Statistics to Trade By

My fellow traders,

I currently use an excel spreadsheet to document and track my trading, and am trying to get a feel about what kind of performance I am having. Currently I track my wins and losses, and figure simple things like win/loss ratio, expectancy, etc.

I am wondering what kind of statistical numbers some of YOU guys use to keep track of your trading. I am not looking for the almighty most powerful numbers necessarily, but even just the little ones you keep an eye on just because.

Any and every measurement of trading interests me.

Many thanks.

Profit factor. Bottom line performance.

And at least 20% losing trades. (so you can have some confidence in the frequency/distribution of losers).

Profit factor, Sharpe Ratio, K-ratio, efficiency factor and a bunch of other less important ones.

Trade frequency, maybe more important then profit factor, or as.

If I'm + or - my deposit is a good one.

As an unrelated note: Tdion started a money management thread a while ago that you might find interesting. The wrapped up description is this... because, by its nature, risking a fixed percentage on each trade means that it will take one longer to recover from drawdowns (You will be trading at your largest when you lose, and at your smallest when trying to make a new high), an alternative is to risk a fixed percentage ONLY until a peak in account balance. When trading below the max account balance the lot size being traded should be kept at that which was used at the peak. One resumes risking a fixed percentage once the peak is broken.

Relevant? No.
Interesting? I hope so.

Obviously you would certainly still need pre-defined controls in place, just in case you weren't willing to lose all your capital.

Thanks for the replies guys, this has helped me a lot and I am looking into some of those numbers.

@Leon, I am also checking this out, looking into tdions thread. If nothing else I think it would be intersesting to track account behaviour at and around account highs/lows.

Thanks!

Tdion is a good friend, and I respect his views, but I must disagree with him here. Position size should reflect the quality of the setup; one's account balance is irrelevant.

MM can not provide an edge. All one is doing in this instance is exchanging one kind of risk for another. If a trader varies his position sizes, it is no better than hit or miss as to whether the larger positions happen to coincide with the winning trades, or the losses.

The nature of fixed fractional betting is such that when you break a winning streak you lose the most and when you break a losing streak you win the least. But no amount of tinkering in a staking system(as Hanover already mentioned) can affect the expectancy of a system.

This brings me to another point (another stat ppl look at a lot) drawdown or max drawdown as a whole which I have seen mentioned in various threads recently, is basically an idea that is meaningless to the independent trader (which I am assuming most are on this board), it has no statistical significance, we can never know in advance what our drawdown will or can be. If you toss a coin 50 times you can expect 25 heads but you might get them after you get 25 tails in a row or sandwiched between 10 and 15 tails or whatever (keep in mind we can 'expect' 25 heads, wont get it evey time)(this is further a function of sample size).
Drawdown is a function of your biggest loss and that is your most important stat. You use that to figure out your fixed fractional stake. And if you trade with reinvestment and only interested in a max terminal wealth (after a certain number of trades, hopefully big enough N size of trades to realize your positive expectancy), then drawdown should not even be considered.
Fact is the fastest way your acct can grow is through optimal f, but if u keep betting optimal f, you will eventually lose, and when you do your loss will be f%. So the paradox is, the better the system (hit rate, win size), the higher the optimal f, hence the higher your loss and commensurate drawdown.
Drawdown is given a lot of attention in analyzing CTA's and hedge funds, but they have reasons to adhere to these guidelines (and basically fool themselves and others into expecting that in the future), we dont need to. The fastest way we can grow is by experiencing the most drawdown. LEt the expectancy of your system sort itself out with a high enough N. Fact is the order in which your losses and wins come wether you are reinvesting or not is immaterial, so if you look at a terminal wealth after say 7 years and you want to maximize it, the most efficient way of getting their is through high drawdowns.

But you ask, "I dont want so much risk?". What is risk? Classically defined as the variance in returns, which is obviously not what risk is. Risk is the chance of you going broke PERIOD. Whether that broke is a lower absorbing barrier of say 20% loss of initial equity or 50% or the entire stake, differs, but there is a point where you ill cry "Uncle", meaning you are 'effectively' ruined. That and only that is your risk. So you need to have your win rates and win sizes down , your frequency will give you a better idea of your sample size landscape over 7 years, and then consult the ROR tales (Balzara and Tushar Chande's books have a comprehensive list), your personal utility function curve here is important to help you psychologically, but basically look at your biggest loss and have an idea of what will your mental disposition will be when you loose 10 in a row (and you will), will you be disciplined enough to follow your systems expectancy after 10 losses? I know system expectancies can change with changing market conditions, but more often then not it is the trader who loses faith or his mind way before his robust system implodes. So in effect if you are betting for the next 10 years and say 5000 trades, on a +EV system, just get the trades in the bag ("good trades" not winning trades), the rest will sort itself out.

But that gets us too multiple systems, which is a must and here is why:- Lets say you use one system, and it has gone through a 3 month drawdown (like 3-29 or something), logically when this system turns around, it will have less money to bet with then before the losing streak started. But if you had another system (not positively correlated) in the mix, it could pick up the slack when system A slacks off and provide system A with adequate capital for another winning run when it comes back to the forefront. This is what multiple systems are for, and they can do a good job of smoothing out the equity curve for you within manageable limits if you give your sample size the time it needs. Matter of fact the more +ev systems the better, just keep the correlations in mind, and also there are times when adding a negative expectancy system to the mix can actually improve performance. But there is a limit to the number of systems you can use efficiently here, there is a point where adding systems is counterproductive.

I guess my point to the whole thing is, if you can get yourself into a situation where you can reinvest fully or partially for a fixed amount of time (to get a large enough sample size - this is a function of your system and its trade frequency), then you are in a much better situation to extract max wealth from your endeavor. Monthly summaries and quarterly reports are good for hedge funds, but if you control your own destiny and capital then you have the ways to get their the fastest. Obv making 20% on a 100m fund is better then 100% on a 10k acct but thats another story.

Sorry for getting off track and the length of this post was just too long to explain a couple of points (my writing sucks, ill get better).

Most of this is from Vince (after he revisited optimal f with a 'practical framework') and Balzara.

Leon,

I actually quite like this idea(or variations on it). I suspect it will smooth the equitity curve. I 'll be doing some tests on this in the near future.

Interesting: Yes

www.fxstat.com
www.myfxbook.com

Very thought-provoking post. Thanks for sharing.

Agreed. Trade frequency, and also position size, both have a huge bearing on returns.

Does position sizing reflect the quality of the set-up when you reduce your $ risk after a loss by risking the same percentage on a lower account balance? Because it will take longer (more trades) to make back what you have lost... it is as if one is varying their position size because of the loss.

"Position size should reflect the quality of the setup; one's account balance is irrelevant." ... Is this not th rule you are breaking by reducing the $ risk (using a fixed percentage) when you lose?

When risking a fixed percentage it takes more trades to reach new account highs, this could be seen as a kind of negative edge/detriment to the trader. We don't need the math of 'percentage position sizing' working against us as well as everything else.

Should I risk less dollars on a trade because I have been losing? Or do my trades have an equal chance of success as before?

Do you see what I'm getting at?

EDIT: PS. What you are doing is seeing the percentage as the benchmark for 'equal risk', although in my opinion, equal risk should be seen in $ amounts, IE. 1 winning trade should completely compensate for 1 losing trade (given a method with an average 1:1 RR)

EDIT 2: I'm a fan of percentages for measuring gains, and increasing risk proportionately to the account balance... but I can't help but see that by reducing dollar risk after losses we are further diminishing the benefit of any edge we may have.

When all is said and done, the estate agent advertising the villa displays the price in currency.

Do you grade your setups then? Why would you take anything less than a top quality set up?

Unless risk is kept at a minimum percentage of equity on every entry and then trade size is increased as the trade becomes profitable. That way you have larger positions when your trade is 'right' and smaller positions when it isn't. Risk isn't increased but gains are.

I dunno, it sounds logical to me.

Maybe a sensible 'in-between' approach to take would be to re-implement fixed percentage position sizing when a certain account balance is crossed below, and then to increase risk again to that which was used at the peak when above. An example:

 

1. An account starts at £500 and reaches £600 risking 2% per trade.

2. At the peak £12 was being risked per trade.

3. One sets a predefined point to reduce trade size to 2% again, such as at the starting balance of £500 where only £10 was risked.

4. When current balance is above this point £12 is risked, when below £10 is risked.

This approach seems more sensible to me than a fixed percentage approach because a string of WLWLWL etc. would keep taxing the trader even though they are winning as much as losing. For example:

Wins 2% of £1000 = £1020

Loses 2% of £1020 = £999.60
Wins 2% of £999.60 = £1019.59
Loses 2% of £1019.59 = £999.20
Wins 2% of £999.20 = £1019.18
Loses 2% of £1019.18 = £998.80

Agreed, the loses are small, but it is still undesirable even in a distribution of even wins and loses. This problem would still arise at the 'predefined risk change point', but at least it would be less frequent. In the example above the trader would have lost £1.20 over 6 trades, but scale that up to a £100,000 account and loses are £120.00. Loses could therefore be very hefty given this scenario and position sizing coupled with a high frequency trading strategy.

EDIT: At least allowing a run of up to 3 consecutive loses at the peak $ risk would conserve a considerable amount of capital over the long-run.

I'm just hanging out here putting off doing one of the most boring law assignments ever created.

Profit factor, Calmar ratio, Sharp Ratio, Correlations, Drawdown (abs & MAX), Standard Deviations, Linear Regression, and Normal Distribution

If you use MT4, FXCM or DBFX platform, all the above can be found on www.fxstat.com

Nothing can be as boring as trading, I mean how morose is it to watch a number going up and down, it's terminal boredom!

ssdd

Depends on how you trade.

Law assignments are basically researching and re-writing what you find in your own words... that is hell for me. Trading involves research, forming opinions based on facts, putting your money down based on your opinion, and then waiting out to see if you were correct all the while reevaluating your trade based on new information.

I know which one I'd prefer to do indefinitely.