[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Trading Systems and Trading Success



PureBytes Links

Trading Reference Links

A few questions and comments on Peter's stuff:

> I wanted to make a point about how a trading system is really just a
> small fraction of the story.  Many traders have probably done a good
> deal of backtesting or system development and many traders have also
> probably tried to find a system that would give them  50% profitable
> trades with a 2:1 r:r.  When they found this system, they probably took
> a couple of trades and had some mediocre results and probably looked for
> another system where they could achieve better results(Story of the
> average trader).

This raises the EXTREMELY interesting question:

   When do we know (when should we conclude) that our system that has been
   working in the past is no longer working and should be abandoned?

Yes it's common for traders to conclude this after the first bad spell,
even if the run of losers is no greater than the greatest over all backtested
history.  But when should we conclude "it's broken now, behavior has changed
and this no longer is working, time to switch"?  Ralph Vince suggests it is
when the greatest ever maximum historical intraday drawdown is exceeded, and
trading should be (could be) resumed when new system equity highs are 
re-achieved (it's proven itself again).  Another way I suppose would be to
utilize some factor (i.e., >1.5x) of the largest ever experienced maximum 
number of losses.

I'm interested in any insights on possible answers to this fundamental 
question.  I don't think it an area that has been properly explored in
trading science.


> Until recently, I was of the belief that it was really important to have
> 60-70% of your trades go your way, but now I realize that this is so NOT
> important.  Sure, if you want to scalp the futures for a living you


Ignoring the need to have steady income, I've found systems that perform
reasonably well across extreme ranges of percent winners versus win 
size/loss size ratios.  However, I've NOT found that the spread of results
is uniform.  My general experience is that the highest profitability
systems (highest average return per trade across all trades, winners and losers)
are frequently the ones with very high winning percentages (80% or higher),
and win size/loss size ratios that are less than 1.0 (average winner is
smaller than average loser).  Excellent systems of this type give me
average trade returns of 700-1200 (single contract).

Trend following systems usually have the opposite characteristic: low
percent winners (20-35%), and win size/loss size ratios that are large
(2:1-4:1 or more).  But I can't get such systems to give me large average
trade returns, the "good" systems are still down in the few hundred per
trade.  And the consistency of return is awful; the drawdowns are much
more severe and the equity curve is choppy as all get out.  As a result,
I've pretty much stopped bothering working on trend following system
development.  (I may still use trend as a component, but "letting winners
run" and "cutting losses short" as a basis for system design...nope, I've
given up on that.)

(I have to say that coffee is an exception when assessed over the last
5 or so years, due to the extreme runups/downs it's had.  Can we really
trust it will do that again soon?  Here again is something I don't like
about trend following: a few big trends in history can lead us to conclude
that we like trend following, but a few big trends isn't really statistically
significant, and the odds of a new one in the reasonable future are pretty
much unknown.)

Could just be I'm a poor system designer in the area of trend following
approaches, and if someone has an approach that scores >500 per trade
on average on a specific market, I'd like to see it.  (The S&P is also an
exception because it's such an enormously upward biased market; as long 
as stops are reasonably deep a system will show piles of profit in the
past in the S&P, and arguable, if we have deep enough pockets, it is the
most awesomely trended market in history and we are insane to not trend
ride it...right?)


> The reason I believe this is because they have mastered
> the art of money management.  The world's best traders will have their
> biggest positions on when the market is going in their favor and their
> smallest positions on when their is no trend.  "Varying the bet size" as
> they call it is the SINGLE MOST IMPORTANT INGREDIENT TO SUCCESS.  I
> emphasize this because over and over again I read about or hear about
> traders who are going into the markets looking to scalp money out of
> these volatile markets and are losing because of: a) the increased
> slippage, b) their own emotional and psychological limitations; c) poor
> money management; d) no real plan on how to increase their position size
> when the market moves into a trending mode.

Your assessment of money management as pyramiding while riding a trend
is all wrapped the fundamental idea that we're trading a trend following
system.  I don't for the reasons explained above.  All my systems that I'm
trading now (3) either exit with profits on a time fuse (i.e., "exit in
3 days" or "exit on first profitable close") or exit on a profit target
("exit when profits = $2100").  No trend riding.

For me, money management is all about compounding position size in a
rational and apppropriately aggressive way from trade to trade, as I
have more or less equity to work with.  There is no question that money
management (=varying position size) is key to taking any small pile into 
a very big pile (i.e, 50-100x larger).  There's nothing wrong with NOT
doing this and taking steady profits (nice work if we can get it).  But
the way I view it:

   - my system isn't going to work well forever;
   - I should maximize my return now while it is working well;
   - I should go for a profit target aggressively (via compounding position
       sizes) and then STOP TRADING to secure my wealth;
   - once secured, start over with a new small pile and try to do it again.

In this sense, running a system is kindof like a macro version of running
a trade.  The key point of my logic is that we can't trust our system
to be nicely profitable forever, so we should try to maximize our profits
in as short a time as we can, take alot of risk along the way in doing so
(in order to minimize the time component), then tuck those profits into a 
nice safe low yielding account, or real estate, or whatever diversification 
is your bag.

No right or wrong here, just my experiences and learnings and personal
conclusions.

-Kevin