hmsb4494
81 posts
msg #91393
- Ignore hmsb4494
modified |
4/19/2010 1:27:38 AM
kevin,
have you done anything with stops less than 10%?
|
Kevin_in_GA
4,599 posts
msg #91397
- Ignore Kevin_in_GA |
4/19/2010 8:18:59 AM
Question, say your plays trigger on a Monday, are you done for the week or do the picks from the filter still have life in them?
+++++++++++++++++
The statistical analysis underpinning this approach is based only on the Monday morning open price. If you want to try to play other stocks (or the same stock again) during the week, there is no analysis that would support it.
That does not mean it wouldn't work, it only means that this filter (and the results from it) are to be used as intended.
|
Kevin_in_GA
4,599 posts
msg #91398
- Ignore Kevin_in_GA |
4/19/2010 8:22:45 AM
kevin,
have you done anything with stops less than 10%?
+++++++++++++
I did look at tighter stops, but the original analysis was from late 2008, so lots of stops triggered. 10% seemed to be about the right seting (eyeballing it - not a "true" analysis).
What I don't like about this is that you are setting up for a reward/risk ratio of 0.5. The filter obviously works, but that is the one part of it that I am not really comfortable with.
|
hmsb4494
81 posts
msg #91420
- Ignore hmsb4494 |
4/19/2010 10:30:55 PM
yeah in my VERY modest account a 4/5 (80%) win rate gives a VERY,VERY modest profit.
|
Kevin_in_GA
4,599 posts
msg #91424
- Ignore Kevin_in_GA |
4/20/2010 7:25:23 AM
Not sure why ...
(5% + 5% + 5% + 5% - 10%) / 5 = 2%
per week.
104% per annum, without any compounding. How is this bad?
|
LebowskiUrbanAchievers
56 posts
msg #91425
- Ignore LebowskiUrbanAchievers |
4/20/2010 8:03:31 AM
MTG falls on news of common and prefferred stock sale.
Already down to 11% this AM. I am getting stopped out.
|
sbuck143
88 posts
msg #91426
- Ignore sbuck143 |
4/20/2010 9:06:58 AM
hmsb4494
- Ignore hmsb4494 4/19/2010 10:30:55 PM
yeah in my VERY modest account a 4/5 (80%) win rate gives a VERY,VERY modest profit.
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
This has everything to do with your position sizing, and nothing to do with the trading system itself. With a winning percentage of 80%, and a reward to risk of 0.5, the expectancy of the system is (as Kevin alludes) is 2% per trade.
This is an inherently profitable system, but it takes a keen mind to realize this is only half (and probably less than half) of the equation to make you money....the other factors are money mgmt and the ability to pyschologically withstand extended drawdowns.
Take for example a $5000 account that starts out with $500 on each position.
Trade 1. 5% win = $25 profit, acct value: $5025
Trade 2. 5% win = $25 profit, acct value: $5050
Trade 3. 5% win = $25 profit, acct value: $5075
Trade 4. 5% win = $25 profit, acct value: $5100
System is looking good, but so far after a month, you're only up $100 (before commissions). YOu've decided its the real deal, so you throw more money at it. For Trade 5, you put $2500 in the play, and you know what's comign next...
Trade 5, 10% LOSS = $250 loss, acct value: $4850
Trade 6, you keep the same position size, wanting to make up that loss. You figure the 80% system already had its one bad week out of 4.
Trade 6, 10% LOSS = $250 loss, acct value $4600.
You learn the error of your ways and go back to $500 on each position.
Trade 7. 5% win = $25 profit, acct value: $4625
Trade 8. 5% win = $25 profit, acct value: $4650
Trade 9. 5% win = $25 profit, acct value: $4675
Trade 10. 5% win = $25 profit, acct value: $4700
So at the end of 10 weeks, through nothing else than position sizing mistakes, driven mostly by greed and the lack of proper money mgmt, you are looking at a [b]6% loss.[/b]
Now lets contrast this to a strategy using consistent money mgmt:
If you kept the $500 position throughout, your account value at the end of 10 weeks is $5100....which exactly fits the expectancy of 2% on your $500 risk over 10 weeks.
|
Kevin_in_GA
4,599 posts
msg #91432
- Ignore Kevin_in_GA |
4/20/2010 12:04:42 PM
To build further off of the above, one should see this filter as a means to idenitify stocks with a statistical edge - feel free to change the stop loss to 5% or less.
Also, given that these are shooting for 5% per week, you could be happy with 4% (knowing that you are more even more likely to hit that target).
My backtesting was on only the top ranked stock - one bet per week. I used a 10% stop loss because of the unusually bad period of the intial backtest (late 2008 and early 2009), and the consistent selection of 2x inverse leveraged ETFs during that period (SKF, SDS, SMN) which needed some breathing room.
|
ciscoslaves
5 posts
msg #91433
- Ignore ciscoslaves |
4/20/2010 12:07:08 PM
Hi, been following the thread occasionally.
I like this general concept, although I think two things here could be improved:
#1 timing
#2 adjusting for general market conditions
#1 - I mean, instead of buying early monday morning, what about waiting for some sort of momentum indicator to turn bullish? Perhaps using the slow stoch.
#2 - The last 60 weeks or so has been one of the greatest rally's we have ever seen, and now that volality is gone, one might think to adjust the 5%/-10% to something like 4%/-4%. Thoughts?
|
sbuck143
88 posts
msg #91434
- Ignore sbuck143 |
4/20/2010 1:21:30 PM
ciscoslaves
- Ignore ciscoslaves 4/20/2010 12:07:08 PM
Hi, been following the thread occasionally.
I like this general concept, although I think two things here could be improved:
#1 timing
#2 adjusting for general market conditions
#1 - I mean, instead of buying early monday morning, what about waiting for some sort of momentum indicator to turn bullish? Perhaps using the slow stoch.
#2 - The last 60 weeks or so has been one of the greatest rally's we have ever seen, and now that volality is gone, one might think to adjust the 5%/-10% to something like 4%/-4%. Thoughts?
=======================================================================
The thing with buying early monday morning is that the set up, the criteria that put a stock on the list are based all off that reference price of Monday mornings open. If you buy at any other time, regardless of how 'good' you think you've refined it, you've officially moved outside the bounds and parameters of the system and the statistics are invalidated.
In other words, the 5% up range could be hit while waiting for your slow stoch to confirm the move. You get in, only to watch it drift back down.
Now what you might could do is add something to the criteria selection process, say ADX above a certain level , or maybe DI+ has to be higher than DI- by a certain amount, but that might filter out too many stocks.
|