Long Term or Short Term Stats?

[alexmr2]

I'm going to keep this as general as possible as I'm sure people will only openly discuss this to an extent. Personally I haven't delved much into stats and don't use them in my manual trading but I have been reading a lot of books recently and thinking about whether long-term or short-term stats would be more useful in trading sports markets, or perhaps both are in different situations?

Firsty long-term, take 12 months performance of a football team for example. My thoughts are that a bigger sample size would give a more reliable estimate of the true odds. If the team happened to have an unusually good week then we may see an overreaction in the market only for them to experience regression to the mean (reducing the luck element) and thus leaving some value on the table? On the other hand, does how well someone performed 11 months ago really have any relevance to how they will perform today?

As for short-term (last 2 weeks, month, 10 games etc) I would assume that this is what the majority of punters look at it? My guess is that they will be looking at the horses form or how well their team perfomed in the last couple of weeks when deciding what to bet on but this is obviously the opposite of what I just described when looking at long-term results.

This is all my own speculation of course, I would be interested to hear anyone thoughts and if it really is a case of getting the balance right if it is something which can be implemented into your trading

[ShaunWhite]

The older a stat gets the less relevant it is. Consider applying some sort of age/relevance weighting so you can still use large datasets without the oldest data being as dominant in your calcs as the newest.

[Atho55]

On the assumption that it`s in the Bookies interest to correctly estimate the odds and they probably have better qualified individuals doing it have you considered compiling your own odds using the bookie Home Win Odds against the Actual result of the match.

With a lump of data you can select the desired Home Win Odds then see all the historical actual results.

Footy Odds 2.2.jpg

This is for odds 2.2 and shows 131 results from the Premier League. By adding up the occurences you can determine the implied probability and the odds to compare against the odds being offered.

You are reliant of course on there being sufficient games having started at the Home Team odds you are looking for.