Position Size vs Time Frame

Ok, so lets take an example of a successful system. I will use the James16 thread to illustrate (as it is the most widely followed). In this thread i noticed many users claiming success on the 5 min time frames (rightfully so) and many vets advising noobs to begin on the daily's first for various reasons (again, rightfully so). Now this left me thinking...

I use 1% on the daily's and weekly's, and 0.2% on the hourlies (the actual percent size does not matter, its the multiple that counts).
Now my usual stop on the hourly might be 35-40 pips and 150-200 pips on the daily. So my lot sizes might be similar, but percent risk is percent risk and same for gain, so unless i'm banking quick profits after BE (to the pt where its detrimental), my risk should be proportioned properly. Why?

a) Well because, I will have more time to properly line up my setup on a daily (PB, PPZ, fibs, TL, confluence), and the daily will give less false signals (my assumption) due to more info in one bar etc.

b) On the hourly's (the lowest i would go, probably taking an entry on the 15 min), I would have to rely on quicker analysis and hence be prone to certain mistakes, and instinct would play a part etc.

Why then trade the hourlies?
Trade frequency WILL define the returns of traders using similar systems.

Even though taking 2 or 3 daily setups will provide an adequate return on a large (this is so relative though) account, fact is, most traders would like to increase their frequency and thus, return on their 'small' accts right? I guess what it comes down to is this ...

If you have an edge on the hourlies and you can 'define' it? Do the inefficiencies that creep up in b) above, actually wipe out your edge? If not, then it does not make sense to pass up these trades.

But this is subjective, some people will argue that the market being fractal in nature, provide the same chances on every TF, some might argue to take 'only' the best setups on the hourlies and even 5 min. Some might say, that it's foolish to vary position sizes using a fixed fractional system unless dependency has been proven (which as we know, is awful Hard to prove with any reasonable level of 'confidence' {RUNS etc}), 'you might win the 0.2's and lose the 1's' etc etc.

What are your thoughts on varying position sizes according to time frames?

My final thought for now is, that since we do look for optimal allocations in an optimal portfolio for our systems (assuming we use more then one system, even if that's an interest bearing bank act), and in such a scenario there exists an optimal position size to 'put on'. Can we not break this down to apply to various time frames within each system? If the data below the dailies is reliable as a consequence of more 'noise', then we can dampen the effects by reducing variance (reducing position size) on lower time frames. Is this a sound approach?

Determining your PS based on the TF used to make trading decisions makes no sense. It also makes no sense to use the frequency of your trades for that. If your system has no real, definable edge, you will eventually end up with 0 in your account, regardless of your position size, the only difference being how long it will take. However if it does have an edge, then PS size becomes crucial to the systems success once you have a period of DD. Therefore the main thing determining your PS should be how much DD you are willing to stomach, and how long it would take for your equity curve to make a new peak.
If your system has a relatively low DD on the 5 min, and a higher DD on the Daily, then you would use a larger PS (% wise) on the 5 min. So it is not about the TF
Just my 2 cents worth (less)

I agree - timeframe as such should not determine how big you will bet.

Determine position size by risk including the statistics of a typical trade on the timeframe. So you might risk 2% as standard for a long tend following system but adjust it for:
- win ratio (higher win ratios with the same expectancy have better kelly criterion so the amount risked may be higher for the same risk of ruin)
- amount of slippage expected in normal and shock situations.

Obviously the % slippage in a shock situation will be higher for a 5m trade than a daily trade (assuming the initial stops relate in some way to bar sizes). So to that extent a low timeframe will need a smaller position size vs the standard amount risked ( range to stop + commission + slippage ). If the shock situation might result in slippage that results in a loss on a trade = 2x standard amount risked on 5m and only 1.1x standard amount risked on daily you need to drop the % risked down towards 1.1/2.0 of the daily risk percentage.

I am referring to the same system, so I am implying a similar win/loss ratio in all timeframes. Plus position size and risk is essentially the same concept to me. 1% and 0.2% are different risk levels and hence diff position sizes.

My premise is, given the same system, there are other intangibles aside from technical factors that affect results on lower TF's.

Now, as far as slippage and shock, i agree, that would definitely make a difference, but then again, you are fighting a bigger spread edge on the lower TF, which goes back to you being +EV in the first place. Which is assumed for arguments sake.

Since we 'could'(?) anticipate higher variance in returns due to the factors i mentioned and slippage and shock (you), can we not reduce this variance by reducing position size (which is the same as resuming risk)?

timeframes are an illusion. There's just time. I scalp into all trades at 0.2% of account per pip, and if they run, I get richer.

0.2% per pip? or was that a typo. If you meant 0.2% per trade, then, that was my original plan, and I will probably defer to that for my first hundred trades anyway.


As far as time goes, I think there should be some comparison of holding period returns for both TF's. A 400 pip move on the daily might net me the same dollar amount profit as a 4 pip move on the hourly using the standard approach, but one takes 5 hours and the other 6 days. If we were to balance the HPR's for both TF's then does not variable risk come into play?

I might be going about this the wrong way, just thinking aloud, but somehow betting the same percent on hourlies as on the weekly's does not strike the right chord with me.

the hourlies and weeklies do not exist, there is just the rate changing over time. If I scalp into a trade on the 10s chart at (yes) 0.2% account risk per pip there is no law that prevents me from holding this trade for a week or whatever, other than the self-imposed one of % allowable drawdown. If I get out in a minute with a 5-pip, 1% gain, that's good, too...

Not according to my watch they're not...
Mathematically, the hourlies are very different to the dailies. The hourlies range more than they trend, the dailies trend more than they range. This tells you one important thing - you cannot trade the same system and expect the same results (notwithstanding spread impact) on different TFs.
But, there's a danger of missing the bleedin obvious point here that there are key points in the day - London & US opens, news releases, the NY cut etc that have a big impact on the market, that are largely invisible if you trade up to higher TFs.

True and I agree with your point, especially the last point you made. In trading there are not such thing as timeframes. This is proven on the fact that the market could be traded using a Quote Board, the truest representation of price action only market. Charts are nothing more than visual representation of the market, that has the ability to be misrepresented. Quote Board/pure price action is the only order flow and volume indicator we fx traders have ... something that is often overlooked when discussing price action.

The idea everyone's trying to make is: If I choose to buy on the 5min or daily chart shouldn't matter so long you understand the market will be long overall. If chose to BUY because I believe the market is heading up. If I take that, it shouldn't matter what timeframe I came to that conclusion. If I made that decision on the 5min then at the same point I could have used any other timeframe to have made that decision. If you're right about your decision on the 5min chart then all higher timeframes should agree with you decision. If you're wrong about the direction then all other timeframes will disagree with that decision. It's the same idea as using multiple time frames trading, but in reverse.

Traders choose higher timeframes because it's easier to interpret the trend but that's not to say I couldn't do the same thing on the 5min chart. The idea is to choose the right direction of the market and hold that position long ... the only way to become profitable.

OK sure... but I'm not talking about your reasons for entry, be they a 5 min chart or what have you. If timeframes were irrelevant, the probabilities of a random entry, ignoring the spread, on a 1:3 trade would be equal whether you're risking 1 pip to get 3pips, or 100 to get 300. And if the market was perfectly fractal, they would be identical. But it isn't, and they're not the same - and the only difference between them is the length of time it takes to achieve the goal.

I've been interested in defining, and objectively identifying trends (even in hindsight).

Can you somehow back up this statement? I saw you make a similar statement once before, then referring to a trend being a candlestick with the body >50% of the range. There are many arguments against defining a trending market that way.

Can I back it up... what, here in public?
In short though, yes I do use that measure (efficiency ratio as Rabid described it) as to the degree of trending. It does have limitations, eg you can be moving in direction of trend and yet volatility > open-close prices, and of course dailies are made up of hourlies, but it has its place.
Even if you don't agree that's a good measure of trend, the point is the daily OHLC vs hourly OHLC data is very different. Which is very useful to know.

bigger frame high sl with high reward small time frame small sl with small reward

in this example if my strategy to short for 61.8 see difference between two time frame

still waiting for a useful reply,


remember



"reading comprehension" helps.

In that case, I'll have a go.

Time is a manmade concept. Timeframe should not be a variable in determining position sizing.

Instead, what should determine your position sizing is your appetite for risk and the edge of your system - if that is even quantifiable. Theoretically, if your edge is the same in all timeframes, position sizing should be the same as well.

The issues you describe when trading shorter timeframes (e.g. being prone to more mistakes,) have a negative impact on your edge. Additionally, it can be argued that shorter timeframes reflect relatively more 'chaos' than 'signal', and therefore the expected edge is reduced. This smaller edge calls for smaller position sizing.

All in all, in reality it may be advised to trade smaller lots on lower timeframes, but not for the sake of the lower timeframe, but because of the lower edge.

Edit: When I'm talking about position sizing, I'm not talking about absolute lot sizes, but about fractions of account size.

That is complete madness. It IS a variable in sizing because you can't ignore variance. Read up on "gambler's ruin" and the Kelly Criterion.

The variance in the faster timeframes will be insanely high while the slower timeframes it will be much more manageable. If you have an edge at both (they can even be equal) ignoring variance will cause one trader to go broke at M5 while a trader at D1 can keep going on and on and on.

Doesn't your edge provide an indication of your ability to 'beat' variance for a given timeframe? Thereby making variance a useless concept in this discussion.

From my observations and calculations I have concluded the longer the time frame, the larger the "Average True Range" and hence the larger the stop hence the larger the risk. Conversely, the smaller time frames allow for smaller accounts.

Example: On a 5 minute chart the my own ATR based stop tends to produce a 10-15 pip stop on something like the GBP/USD. So a $10-$15 risk on each and every trade correlates to a 10/.01 or $1000 account.

To me, risking 1% on each trade tends to keep me cool calm and collected and it'd take quite a few losers in a row to endanger the account.

This is all based on the standard $10,000 contract.

I figure that I'd need at least $2000 to trade two contracts with each trade risking 1%.

Working on 15 minute charts involves 30-45 pip stops so I'm into a $3000 account for 1 contract.

This all stands to reason when one see the whales like Soros and Rogers who basically operate off month charts.

It is true that Bid-Ask spread hangs like a millstone around the necks of micro traders but that's the price you pay to learn. The emotions/mechanics are all the same and one needs to start somewhere.

Ignorant, arrogant and patronising. That's got to be a hattrick.
If timeframe affects profitability (it does), adjust position size accordingly. If it doesn't, don't. It's that simple.

thanks for that answer, sorry to rub you the wrong way, did not mean that