Fixed Fraction vs. other creative money management methods

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
Post Reply
leanhog
Full Member
Full Member
Posts: 11
Joined: Wed Apr 16, 2003 9:05 am
Location: B.Hills , California

Fixed Fraction vs. other creative money management methods

Post by leanhog » Wed Apr 16, 2003 9:23 am

I have been running simulations with Trading Recipes and am finding that
reverting back to a simple fixed fraction bet size selection and possibly using closed equity as opposed to Liquidating value has been as effective as some very advanced methods of bet size selection.
Wanted to hear opinions of others with their research into this area.
Thanks.
Jerry :?

Forum Mgmnt
Roundtable Knight
Roundtable Knight
Posts: 1842
Joined: Tue Apr 15, 2003 11:02 am
Contact:

Bet Size Overkill

Post by Forum Mgmnt » Wed Apr 16, 2003 11:07 am

I found that there are certain aspects of money management that are crucial and others don't matter so much.

It is difficult to answer the question in isolation. Often the different approaches represent different risk amounts.

For example, using core equity equates in some sense to using a larger bet size with closed equity; so direct comparisions of only the two bet size approaches won't yield a complete picture.

For certain systems, using closed equity is significantly better than open or core equity.

What is really important:

Betting small enough
Using bets of roughly the same amount

Ted Annemann
Roundtable Knight
Roundtable Knight
Posts: 118
Joined: Tue Apr 15, 2003 7:44 pm
Location: Arizona

Fixed fractional vs. other creative money management methods

Post by Ted Annemann » Wed Apr 16, 2003 12:06 pm

Several of my preferred systems seem to perform better and better (higher and higer reward/risk ratio) as I increase heat, using a constant-percent-risk betsize protocol. Unfortunately the drawdowns get bigger than I can psychologically stand.

So I've begun to try increasing my AVERAGE heat without increasing my MAXIMUM heat. My idea is, max drawdown is a strong function of (maximum heat) and a weak function of (average heat), whereas annual rate of return is just the opposite: a weak function of (maximum heat) and a strong function of (maximum heat). According to this idea, leaving max heat alone will leave max drawdown alone, while increasing average heat will increase annual rate of return. Voila: improved reward/risk ratio!

In the few tests I've tried so far, the approach seems to work quite well. Even if it's based on a nonsensical theory, which it may be, the result looks pretty darn good 2 me.

What I've tested so far, is a couple of very well known ideas. They're in the Turtle manual. They're in the Aberration Plus software & manual. They're built-in functions of Trading Recipes testing software.

The big idea is: increase the per-position bet, but limit the TOTAL NUMBER of bets. This means some entry signals will be ignored; some trades will not be taken.

I've tested the idea both on a portfolio-wide basis ("Total Portfolio Heat", Recipes keyword TOTALRISK), and on a per-commodity-group basis ("Sector Heat", Recipes keyword GROUPRISK), and both at once. Computer says: good.

Not a new idea. Not fancy, not elegant, not brilliant, not unorthodox. Nevertheless it helps. Quite a bit.

Forum Mgmnt
Roundtable Knight
Roundtable Knight
Posts: 1842
Joined: Tue Apr 15, 2003 11:02 am
Contact:

Post by Forum Mgmnt » Wed Apr 16, 2003 12:59 pm

Ted,

You've hit upon something that relates to what I consider the #1 problem with reports of the performance of the Turtle System: Most people have tested and reported performance for the system without the correlation or directional unit limits.

Many of the signals we had were ignored because we were already at the limits. These limits kept us out of a lot of bad trades. They also made sure we had our money in the strongest (weakest for shorts) markets within a correlated group.

Most people have just tested the system entries and exits across a basket of futures. This does not work anywhere near as well as when you use the correlated instrument limits which are akin to your "Sector Heat", "Maximum Heat", and "Total Portfolio Heat"

The correlation limits are better than "Total Portfolio Heat" alone, since they limit the risk that will be incurred by one single market event more effectively. "Sector Heat" is similar but not exactly the same. It probably accomplishes much the same thing and might have just as good performance with the right settings.

Tuning these will definitely pay off.

Howard Brazzil
Roundtable Fellow
Roundtable Fellow
Posts: 54
Joined: Wed Apr 16, 2003 12:45 pm
Location: Houston, TX USA

Post by Howard Brazzil » Wed Apr 16, 2003 1:28 pm

Several of my preferred systems seem to perform better and better (higher and higer reward/risk ratio) as I increase heat, using a constant-percent-risk betsize protocol. Unfortunately the drawdowns get bigger than I can psychologically stand.
That's an interesting observation, and one of the few examples of the elusive "free lunch." It indeed seems to be the case that when strong systems are traded more aggressively, max drawdown increases at a slightly slower rate than does reward. You also pointed out the catch: you have to be able to psychologically withstand the higher absolute drawdown - regardless - in order to benefit from the free part of the lunch.

I think this phenomenon is related to the issue of contract resolution. As the risk per trade is increased, you come closer and closer to trading the actual number of contracts specified by your position sizing routine, and given a system with a strong positive expectation, traded more aggressively, this process occurs at a faster and faster rate.
...So I've begun to try increasing my AVERAGE heat without increasing my MAXIMUM heat....Voila: improved reward/risk ratio!
Ted, have you found that this technique improves risk/reward ratios other than MAR, such the Sharpe Ratio, which is more sensitive to overall volatility?

Kiwi
Roundtable Knight
Roundtable Knight
Posts: 513
Joined: Wed Apr 16, 2003 1:18 am
Location: Nowhere near

Post by Kiwi » Tue Aug 12, 2003 9:27 pm

Howard,

I have also found that for some systems increasing heat/percentage bet size increases return faster than drawdown and thus MAR. I think your reason may be one component of it. The ones where this test proves out are short term with hundreds of trades per year and I believe that the reason is simple:

AvgMaxDDPercent = (New Heat/Old Heat) * Old Drawdown Percentage.

which is linear. By contrast the Return is cumulative so its something like:

Old Return = (1+ (AvgTrdReturn))^TradeNumber
New Return = (1+ (AvgTrdReturn*NH/OH))^TradeNumber

so you get the power of compounding working for you :-)

John
Last edited by Kiwi on Wed Aug 13, 2003 4:05 pm, edited 1 time in total.

Mark_et_Lizard
Roundtable Fellow
Roundtable Fellow
Posts: 62
Joined: Sun Apr 20, 2003 9:16 am

Post by Mark_et_Lizard » Wed Aug 13, 2003 8:58 am

I've noticed the same thing, no matter how much I turn up the heat with a profitable short term system MAr always goes up. Thanks for the explanation.

Post Reply