Yes when scaling into continuation you have to put in 1/2 as much as your original postion http://www.amazon.com/gp/reader/007...s=average+half+as+much&go.x=15&go.y=13&go=Go!
Half of what? 15% 10% 5% Please clarify the percentage per intial position....also is this a backtested percentage based on position sizing or...just curious...
My 2 cents: When trading with long term timeframes (say daily or weekly), after any reasonable growth of capital curve by a certain percent per day/ week, pyramiding would be much less risky. Probably many traders would do it.
======= ElectricSevant/mrs; Yes on pyramid in bull maket, less in a bear market . However its done , keeping in mind , that say on 15 day average derivative position, early work is best - do not be late. Late is bad. Also do it differently in different seasons, like frost/hunting season. The plan of the diligent tend only to advantage, Solomon,trader king.
The case for pyramiding (and averaging) is extremely simple and difficult to refute. Let us start with an obvious assertion - one should allocate more capital to very good opportunities than to more marginal opportunities. Then move on to a widely accepted assumption - a >0 number of trades will, at some point before exit, experience an increase in trade expectation. From these two claims we necessarily deduce that at some point in some trades one ought to increase position size - in other words, one should "pyramid" or "average". I challenge anyone who disagrees to find the flaw in these assumptions or reasoning.
"one should allocate more capital to very good opportunities than to more marginal opportunities." yes this is a nice assumption. Can this situation be identified at the time or is this assumed as well. And how does this relate to the increased risk exposure or the risk of ruin, as position increases after an equity decrease. simulation work may give more fruitful answers than ill founded assumptions.
I don't have strict rules for pyramiding. Usualy if i can successfuly catch a higher low (or lower high if i'm short), confirming that the original trend is still going, then i add a little at this moment. I add half of my original position to keep my average price low. New stop is set more tight. Can repeat the process many times if there are opportunities.
In experimenting with position sizing in my own backtesting, I found that you can manipulate the variance, drawdown, and bottom line profits. It's just a math thing. I have never seen the math before, but i'm sure someone's done it (maybe someone can provide a reference). I just tried on a napkin and it got really complicated really quickly. It's gonna take some more thinkin' to work out a simple model. For my current trading (cherry picking with intraday holding, ~15 trades a month) I just want to max out the profits, since I am not trading so big anyway. In my backtesting I found entering and exiting full size does the trick here. I found to drastically reduce the risk I should take 1/3 off after a little move my way, but this disproportionately reduced the bottom-line profits. Adding came somewhere in between the two cases. One thing is clear though. An argument without numbers and context doesn't mean a damn thing.
Kaufman in his book "Trading systems and methods" provides some interesting numbers and charts to analyse various types of pyramid methods together with informative risk/ reward ratio and equity curve. Jones' "The trading game" also analyses it with example data.
If a tradeable edge can be identified - an assumption upon which all simulation testing, indeed all trading, relies - then clearly some sort of trading expectation can be identified in real time. The only further assumption needed is that there is more than one edge (or that the edge changes over time during the trade holding period) and that they are not identical in expectation. So, if more than one tradeable edge exists, and they do not have similar expectation, then yes, one can indeed identify the situation at the time. Position-sizing from a risk-control point of view is done just the same as normal. I.e. the risk/reward and expectation of the trade is calculated, then an appropriate position size taken relative to the current market-to-market equity base. In some cases the loss will be sufficiently large that the increased trade expectation will not justify increasing size, and may even require one to decrease it. But in other cases the increased expectation may justify increasing the position even on a somewhat reduced equity base. For example if your trade expectation quintuples, and you are down 0.1% on the trade, it would obviously be foolhardy not to increase size.