The Trading Tribe

Ed Seykota's FAQ
Home | Send Mail to FAQ | Resources | Ground Rules | FAQ Index


Contributors Say Ed Says




Jan 14, 2017

Integrity Jump-Start Kit

Hi Ed,

I'm wondering if you can support me in getting back some integrity. Or at least a jump start.
Thank you for raising this issue.

You might consider implementing the Integrity Jump-Start Kit for a week to see how it goes - and then reporting back to FAQ with your experience.

Integrity Jump-Start Kit:
1. Tell the truth.
2. Keep your agreements.
Jan 13, 2017

Elliptical Equilibrium

Dear Ed,

i need some insights from you about your Govopoly model that is in elliptical equilibrium :

- two phases of 41 days each
- the first it your when duckweed fill the pond when exponential is leader
- the second  when clear water transform the pond in a mountain lake when logarithmic is leader and then starts from the beginning again.

I need some insights about this thinking  in order to build a system model

Thank you for raising these issues.

I do not know what you mean by the term, Elliptical Equilibrium.

You might consider developing some familiarity with the basic vocabulary and concepts of System Dynamics.

For example, see:

Jan 12, 2017

Optimizing on Recent Kelly

Dear Ed,

I replicated both strategies of the Trading System Project to the penny.

I optimized the parameters of both systems for 25 commodities for the past 27 years (1990-2016).
I back tested the Support-Resistance System with a portfolio of the 25 commodities using the optimized parameters for each.

I used 2% risk per trade, 1,000,000 dollars as starting capital and a 0.5 skid fraction for executions in the portfolio backrest.
Trading costs, fees and taxes were ignored in the test.

The results of the backrest are:

ICAGR: 17.67%
Max DD: 42.03%
MAR: 0.4204

I optimized the portfolio of 25 commodities using the same look-back periods.


I took the backrest with the optimal settings (optimal: Slow=160, Fast=110, applied on all the commodities).
All other things unchanged, the results of the backrest are:

ICAGR: 10.57%
Max DD: 34.00%
MAR: 0.3109

I also tested the portfolio using the Kelly criterion (K% = W-(1-W)/R) for position sizing instead of fixed fraction.
The way I calculated K% was as follows:
a) Individual back test for each commodity, trading one contract.
b) I calculated ‘R’ from risk adjusted total profit/total loss. That means I divided every profit and loss with the initial dollar risk of that trade. This way I got total profit/total loss in terms (multiples) of risk.
c) I created a theoretical portfolio that trades all those products that have positive individual K% (positive expectancy). This way the theoretical portfolio was dynamic. As time went by it selected positive K% products and left the ones with negative K%. All trades used one contract as position size.
d) I calculated the portfolio K%, using all trades that were taken in the theoretical portfolio.

I used individual optimum parameter settings (like in the first case).
I used the calculated theoretical portfolio K% as the fraction for position sizing for the next trade in the ‘real' portfolio backrest.
As in the case of the theoretical portfolio, trades were taken only in those commodities that showed positive individual K% in individual one contract tests.
All other things unchanged, the portfolio gave the following results:

ICAGR: 27.01%
Max DD: 86.67%
MAR: 0.3117

*Note: the closed balance of the account suffered a drawdown greater than 100% (103.76%) on one day. Entry signal was not issued on that day. A next day closing of another position pulled out the balance from negative territory.

I tried to play with the risk parameters of this last test.
a) I weighted the portfolio K% for the next trade with the K%-s in the individual tests, so more risk was taken for commodities performing better in the past.
b) Tried to maximize the portfolio K% for the next trade.
c) Tried different fractions of portfolio K%.
All of the above reduced volatility, however reduced return as well, altogether ending up with lower MAR values.

Considering the results above, I was wondering:

1) What is the difference between choosing individual optimums over choosing a single optimum for the whole portfolio? The robustness of the latter? Individual optimums were chosen based on robustness as well and this way the system produces better results. Is it the trader's preference?
2) Do you see any logical shortcomings of the way I calculated K%?
3) Does the Kelly criterion have any added value over fixed fraction bet sizing when speaking of trading? Under ‘added value’ I mean in terms of MAR.

Ed, thank you for your time, as always.

Best Regards,
Thank you for sharing your research.

Dynamically optimizing portfolio selection on recent performance may add considerable complexity to the system and may also risk missing moves that sometimes arise from long-dormant instruments.

For example, your one-day 100% drawdown may indicate a glitch somewhere in your position sizing algorithm.

You might consider taking your feelings about <searching for a perfect fit> to Tribe as an entry point.

Gloves and Mittens

Gloves sometimes offer
a good fit.

Mittens always offer
a pretty good fit.

Jan 9, 2017

Turtle Viability

Hi Ed,

Thanks for taking the time to read this email.

I'm a 19-year old student from Canada who has been trading under my father's tutelage for the past 4 years (both of us would consider ourselves trend followers) with varying success.

I recently stumbled upon this video back-testing Richard Dennis' Turtle Trading Strategy from 1983 to 2014, with the results showing poor performance for the past 6-7 years. With my very limited understanding of programming, the video's maker did seem to implement the trend-following strategy correctly.

So what I'm basically trying to ask here is: Is the Turtle Trading Strategy no longer viable today? Did the video maker not implement it correctly? Why does trend-following as seen in the turtle strategy not appear to "work"?


P.S. Huge fan of your excerpts in Trend Following and Market Wizards
Thank you for raising this issue.

I wonder exactly what you mean by "viable" and how you measure viability.

You might also consider reading the article directly below.

You might consider taking your feelings about uncertainty to Tribe as an entry point.

Jan 9, 2017

Death of Trend Following (Again)


It gets to a point where you just have to laugh. After two years of slow returns, the boo-birds are coming out again and investors are pulling money out of trend following funds. November saw the biggest monthly outflow in three years.

I’m coining a phrase for this repeated investor irrationality when it comes to bailing out of trend following due to sluggish short-term performance - “Same Sh--, Different Drawdown”.

Historical trend following performance speaks for itself. Saying it doesn’t work or it’s dead means you haven’t done any research or work. Sadly, many investors don’t do any work but simply chase short-term performance.

Over the past couple of years, trend following has struggled so now they’re throwing a fit and pulling money out. Forget the long-term diversification benefits or outperformance of buy-and-hold. Short-term performance is the only thing that matters to most people.

Fortunately, investor behavior can serve as a contrarian indicator. Right now, I believe it’s serving us, yet again, by signaling a buying opportunity for trend following.

Excessive Negativity Serves as a Buying Opportunity

Investors have withdrawn capital from trend following funds and have declared it dead many times over the past 15 years: mid-2000, early 2002, mid-2004, mid-2007, early 2010, late 2012 and early 2014.

There are many examples of this before 2000, but I think you get the point. Whenever trend following, or any other investment for that matter, goes into a losing streak, people declare it dead and bail out - often, right near the bottom. It’s actually hilarious to me.

Trend Following Dies



Thank you for sharing your insights.

For additional information on this topic, you might like to see this article.
Jan 9, 2017


Dear Ed,

I recall our talk about seducing women. I consider seduction a form of control, since one person tries to move another person to act in a certain way, according to his/her own interests.

And in Tribe we do not control people. I remember having a hard time trying to integrate both.

Maybe you can share your ideas about control and seduction in a TTP frame. As I write these lines I recall principles of Ericksonian hypnosis (fully accepting the client and their beliefs), and Kaa the snake from Jungle Book (trying to lunch Mowgli).

Best regards,

Thank you for raising this issue.

In the TTP Intimacy-Centric model, we experience, share and receive feelings without trying to change them.

We strive for connection, rapport and intimacy rather than for control.

In the Control-Centric model, you try to bend another person to your will, typically through manipulation and gaining trust.

For an example of Kaa working the Control-Centric model on Mowgli, see:

To Top of Page