The market is incredibly efficient at this level."

When it comes to sports betting, I think the lesson is that accurately valuing the outcome of an event is very difficult, and when it comes to football or big sporting events, your opposition includes some highly sophisticated participants with access to far more useful data than yourself. 

One Tweet appeared on my timeline at the weekend from an account claiming to offer "Powerful tools for experienced players" and these words:

In the Advanced Goals Statistics module, select a live game then the score and the minute will be taken over automatically and used to calculate the next 5 minutes percent to see what happened on previous seasons similar games

Further, you can also apply other complex filters

Who knew that investing was so simple? I won't give any publicity to this vendor - you can find the thread on Twitter if you are interested - but of course the idea that this kind of thing is useful is based on the flawed logic that what happened in previous seasons has some kind of bearing on future matches and similar games. 

Common sense should tell you that Manchester United v Liverpool in August 2020 has no relevance to the same fixture played in May 2022 due to, for a start, the fact that players, tactics and sometimes managers may all be quite different a week apart and certainly a season apart.

So I asked the question:

Help me understand how what happened in a previous season with different players, managers, weather etc. has any predictive value?
It will come as no surprise that what followed was not an answer, but deflection - "I will help you if you tell me why for a given event (even live event) the odd 2.10 may be good to play and let's say 1.90 may not be good."

Oh dear. Expected Value is a basic concept for anyone taking betting seriously, but in exchange for my answer giving a clear example of why 2.1 might be value and 1.9 not, I never did get an answer as to how a game from a previous season could be relevant in any way to a current game. 

The software can be licensed for just 27 for 3 months and apparently lets you enter a game state and in return will give you historical outcomes, which might be interesting as a curiosity but is absolutely useless as a predictor for how a current match might play out. 

A couple more thoughts and observations. Entering the number of minutes elapsed is meaningless, because what is important is how many minutes remain in a game, not how many have elapsed as I have explained before. The 90th minute in a game with several minutes to be added on is quite different to the 90th minute in a game with none.

There's also the problem with these in-play bets that you are competing against some extremely sophisticated models which actually are profitable. The money they are consistently winning is the money from those who think that matches from previous seasons have any relevance. If profitable betting was really this easy, we'd all be millionaires and Premium Charge payers, but this use of historical data is not an approach that will get you to that point.     

No comments: