An interesting piece in the inbox this morning: -
http://www.safalniveshak.com/latticewor ... l-numbers/
'Law' of small numbers
Who's to say phelps won't false start and get eliminated?If you were observing Michael Phelps, the swimming legend, compete against few amateur swimmers, even few observations would be enough to make a generalization about future outcomes of such competitive events i.e., Michael Phelps will trounce each one of them every single time. Here you don’t need to worry about the law of small numbers. Why?
The answer lies in understanding the role of skill and luck in any activity.
-
- Posts: 78
- Joined: Sun Nov 17, 2013 11:58 am
Peter,
When applying the theory to trading, what volume of back testing would you say is needed to prove or disprove a method. Additionally if you were scalping pre race, how many races would you need to look at to see if you had "cracked" it, opposed to just being lucky?
When applying the theory to trading, what volume of back testing would you say is needed to prove or disprove a method. Additionally if you were scalping pre race, how many races would you need to look at to see if you had "cracked" it, opposed to just being lucky?
All the strategies I have developed have reverted to mean fairly quickly. So I look at the standard deviation of returns and monitor that. When the stabilise I know I am getting valid results
Peter,
I don't understand 'mean reversion'..indeed i don't understand the entire sentence above! Could you explain more fully please?
Many thanks...
Peter,
I don't understand 'mean reversion'..indeed i don't understand the entire sentence above! Could you explain more fully please?
Many thanks...
Here you go:-Hambleden wrote:All the strategies I have developed have reverted to mean fairly quickly. So I look at the standard deviation of returns and monitor that. When the stabilise I know I am getting valid results
Peter,
I don't understand 'mean reversion'..indeed i don't understand the entire sentence above! Could you explain more fully please?
Many thanks...
https://en.wikipedia.org/wiki/Mean_reve ... finance%29
Hah, I guess he didn't specify which Peter he asked
Hambleden, there's nothing wrong with asking but you should always try to find answers on your own first, it really only takes a few seconds to type "mean reversion" in Google search and there's your answer. It's much faster than asking on forums and waiting for somebody to google it for you. Not trying to be rude, just saying that it's very important to be self-reliant in this business.
Hambleden, there's nothing wrong with asking but you should always try to find answers on your own first, it really only takes a few seconds to type "mean reversion" in Google search and there's your answer. It's much faster than asking on forums and waiting for somebody to google it for you. Not trying to be rude, just saying that it's very important to be self-reliant in this business.
- marksmeets302
- Posts: 527
- Joined: Thu Dec 10, 2009 4:37 pm
I think what Peter ("Euler") means is that the distribution of his returns fairly quickly converges. In other words, after a couple of runs he will have a good idea if he's onto something or not - the returns (profit or loss) will center around a positive or negative number. There are statistical tools to determine the confidence you can have in your returns to be in a certain range. Usually these calculations are not necessary though, because your gut will tell you if you can trade a certain strategy. The article Peter refers to is a warning that your gut may be basing this decision on too few observations.I don't understand 'mean reversion'..indeed i don't understand the entire sentence above! Could you explain more fully please?
This is different from mean-reversion as a strategy - here you assume prices will wiggle around some value, instead of trending in one direction.
Hi KaiKai wrote:Hah, I guess he didn't specify which Peter he asked
Hambleden, there's nothing wrong with asking but you should always try to find answers on your own first, it really only takes a few seconds to type "mean reversion" in Google search and there's your answer. It's much faster than asking on forums and waiting for somebody to google it for you. Not trying to be rude, just saying that it's very important to be self-reliant in this business.
Yep I knew it was Peter Webb..I just saw it and popped a reply up..
Also worth typing in "Mean Reversion" in the betangel search tool above..there was debate some time ago regardinng mean reversion in the context of pre race odds too...worth a read
Regards
Peter
viewtopic.php?f=42&t=2285&hilit=mean+reversion
- marksmeets302
- Posts: 527
- Joined: Thu Dec 10, 2009 4:37 pm
yesterday one of my strategies that usually has 70% winners had 70% losers on 24 races. Had I deployed it yesterday for the first time I'd thrown it out.
Made me wonder how many strategies I've dismissed too soon.
Made me wonder how many strategies I've dismissed too soon.
We all probably have had loads, when you try something and it goes on a long losing run it's hard to keep going with it even thou it could end up beening the best one you have ever tried.marksmeets302 wrote:yesterday one of my strategies that usually has 70% winners had 70% losers on 24 races. Had I deployed it yesterday for the first time I'd thrown it out.
Made me wonder how many strategies I've dismissed too soon.
Burdo,burdo77 wrote:It's easy to throw away strategies that could be winning ones. I have data to suggest I have an edge, do you think I can get my settings right in automation? Nup, causing me a headache ATM!
What I tend to do when Im building a strategy (before I even touch the keyboard) is to draw it out on a piece of paper (like a flowchart). Im very much a visual person and this works for me, these flow charts can become enormous. Once I'm happy with it, I start to build it (in practice mode) from the ground up. (I have more than one Betfair account so this doesn't stop me doing my normal trading). If it then functions as I would expect and I have my POC (proof of concept), I switch it to the live environment for further testing and enhancement.
Funnily enough, I did exactly this on a new strategy yesterday and went to bed thinking about it last night (sad I know!), when I woke up this morning i came up a brilliant idea (i hope!) , so Ill throw that into the mix today.
What I'll do is run two betfair accounts, one with the spreadsheet from yesterday and one with the new features and then compare/contrast/refine.
Sorry Ive waffled on a a bit there , and you may already be doing this, but it may help a newbie
regards
Peter