Money Management 101

Don't thank me yet, as my suggestion of not doing it may seem unhelpful. My argument against it would be the same I would use against martingale. Why do you want to do this?

I thought that this old thread was worth resurrecting. The OP's posts on different MM methods are a worthwhile read IMO.
_______________________________

Basic risk calculations

Essentially it all comes down to A/B, where:
 'A' is the amount you want to place at risk (fixed $ amount, or %equity)
 'B' depends the method you want to use (per trade, per pip, per ADR$, etc)

Hence you combine your preferred choice from 'A' with your preferred choice from 'B'.

With regard to 'A', using %equity allows compounding. Hence (for better or worse) the absolute dollar amount at risk grows as your account increases, and shrinks as it decreases. Whereas fixed $ might be more suitable when accounts have grown large, and large equity swings (in absolute dollar terms) become unnecessarily risky.

With regard to 'B': most traders prefer to equalize the amount of risk per trade, i.e. they risk $X, or X%, per trade. I used to adopt this approach, and I still don't see anything wrong with it; and it inherently gets around any problems normalizing sizes across different pairs/instruments. The calculation is arguably simpler, if you're not using EAs or scripts to place your trades. However, it necessitates the use of stoplosses, or at least some dummy loss-exit point, to allow position size to be calculated.

However, since returning to trading this year, I've gravitated toward a 'per pip' methodology. Reasons:
 My focus is on making pips (and hence $), rather than making winning trades.
 Everything else being equal, I see myself being stopped out of a (for example) 20 pip SL approx 5x more often than if I use a 100 pip SL. Hence I'd feel more comfortable risking a lesser $ or % on the 20 pip trade than the 100 pip trade. (Note that this is purely personal preference, and it has no bearing on the RR where profit exits are concerned).
 Moreover, it eliminates the need to use hard stoplosses. (However, I still exit losses quickly, and then look to re-enter, when price moves further against the original position, to the next entry level -- thus avoiding the pips that I would have lost in the interim).
 If I'm entering an exiting the same macro-position many times, pip-based sizing is more workable than a per-trade method.

For example, if I risk 1% of my account per 100 pips (on a pair that moves at $10/pip per full lot traded), then I'm trading 1 lot per $100k in my account, which is effectively 1:1 leverage. Risking 2% equates to 2:1 leverage, and so on. However, different pairs/instruments have different $/pip values -- which highlights the need to normalize sizes across different pairs/instruments -- and different average daily ranges. To get around this I use a 'per ADR$' rather than 'per pip', where ADR$ is the average daily range (in pips) of the instrument, multiplied by the $/pip (per full lot traded) for that instrument, i.e. it is the average number of 'dollars' that the instrument moves in a day. (Hence, in reality, I'm attempting to size based on absolute dollars rather than pips).

The calculations, with regard to 'B', for each approach are as follows:

a) The conventional 'per trade' approach:
number_of_lots = dollar_amount_at_risk / (pips_between_entry_and_SL x dollars_per_pip_per_full_lot)

b) The 'per pip' approach:
number_of_lots = dollar_amount_at_risk / (100 x dollars_per_pip_per_full_lot)
(or you can simply use a rough fixed value of 10, for dollars_per_pip_per_full_lot)

c) The 'per ADR$' approach:
number_of_lots = dollar_amount_at_risk / (average_daily_range_in_pips x dollars_per_pip_per_full_lot)

I've highlighted the part that differs in red typeface.

In each case, and with regard to 'A':
 If you're trading fixed $ amount, dollar_amount_at_risk is simply the amount in $ that you wish to risk.
 If you're trading %equity ("fixed fraction"), then dollar_amount_at_risk = desired_percent_at_risk / 100 x account_equity_in_dollars


Again, I'd point out that (if RR is important to you) none of this has any bearing on RR. It is merely setting the lot size, given the initial risk that the trader wants to take, i.e. the risk part of the RR (return divided by risk) equation.

Note: By "dollars", I'm referring to the currency that your account is denominated in, be it US dollars, Australian dollars, euros, pounds, yen, whatever.

Great thread here and great last "summary" post, Hanover. Thx for this.

Am I correct when I say that this means that your 1R risk (%equity or X$) is just the ADR? So if I would like to standardize across pairs but use a set SL I could adjust to a multiple of adr in the calculation and set SL as this multiple which equates my 1R = dollar amount at risk / % at risk?

To me, it feels like MM is a hot topic but is somewhat of lesser importance. I guess it is essential when you are trading a system with a positive expectancy that you maximize your equity curve while keeping your drawdown within a determined limit. A %risk will go a long way with the value determined by a Monte Carlo / Bootstrap.

The task is basically to arive at a system with a positive expectancy and only then can we think about a form of MM. Trade management (entries and exits that maximize the system's expectancy) seems much more important which brings me to this post:

http://www.forexfactory.com/showthre...77#post1452677

Thx for the great stuff,

Best,

Assuming that I understand you correctly: that's exactly right. The amounts in red typeface in a) b) c) in my previous post simply represent the number of pips you want your 1R risk to equate to. a) b) and c) are merely three possibilities; in reality, it could be any number of pips that you wish. If you set your SL to the ADR, then pips_between_entry_and_SL is equal to average_daily_range_in_pips, hence calculations a) and c) would give an identical result. From what I've read, setting SL to a multiple of ADR is fairly common practice with many traders.

Note: I'm careful when using the term "MM" on forums. The terminology can become confusing, as a lot of folk equate "money management" with exits (stoploss/trade management) as well as position sizing. I prefer to make a clear distinction between exits and MM (position sizing), for these reasons here:
For me, trading is entries and exits; that's what separates it from other activities that likewise generate distributions of wins and losses, such as casino games. Like Dr Jones, I see MM as a function that can be treated independently, and separately, and in this way it becomes an exercise in pure math. Hence different MMs can therefore be tested by the application of math, leaving no room for guesswork or conjecture, which I find helpful. You can take a given trading system (entries and exits), and then apply any desired MM to it, without interfering with the system rules.

Mathematically, that's exactly right. However, IMHO the primary goals of MM should be (1) preservation of capital, and (2) an even distribution/diversification of risk (i.e. avoiding the possibility of one loss wiping out the gains made by several smaller wins). There's a good mathematical argument for taking more risk on higher probability setups; however, it pre-supposes that a trader has advanced far enough to know which of setups deliver statistically superior results to the others.

However exhaustive the Monte Carlo testing, markets can be fickle, and drawdown going forward frequently exceeds drawdown in testing. The Kelly formula delivers optimal return for any given set of parameters, but is almost universally deemed to be much too aggressive. Also, I believe it's important to keep position sizes within emotional as well as mathematical tolerance levels: as the saying goes, scared money never wins. Ultimately, there's not necessarily any right or wrong answer: how much you risk is a personal decision.

About whether 1R is better represented by $X or %equity: most traders use %equity, and I certainly don't disagree. Assuming positive expectancy from one's entries/exits, accounts can be safely compounded using X%equity risk, where X is a sensible value. Having said that, when account size gets large, the old adage that "1% of $1 million is the same as 1% of $10,000 -- it's still 1%", while mathematically correct, is arguably unhelpful: there's a big difference in 'real' terms between $10,000 and $100, especially when it comes to paying household bills! Hence there may be an argument for reducing %equity risk as the account balance grows. Again, it ultimately comes down to personal comfort and preference.

If your thinking is similar to mine, you might find this post interesting (I hope that referencing one of my own posts doesn't sound like an exercise in self-promotion, it's not intended to be).