 |
benwm
Registered: Mar 2008
Posts: 1309 |
02-25-11 10:13 AM
1a2b3cppp
Are you talking about martingaling into individual stocks or general indices? And perhaps you could describe a little more specifically how exactly this works...
For instance, how many points in ES do you add, or how big a decline for individual stocks do you add (so we are all on the same songsheet assume initial entry price $100)?
eg. Do you buy 100, 95, 90, 85 in size 1,2,4,8,..?
or buy 100, 95, 90, 85 in size 1,1,1,1,..?
or something else assuming 1st purchase at 100 in 1 lot...?
With more specifics perhaps we can give you / us a better understanding. Generally, it is a flawed strategy, but if you KNEW that a particular stock was a winner, that is your edge, this knowledge. But in this case, it is still a poor strategy relative to others because you are limiting your upside to a small % return of the amount you have allocated to the strategy.
|
| |
|
Edit/Delete • Quote • Complain |
nLepwa
Registered: Feb 2010
Posts: 301 |
02-25-11 11:43 AM
Quote from abattia:
Would it be accurate to rephrase this as follows?
"If your mean reversion edge works, averaging down can make sense."
Yes.
Ninna
|
| |
|
Edit/Delete • Quote • Complain |
PocketChange
Registered: Jul 2008
Posts: 2036 |
02-25-11 12:42 PM
Martingaling / Averaging down using exponential scaling is unrealistic outside of a tight range. 1,2,4,8,16,32... 128 or 1,3,9,27.
Averaging down in Linear progressions at fixed intervals (1,1,1,1,1) will expand the range your capital can cover but requires a 50% retracement to break even.
Play around with a combination of a 1 step average down with pyramiding exits with directional stop and reverse.
Lets say ES is 1310.
Your strategy is flip flopping for a break out inside a tight range.
Entry: Long 1 @ 1310: Target Exit 1312: (Net 2)
Average Down: Long 2 @ 1308 Target Exit: 1309 (Net 1)
Stop -3 @ 1307 (Net -5)
Reverse Short 1 @ 1307: Target Exit 1300 (Net 2)
Average Down: Short 2 @ 1309: Target Exit: 1306 (Net 2)
Stop +3 @ 1310 (Net -10)
Every time you flip and reverse you take a 5+ point stop.
Once you reach your target exit you employ a pyramiding trailing exit strategy. After your 3rd flip you increase the initial packet size from 1 to N+1.
A modest amount of capital $10K can trade this strategy for 7 flips before busting. The winning ratio is high ... 99.1% or something close to that.
You eventually lose money if you just take a fixed 1 or 2 point scalp without pyramiding the exits for potential rainmaker. You risk the bank once beyond 9 flips in draw down and trading size. Its a thing of beauty when you catch a 10 point move scaling size.
This algo works well for volatile markets but needs to hibernate when the market goes flat to keep from flip flopping.
Each play is a sequence of trades... and for all practical purposes is a $10k bet that the market will break out of a 3 point trading range before crossing the entry and hitting the edges 7 times.
In Vegas terms you are betting that the roulette table doesn't precisely hit RRR, BBB, RRR, BBB, RRR, BBB, RRR.
|
| |
|
Edit/Delete • Quote • Complain |
nLepwa
Registered: Feb 2010
Posts: 301 |
02-25-11 12:50 PM
Quote from PocketChange:
Martingaling / Averaging down using exponential scaling is unrealistic outside of a tight range. 1,2,4,8,16,32... 128 or 1,3,9,27.
Averaging down in Linear progressions at fixed intervals (1,1,1,1,1) will expand the range your capital can cover but requires a 50% retracement to break even.
Play around with a combination of a 1 step average down with pyramiding exits with directional stop and reverse.
Lets say ES is 1310.
Your strategy is flip flopping for a break out inside a tight range.
Entry: Long 1 @ 1310: Target Exit 1312: (Net 2)
Average Down: Long 2 @ 1308 Target Exit: 1309 (Net 1)
Stop -3 @ 1307 (Net -5)
Reverse Short 1 @ 1307: Target Exit 1300 (Net 2)
Average Down: Short 2 @ 1309: Target Exit: 1306 (Net 2)
Stop +3 @ 1310 (Net -10)
Every time you flip and reverse you take a 5+ point stop.
Once you reach your target exit you employ a pyramiding trailing exit strategy. After your 3rd flip you increase the initial packet size from 1 to N+1.
A modest amount of capital $10K can trade this strategy for 7 flips before busting. The winning ratio is high ... 99.1% or something close to that.
You eventually lose money if you just take a fixed 1 or 2 point scalp without pyramiding the exits for potential rainmaker. You risk the bank once beyond 9 flips in draw down and trading size. Its a thing of beauty when you catch a 10 point move scaling size.
This algo works well for volatile markets but needs to hibernate when the market goes flat to keep from flip flopping.
Each play is a sequence of trades... and for all practical purposes is a $10k bet that the market will break out of a 3 point trading range before crossing the entry and hitting the edges 7 times.
In Vegas terms you are betting that the roulette table doesn't precisely hit RRR, BBB, RRR, BBB, RRR, BBB, RRR.
This strategy works well when there's a (relatively) strong bias in the underlying. Else you're breaking even minus commissions.
But since most assets tend to trend more than normal it works when the volatiliy is high.
If you can keep trading costs under control and can diversify on a lot of assets and then maximize capital utilization then the equity curve looks good ok.
Your execution needs to be perfect however. It would be very hard to trade for retail traders.
Also you need to dynamically adjust the flipping gap taking into account volatility.
Ninna
|
| |
|
Edit/Delete • Quote • Complain |
PocketChange
Registered: Jul 2008
Posts: 2036 |
02-25-11 01:28 PM
This has always been a problem with the algo... you don't know your in a flat zone until you've reversed. One variation was to hibernate the trade sequence at each stop and require the market price to move into a new price zone before resuming with the reversal.
The primary benefit of this style of an algo is the ability to front load the risk and not be time locked in a position or play.
Psychologically I prefer to scale size in an already profitable position. The mind set of gambling with 1/2 your profits is much more comfortable than gambling to cover your loss.
Quote from nLepwa:
Also you need to dynamically adjust the flipping gap taking into account volatility.
Ninna [/B]
|
| |
|
Edit/Delete • Quote • Complain |
| Receive
an email whenever a new post is added to this thread by subscribing
to it. |
|
|
|
|
