ROMAD, the ultimate statistic?

this is a reprint of a thread i posted a few years ago. i am reposting it because i think we can delve further into the discussion with so many bright traders at FF these days btw, i still hold absolutely firm in my view of ROMAD.
________________


youve all have heard me talk about my favorite trading system statistic, what i used to call "annual return vs. max drawdown".

i got around to reading "Trading Risk" by Grant. whaddya know, this is his favorite statistic also. he calls it ROMAD. i like this name better

it got me thinking about ROMAD, and trying to visualize why it is the most important statistic. its incredibly simple, yet still took me a moment to grasp.

the reason ROMAD it is the most important stat is because it mathematically defines risk vs. reward (actually, it defines reward vs. risk, but its the same difference). if you think about it, what is the risk of a system? its the max drawdown! and as most of us are trading for bottom line results, reward IS annual return. annual return divided by risk = ROMAD.

heres an example...

a rational (and i use that term loosely ) investor will have a maximum amount they are willing to risk. lets say you have a 10k account, and the maximum you want to risk on the total system is 3k. in other words, you are willing to bear a 30% drawdown before you quit using the system.

with this in mind, pretend you have two systems to choose from.

systemA / ROMAD = 1.8
systemB / ROMAD = 1.0

since we are only willing to risk 30k, no matter which system we choose, systemA will give us 80% higher annual returns than systemB. because for every one dollar risked (in the form of maxDD) you get back $1.80 back (in the form of annual return) with systemA. and for systemB you only get back $1 for every $1 risked. thats a complicated way of saying systemA has a better risk vs. reward!

ROMAD = avg. annual return / max drawdwon
ROMAD = REWARD VS. RISK

That is exactly how I look at it, although what I generally do is consider 2-3 month periods together in terms of both performance and risk, because that is how long it takes for me (usually) to move up the volume a notch. You can then average out the reward/risk over all the periods.

One other thing: maximal drawdown is misleading. What if it is a statistical anomaly, and your usual drawdown is much less? There are ways to smooth these things, so that you can be AWARE of the maximal drawdown but also aware of how likely it is in probability terms given the statistics. I usually have a fixed figure in mind about how much I am prepared to lose per trade. This number is dependent on market volatility and price action, but also on what I ate that day, how sunny it is, whether I've met any ugly women in the elevator, or maybe pretty girl the day before..etc.

What I'm trying to say is that you may have different appetites for risk at different periods, if you like to place your calls yourself. Discretionary trading is an art, and measuring its success should bear respect to that fact.

ROMAD. Made me instantly think: Return On MADness.
Fits the bill!

I do agree on it being a key characteristic.
I try to generate these stats every month or 2 as well to check how I'm doing and get an eventually presentable graph, but it's a hassle if you shift money into and out of the account, as that makes the Excel work with the exported stats terrible, at least with Oanda. (I.e. you have to manually compensate / differentiate between wires and actual trading..*sighs*)

I'd be against obscuring this by "smoothing". Regardless of how large an "exception", money gone is money gone. Magicking Math won't change that. What are we, Bear Stearns!?!?
Keep it clean I say!

true, ROMAD its flexible in that you can adjust the time period for the max dd and return. i use one year because annual returns are the norm in the investment world so it makes for easy comparison.

you make a great point here. max dd is always a guess at best...obviously you can look to history for an idea on what to expect, but that just lets you make an educated guess over a wild guess

i think i forgot to mention ROMAD stands for "return over max drawdown"

i also forgot to mention that the ROMAD stat is used mainly when developing quantitative systems that can be backtested. when im discretionary trading i throw statistics out the window because things like patience and discipline become far more important. the quantitative strategies are the ones i calculate to the Nth degree.

nice idea
what about number of trades?
if system C have ROMAD=3 with 1000 trades
and system D have ROMAD=3 with 300 trades

which system is better?