Money management - relationship between profit and position size

[It's Snake Pliskin]

Some fantastic stuff here fellas.

One question, have any of you tried over any decent length of time, exiting a position if the trade does not go in your favour within a couple of days? (was it Jesse Livermore who used to do this?).

I personally found my most successful period came when doing this, I initiatied only 20 trades over a 3 month period and made a profit of just over 30%. Wish I could compound that over a year or two

Since, I have been using a traditional initial stop based on chart patterns (below a double bottom, support etc) and do not exit until either the stop is hit or the position moves in my favour before I adjust that stop, however I find this far less profitable (maybe simply due to volatility and now being a time where very short, high frequency trades are going to make you the $$ IMO). However, over the past week I have moved back to my initial exit strategy.

Any thoughts?

Momentum and lack of momentum = apples and apples

Momentum and time = apples and oranges

 

[sleepy]

My point is that it is better to design intelligent ways to exit rather than allow the maximum loss stop to be hit regularly.

This approach is advocated by Curtis Arnold and Nick Radge and again highlights the importance of Managing the Trade.

For example, in Nicks Power SetUps very few of the stocks that are closed out hit his original stop loss (i.e, maximum loss).

One question, have any of you tried over any decent length of time, exiting a position if the trade does not go in your favour within a couple of days? (was it Jesse Livermore who used to do this?).

Again both Curtis and Nick recommend defending an open position (i.e, moving the original stop if possible) if it doesnt move in a positive fashion (i.e., higher closes) in the first 4 days.


sleepy

 

[MRC & Co]

Yeh, I see Nick does it, which is pleasing.

Wanted to know how many others apply the same logic as I have read many who say leave your initial stop in place to give it time to move, but this appears to be counting on more luck and you being right, as opposed to getting on momentum (or a change thereof) or a breakout/breakdown, which is what chart reading is all about afterall.

I definately know which one I have found more successful.

Just another example is last week I initiated 7 trades, 3 did not move my way and I got out with practically no loss on 2 and was stopped out on 1, 4 did move my way and are up by an average of about 5% each in just 4 days. Not much and definately far from done yet, but it is a far more positive position to be in than with dogs lingering around my initial stop offsetting those running away.

Cheers

 

[Timmy]

Some fantastic stuff here fellas.

One question, have any of you tried over any decent length of time, exiting a position if the trade does not go in your favour within a couple of days? (was it Jesse Livermore who used to do this?).....


Any thoughts?

MRC - I do have some thoughts on this approach, not original, but thoughts I have found useful so I will spout them here if OK.

Basically the philosophy is that, having entered into a position, the price must move to prove the position right in order for the position to be held. The idea is that the market must behave (price move in my favour) to show the position is 'right'.

This is a different approach to setting a stop and letting the price action either take it out or go into profit. In this approach the market (price) is allowed to move to show that the position is 'wrong'.

I am not sure if I have explained it well but the first approach comes from an assumption that I must see confirmation from the market to allow me to hold the position.

The second approach I am assuming the position is right and then allowing the market to prove the position is wrong. The difference between the two approaches is subtle but very important (at least to me). I hope I have made sense.

In summary - I think your approach of not giving the position time, but requiring that it perform and thereby confirm your position is the correct approach and one I use too.

 

[MichaelD]

Developing a trend based system on CBA and BHP, why do you need stops?

'Cause CBA was $62 6 months ago whilst in a long term uptrend with no end in sight and then got down as low as $37.

 

[nomore4s]

MRC - I do have some thoughts on this approach, not original, but thoughts I have found useful so I will spout them here if OK.

Basically the philosophy is that, having entered into a position, the price must move to prove the position right in order for the position to be held. The idea is that the market must behave (price move in my favour) to show the position is 'right'.

This is a different approach to setting a stop and letting the price action either take it out or go into profit. In this approach the market (price) is allowed to move to show that the position is 'wrong'.

I am not sure if I have explained it well but the first approach comes from an assumption that I must see confirmation from the market to allow me to hold the position.

The second approach I am assuming the position is right and then allowing the market to prove the position is wrong. The difference between the two approaches is subtle but very important (at least to me). I hope I have made sense.

In summary - I think your approach of not giving the position time, but requiring that it perform and thereby confirm your position is the correct approach and one I use too.

I understand what you're saying Timmy, I think lol:

To me it depends on the type of signal & timeframe you're using to trade with.

If you're using some form of breakout set up like Nicks power set ups, you want the position to move in your favour straight away, especially if using shorter time frames.

But if I'm using (for example) a Wyckoff set up over a longer time frame (different magnitude of move) I'm willing to let the stock prove me wrong. E.g - Buying off a No.3 spring or at a last point of support, as the stock may take a bit longer to move but when it does I want to capture a larger portion of that move - I don't want to be shaken or scared out too early :

 

[Timmy]

I understand what you're saying Timmy,

That's a relief ... now if you could just explain what I said to me ... LOL

 

[MRC & Co]

Yes, refer to my post above Timmy. I was basically stating what nomore4s said, if reading the charts and looking for a breakout, the position should move practically instantly, further, especially this is particularly relevant for shorter timeframes, of which I see as critical in this type of environment, as stated earlier also.

Seems logical to me and is what I have used with the most success, that's why it was good to see someone both successful and more experienced back up my sentiments (as stated in Adaptive Analysis which is what I guess you are referring too Timmy). Wondering how many other seasoned campaigners, Frank, Wayne etc use the same technique......

 

[Timmy]

Seems logical to me and is what I have used with the most success, that's why it was good to see someone both successful and more experienced back up my sentiments (as stated in Adaptive Analysis which is what I guess you are referring too Timmy).

Actually, I first read about the idea from the Phantom of the Pits (PDF doc attached).

 

[MRC & Co]

I have heard of the Phantom. Thanks for that PDF mate.

 

[Frank D]

Michael,

who the F sake is going to BUY CBA around $62....

don't take to comment out of context....it was about investing over 20years

run stops if you are trading leverage on CBA and take profits on leverage around channel highs each year...

Long term investing:- accumlate CBA around higher timeframe levels every year on pullbacks.

 

[WaySolid]

Howard, I will need to think about this a bit more. I understand the logic of the idea that the maximum stop is the worst way to exit a trade; though I have never personally found a better way to exit with my trades that are in loss making position.

The following is all concerning trades in loss (<0 R)

Take a system with profits set at 3R, I'm loathe to close these trades between 0 and -1R except MOC (day trading) as the upside is much larger than the downside at these points, I would be much more inclined to choke a trade when it's up 2R than down 0.5R, confirming the addage that you should chop your winners and let your losers run. If you have some value adding skill closing a loser <0R and have edge tested your exits then I'm thinking that's a stop and reverse trade, otherwise let them run.

Some interesting things my numbers tell me, that you don't tend to read about that much.

1) On short term systems (No positions held O/N) profit targets can work very well.
2) I'm yet to find a trailing stop idea that adds value.

 

[howardbandy]

Hi WaySolid --

For your positions that hold a fairly long time, look at the Chandelier Stop.

For a Long position --
Use either the highest high or highest close as the anchor point.
Decide on some measure to determine how far from the anchor point you want your trailing exit to be. 2 * ATR(14), for example.
On the entry bar, set the exit point.
On every subsequent bar, raise the exit point if it is higher than the previous value, otherwise leave it alone.
Exit intra-bar when the Low is lower than or equal to the exit price.

Here is the code for the AmiBroker development platform, taken from my book:


// ChandelierStop.afl
//
// The Chandelier Stop is a trailing stop that rises
// as the price rises, with the distance between the
// price and the stop determined by the volatility
// of the price.
// Volatility is measured by Average True Range.
// The chandelier is "hung" below either the highest
// high or the highest close.
//
SetTradeDelays(0,0,0,0);
BuyPrice = C;
SellPrice = C;


// Use a moving average crossover to generate the Buy signals.
MA1 = MA(C,5);
MA2 = MA(C,25);
Buy = Cross(MA1,MA2);

// ATRRange is the number of bars used in the
// calculation of the ATR.
ATRRange = 10; //Optimize("ATRRange",5,1,30,1);

// ATRMult is the number of ATRRanges below
// the highest value to place the stop.
ATRMult = 3.0; //Optimize("ATRMult",2,0.5,5,0.25);

// Trail the exit from the highest High
// -- it could be the highest Close.
TrailPrice = H;

// Compute this bar’s Chandelier Exit Price.
ThisBarsExitPrice = TrailPrice - ATRMult*ATR(ATRRange);

// The trade’s exit price is the highest
// of the bar’s exit prices.
TradeExitPrice = HighestSince(Buy,ThisBarsExitPrice,1);

// Sell when the price drops to the CurrentChandExit.
// If there is a Stop Order in place during the day,
// use L as the triggering price.
// If there is not an intraday stop,
// use C as the triggering price.
TriggerPrice = L;
Sell = TriggerPrice <= TradeExitPrice;

Buy = ExRem(Buy,Sell);
Sell = ExRem(Sell,Buy);


shape = Buy * shapeUpArrow + Sell * shapeDownArrow;
Plot( Close, "Price", colorBlack, styleCandle );
Plot(TradeExitPrice,"Chand",colorRed,styleLine);
PlotShapes( shape, IIf( Buy, colorGreen, colorRed ),
0, IIf( Buy, Low, High ) );
GraphXSpace = 5;
//Figure 7.12 Chandelier Stop

Thanks,
Howard

 

[WaySolid]

Howard thanks for sharing, I used an ATR exit when I was trading ASX stocks (av hold time 14 days for me), here I think I was like all classical trend followers (no time stops) always letting the market take me out of a trade and never trying to pick a top. It was very important to catch my fat tails when I wasn't turning over my capital that quickly, now I have discovered faster time frame trading a whole new universe has opened up for me.

 

[bassmann]

Hi all,

I’ve been researching & testing everyone’s recommendations the last few weeks as I think it’s only fair I do so before coming back and asking for further advice.
Just thought I would give an update of where I am at, remember my aim here is too reduce my systems risk whilst maintaining (or near to) its profit characteristics.


Reading
MichaelD - currently reading Van Tharp as you suggested and about 2/3rd of the way though. Finding the reading is at the right level for me and is confirming things I already do whilst highlighting things I am not in a structured manner. Much for me to cover off here in the coming months and I would recommend for others.
• The first area to look at is to measure my system as Tharp prescribes in the earlier chapters.
• The second being to look at the components of my system that are ‘Setups’ vs.’Entires’ since I have not thought of it this way before. I haven’t completed the section on Entries yet, but at this point my takeout is my own system has a good setup and lacks an Entry since my setup is my Entry. If I can refine this I can potentially Enter closer to my stops and reduce Entry Risk.

HowardBrandy - Ralph Vince is on the reading list as well.



Risk
We’ve had previous discussions around Risk per Trade, Portfolio Risk and touched on how to use margin with Risk. I’m working on a model to work out ‘What Ifs’ based on this idea.
One question I have is, do people use Portfolio Risk theory just for position sizing and let there exits manage the trade, or do the Risk rules hold true for open trades as well? If so, does it apply to losing, winning trades and/or both? My gut feeling is it should apply Risk to cover your initial risk and protect profit.

But this brings about the following issue.
What if there is a big move on one stock in the portfolio, say 50% in a day.. If you work on a principle of 2% risk this may increase substantially and would suggest an exit. This goes against the principle of cutting losses short and letting profits run.

Alternatively, do a different set of rules apply to manage Risk on open trades?
Interested to hear others thoughts on rules for managing risk once in the market.



Stops
The stop in my system is volatility stop based of my own making. It has smoothing built into both the Stop and the Trigger (i.e. 10 day smoothing of the Close) and does not allow intraday exits, so a Close below the Stop may will not cause and exit. It’s also based on High for X days. I am using this as the initial stop and profit stop. I’m sure many of you think may be thinking smoothing and the accompanied lag are ultimately increasing my risk (and they probably are), but no matter what I try I cannot get better results with more commonly available methods.

I did try adding a % based initial stop to act as an absolute floor for a worst case scenario that would trigger intraday. This did not provide any value to the system so I have dropped this tact for now.

I noticed that quite often my Stop tightens a few days into the trade (lag inevitable caused by smoothing), especially if the breakout was substantial. This means the high risk I have on entry is short term and may cause Fixed Fractional sizing to calculate incorrectly.


Position sizing
I have done further position size testing on Fixed Position sizing and Fixed Fractional sizing. In each case I have factored brokerage and excluded outliers (best 3/worst 3 trades) and I have simulated at different risk levels.

The results on my system show that Fixed Position sizing gives better results than Fixed Fractional sizing.
Additionally, to my earlier point, 5 fixed positions vs. 10 fixed positions give far superior results again. Month by month profits are evenly distributed and DD acceptable.
Both conclusions are based on the system CAGR, DD and PF.
Despite this my system theoretically carries much higher risk than what my reading and you guys recommend as managed.

The whole theory around Fixed Fractional makes sense to me as so I need to unravel why its not for my system. I have narrowed this down to 3 issues.

1) Time in the Market - My system has an on/off status which over the long-term is on about 66% of the time (this would be less in the last 12 months given the environment). Of this 66% the actual days in the market is much less given time between signals and so forth. It you think about how this may impact 5 vs.10 positions in a Fixed Position size model it takes longer to generate 10 signals so I commit my capital to the market faster and increase inventory turns with fewer positions. This rationale seems to hold true if you are not in the market all the time. The net result is the effects on my CAGR.

2) Fixed Fractional - Fixed Fractional is not working because to achieve my desired risk level, its creating too many positions and suffering from the ‘time in market’ issue in the point above but probably in a more managed way than say a portfolio of 10 positions.

3) Stops – my method of calculating my stop may not be suitable for Fixed Fractional Sizing due to its lag effect from the smoothing. At the time position sizes are calculated the lag means short-term risk is large only to contract a few days later.


In my mind I am making significant progress here. The most important point being I now have a picture of what I am aiming for.
Any comments/feedback much appreciated.

Regards,

Bassmann

 

[peter2]

Your research will give you a good understanding of what to expect using your system and it seems that this is important to you. All the research that you do will not prepare you adequately for what can happen when you actually trade. Strange things will happen when you start trading.

Adequate preparation will allow you to trade your system correctly. Over analysis will make you doubt the effectiveness of your system and you won't trade it correctly. I think you are in danger of over analysing your system parameters. You can only minimise risk not eliminate it. In fact, you have to embrace risk to trade well.

do people use Portfolio Risk theory just for position sizing

IMO no, pos. sizing is determined by the individual trade risk (entry - SL). Portfolio heat determines the maximum number of open positions that are exposing your trading capital to risk. I do not include open profits in my calculation of P heat.

The results on my system show that Fixed Position sizing gives better results than Fixed Fractional sizing.

More risk gives you more rewards. If fixed pos. sizes gives you bigger profits then you are risking more using this model.

 

[nizar]

What if there is a big move on one stock in the portfolio, say 50% in a day.. If you work on a principle of 2% risk this may increase substantially and would suggest an exit. This goes against the principle of cutting losses short and letting profits run.

This doesnt make sense at all.
How does risk increase substantially if the stock skyrockets??

The 2% rule is the MAXIMUM loss of 1R that you are willing to tolerate on any trade (Of course due to slippage and/or opening gaps, you could (and will) potentially lose more than 1R). So its the worst case scenario.

If the stock significantly appreciates past your entry price (for example, up by 50% in one day), then your should have exited the trade OR your exit should have moved up. Actually there are possibilities as to what you can do here but this depends on your exit and your timeframe.

For me, I use a trailing exit, which follows the price upwards BUT then flatlines when the price declines, so I am protecting my equity as the trade moves in my favour.

Contrary to what I have quoted from you above, A trailing exit is actually exactly how you cut losses short and let winners run.

An example of how my trailing exit works is shown on my blog on the first page.

 

[PolarBear]

Hi,

Howard:-- trailing exits do not work well with short holding periods

Howard, could you explain why you think this is - I'm particularly interested in whether this is the case (and why) for strategies based on scalping indexes.

thanks
Daniel

 

[Trembling Hand]

Howard, could you explain why you think this is - I'm particularly interested in whether this is the case (and why) for strategies based on scalping indexes.

Daniel I suspect Howard is talking about a longer time frame than scalping with an indicator based system. Using the signal from the indicator to tell you to get out. (could you define your idea of scalping for us)

I have asked the same question because in testing have found systems based on price patterns that have been greatly improved buy stops but didn't get a response from Howard.

 

[tech/a]

I'm of T/H's thinking here.

Indicator trailing stops would logically be poor in my mind.

However Price/Range and or Volume based work well for myself.

But I havent tested this other than my own trading.
I suspect target exit are the best performing as there is no "give back"-----only in some cases lost "Potential" Profit.

Would be interested in Howards comments.