What is decent return?

[montmorencyt2w]

Another quick one , a guy called jack schwartz( i bel;ieve thats his name) did a lot of work ( and wrote a book, which i have ) in which he researched short term "seasonals" on ftse 100 index , i.e, which days of the year "traditionally "finished up /down etc with percentages and even down to half hours , days of week , days of month etc , i think if i remember rightly beginning and end of both week and month tend to be positive , he had allsorts in there , like xmas effect etc also

Check out "Mutual Fund Monday".

 

[henry766]

there allsorts of wiered and wonderfull seasonals out there , some probably coincidence , but I think one premise is that busy times tend to be positive ( closes and opens etc ) in general

 

[daxdaytrader]

A decent return is one that has a low probability of being possible by chance alone.

see http://www.evidencebasedta.com/

 

[bevok]

The problem with most of these so-called seasonal is that there is far too little data in the sample to call it statistically significant. A lot of it doesn't stand up to reason either. I can understand the logic of sell in May and go away, or the old "Voice from the Tomb" ags system, but buying the FTSE100 on the 12th of May because its gone up on that day 80% of the time in the past 25 years?

Looks like an interesting book.

 

[meanreversion]

The seasonals might give you a clue to entry, but what about position sizing and exits -- do they tell you about that as well? Sounds a bit flaky to me.

 

[bevok]

Indeed, but its like almost all of the books you see in the Investment section at a bookshop, or these systems on the internet - 70% accurate, 80% accurate. What, you mean price went up at some point in the following 30 days, five minutes, one year? Oh but it fell 20% before that? Absolutely without value the way its presented if you actually want to make money from it, rather than 'forecast'.

 

[meanreversion]

Trading system simply needs to have positive expectancy, with acceptable levels of drawdown. Percent winning trades is just not that relevant, but I think we've been over this many times before.

 

[jm99]

henry - you are quite right re worthwhile bets - its very much live testing only for now. Once I am satisfied I will increase betting to £100. Worst case is I loss £1k which is nothing these days is it?

I am just following advice from more experienced posters who advised to bet small to start with.

The beauty is while testing I can place the bet do a days work and not have to sit there watching the screen all day. Its all a bit of fun really at this stage.

Thanks for reading and posting on my thread.

so can you update us on how the 1st month of live testing went??

 

[jm99]

are u using ftse cash data for your backtesting?

and you never answered this question

the reason for asking was because if your using ftse cash data your testing will likely be inaccurate