A Practical Framework For Managing Risk In Your Portfolio
A recent thread on X.com about stop losses turned into a small
controversy, as this topic tends to do, because it quickly polarizes
people. Here is the link to my post:
https://x.com/ManuelBlay3/status/2085467290200625336
On one side are those who insist you need a stop and that skipping
one is simply reckless. On the other are those who consider a stop a
tool for people who end up getting run by algorithms and market makers,
placed exactly where everyone else’s stop already sits. What follows is a
distillation of that back-and-forth, an attempt to bring some clarity
to when a stop genuinely makes sense and when it does not, which depends
entirely on your personal approach to trading and the strategy you are
actually running.
Path One: Protection Through Breadth, Not Through Stop-Loss
If your edge is fundamental, meaning you are buying businesses
because of earnings power, balance sheet quality, or a mispriced cash
flow stream, a mechanical price stop is often the wrong tool.
Fundamentals move slowly. A drawdown in a sound business is frequently
noise rather than signal, and a rigid stop turns a temporary markdown
into a realized loss right before the thesis has time to play out. The
logical alternative is to buy on good fundamentals, give the trade time,
and sell only when those fundamentals deteriorate. Applying a
price-based stop in that context is not caution; it is self-sabotage.
The protection here comes from position count and position sizing,
not from a stop loss rule. A portfolio of twenty stocks or more,
reasonably sized and not concentrated in a single sector, already
absorbs company-specific risk the way an index does. One name can fall
sharply, and the portfolio survives.
This is also the honest answer to the sharpest objection raised in
the discussion: whether rejecting stops means accepting a catastrophic
loss from a surprise event such as a bad offering or an accounting
scandal. It does not, provided the position was sized correctly in the
first place. A severe loss on a name that was four or five percent of
the portfolio is painful but survivable. A severe loss on a name that
was the entire position is a different problem altogether, and no stop
loss discussion fixes bad sizing after the fact.
There is a practical problem hiding inside this path, though, and it
deserves to be said plainly rather than assumed away. Twenty stocks is
the minimum for proper diversification, yet finding twenty genuinely attractive fundamental ideas through discretionary judgment alone is no easy
task. How do you consistently decide what is good and what is bad? How
much weight should be given to earnings, quality, value, momentum, or
any of the other factors that matter? Most investors, working through
annual reports, model one company at a time and run out of high
conviction ideas well before they reach twenty. Depth of research and
breadth of names pull against each other, and something usually gives: either
the position count shrinks below what real diversification requires, or
the quality of research per name thins out to fill the roster.
But if building a portfolio of twenty high-quality ideas is difficult, deciding when those same twenty stocks should be sold is even more challenging.
Buying is only half the battle. A company rarely goes from attractive
to unattractive overnight, and determining when its fundamentals have
deteriorated enough to justify a sale is often a matter of judgment
rather than a clear-cut decision.
And this is precisely where human nature becomes dangerous.
I was reminded of this while reading the latest “Market Wizards”
book. One trait appears again and again among even extraordinarily
successful traders: learning to control risk was difficult. It wasn’t
simply a matter of knowing that stops or exit rules were necessary. Even
when traders had them, the temptation to override them was often there.
I recognize the problem because I experienced it myself. When I
started trading, a stock could hit my stop, and I would think: let’s
give it a little more room, it will recover. And the dangerous thing is
that this often worked, maybe eight times out of ten. But the other two
times, the stock didn’t recover. The loss kept compounding and
eventually became enormous. Those few disasters could wipe out much of
what had been gained from all the occasions when overriding the stop
appeared to have been the correct decision.
The problem, therefore, is not merely designing an exit rule. The problem is obeying it.
The solution I have settled on is what I call “quantamentals”. The
underlying inputs are still fundamental: growth, value, quality,
profitability, earnings revisions, the same building blocks any
fundamental investor would recognize, but they are processed on a
quantified, computerized basis rather than judged name by name through
memory, instinct, and conviction.
This approach solves two problems at once. First, it makes genuine diversification practical.
Instead of developing twenty separate high-conviction narratives, the
computer evaluates the entire investment universe using the same rules
and ranks each stock accordingly. I don’t need twenty compelling
stories. I need twenty stocks that objectively score better than the
alternatives.
The second advantage may be even more important: quantification imposes discipline on the sell side.
Every rebalance produces stocks that deserve to enter the portfolio and
stocks that no longer deserve to remain in it. If five better-ranked
stocks need to come in, cash must be raised by selling stocks whose
rankings have deteriorated. That completely changes the psychology of selling.
There is no “let’s wait another week.”
No “maybe it will come back.”
No emotional attachment to the original thesis.
The rule for selling is not a price level; it is a rank-based exit.
A position is sold when its “quantamental profile” deteriorates past a
defined threshold, not when the story I originally told myself about the
company stops feeling true. That distinction matters more than it
sounds. A stock held because of a narrative gets defended long after the
evidence has changed, because human beings are remarkably good at
finding new reasons to defend an old decision. A stock held because of a
quantified score has no story to defend. It has a ranking, and when
that ranking deteriorates sufficiently, it goes.
I have come to think of the portfolio almost like the shelves of a
supermarket. Shelf space is scarce. If some oranges are deteriorating
while fresher, better merchandise is waiting to take their place, the
supermarket owner does not become emotionally attached to the old
oranges. He clears the shelf and replaces them.
A quantitative portfolio works much the same way. Capital is scarce
shelf space. A stock does not need to become a disaster before it
deserves to be sold. It merely needs to become less deserving of that
scarce capital than the alternatives. This is a subtle but crucial
difference: I am not necessarily selling because something
terrible has happened to the company or because its price has crossed an
arbitrary line, but because better opportunities now exist relative to
the rest of the investment universe.
This is why, in a diversified quantitative strategy, risk management
does not have to mean placing a mechanical stop beneath every position.
Risk is controlled through small position sizes, diversification across
sectors, systematic ranking, and the disciplined replacement of
deteriorating holdings. The computer does not hope. It does not fall in
love with a stock. And, most importantly, it does not second-guess the
sell signal.
Jim Simons and Renaissance Technologies built the most successful
track record in the industry, largely without discretionary stop-losses
on individual positions. Their protection came from an entirely
different architecture: thousands of small, statistically independent
bets, each with a modest edge, held briefly, sized so that no single
position could meaningfully damage the fund, and diversified across so
many uncorrelated signals that the portfolio behaved closer to an
insurance book than to a stock picker’s book. Risk was managed at the
level of the whole system through position sizing, correlation control,
and turnover, not at the level of any single trade’s exit price.
But some of you may still think that a stop loss would add an
additional layer of protection to an already well-diversified portfolio
based on “quantamentals”. I am sorry to disappoint you. The addition of
stops to such strategies always results in a sharp degradation of
performance.
The image below shows one of my quant strategies without a stop-loss.
A nice equity curve with an annual performance of 32.98% and a -42%
drawdown.
Now, if we add a 15% stop-loss, we effectively ruin the strategy, as
the chart below shows. Performance is nearly halved, falling to 15.24%,
while the maximum drawdown improves by only 31%, declining to 29.35%.
This relatively modest reduction in drawdown comes at far too high a
cost in performance.
The deterioration in risk-adjusted returns is equally striking: the
Sharpe ratio falls from 1.36 without a stop loss to just 0.79 with the
15% stop.
Furthermore, turnover explodes to 736%, adding substantial
transaction costs and slippage and making the strategy extremely
difficult to trade in practice.
Therefore, if one invests based on “quantamentals”, one should steer away from using stop-losses.
Path Two: Protection Through The Stop-Loss
If instead your edge is technical, such as buying
breakouts, moving-average crossovers, price patterns, or swing trading,
and, to make things worse, you have few positions open at the same time,
a stop is not optional. Furthermore, such “technical” traders tend to
have a very concentrated portfolio, often with fewer than 5 positions.
In such a case, a stop-loss is a vital tool for survival.
But not all stops are equal, and this is where many traders get it wrong. There are dummy stops and intelligent stops.
Dummy stops
A dummy stop is an arbitrary number applied uniformly across every
trade, most commonly a fixed percentage such as ten or fifteen percent
below the purchase price, regardless of the stock, its volatility, or
the chart. It is easy to implement, and it is also blind. A fixed
percentage means almost nothing on a stock that regularly swings eight
percent in a week, and it means far too much on a stable low-volatility
name where that same move already signals something has genuinely
broken.
Intelligent stops
An intelligent stop is derived from the security itself and/or
context rather than imposed on it. Three layers are worth building, in
increasing order of sophistication.
Structural or chart-based stops. A more discerning
trader places the stop beneath a level that has actual meaning on the
chart, a prior base, a broken resistance that should now act as support,
or a key moving average such as the 50-day, the 150-day, or the 200-day
line, chosen depending on the entry pattern. The logic is simple. If
price falls back through the exact level that justified the trade, the
original reason for owning the stock is no longer valid, and the
position should be closed on its own merits rather than at an arbitrary
distance.
One objection worth answering directly is the claim that placing a
stop at an obvious technical level is pointless because everyone else’s
stop sits at the same price, and that level gets run before the real
move happens. There is truth in it. Round numbers, prior lows, and
textbook moving averages are exactly where liquidity clusters, and price
can and sometimes does push through a crowded level before reversing.
The correct response to that is not to abandon technical stops; it is to
stop placing them exactly on the obvious number. A volatility buffer
below the structural level, or the confirmation requirement described
above, both exist specifically to avoid being the easy liquidity at a
level everyone can see on the same chart.
Volatility-based stops. Instead of a fixed
percentage, the stop is set as a multiple of the stock’s own average
true range or realized volatility, giving the best of both worlds,
avoiding a premature exit on a normally noisy name while still keeping
risk genuinely managed. A volatile name gets more room, a quiet name
gets a tighter leash.
Confirmation-based stops. This is the most refined
layer and the one closest to classical Dow Theory thinking. A single
instrument breaking a significant technical level can be a fakeout, a
stop run, or noise driven by low liquidity. Requiring confirmation from a
related asset filters out a meaningful share of false signals. If the
stock breaks its level but the related asset holds firm, the break is
more likely noise. If the two assets break together, the signal is real,
and the exit should follow without hesitation. This is the same
principle that governs price and Advance Decline line confirmation in
classical Dow Theory, applied here at the level of a single position’s
exit rule.
The image below, whose full explanation you may find in this post,
provides an excellent example of stops based on the principle of
confirmation. The piercing of SLV (silver ETF) at a relevant low,
unconfirmed by GLD (gold ETF), proved to be a fakeout.
The Third Path: The No Stop Fallacy For Day Traders
A different objection surfaces often in these threads: the claim that
day traders do not need stops at all because closing every position
before the close removes overnight gap risk entirely. This is true as
far as it goes, and it is also close to the most unhelpful answer people
give to the stop question, because it treats day trading as a fixed
strategy rather than a capacity-constrained one.
A day trader working with a modest account is right that flattening
at the close solves the gap problem, and there is nothing wrong with
building a living around that approach. But day trading does not scale
the way the argument implies. Executing meaningfully large size within a
single session, without moving the price against yourself, becomes
progressively harder as the deployed assets grow, because intraday
liquidity in any single name is finite and a large order worked in a few
hours leaves a visible footprint. The trader who is right to skip stops
at a modest account size is not automatically right once that account
reaches real scale, not because the logic about gaps changed, but
because pure day trading stops being available at that size. Growing an account seriously eventually forces a longer time frame,
and a longer time frame is exactly where the gap risk the argument
dismissed comes back, along with the need for a genuine stop discipline.
Furthermore, day traders who trade without stop losses often overlook
the fact that a stock can collapse dramatically even within a single
trading session. If you run a highly concentrated portfolio, which is
often the case with only three or four positions, a single stock falling
30% intraday can inflict catastrophic damage on your overall portfolio.
In other words, the absence of overnight risk does not eliminate the
need for risk management. Even over the course of a single day,
concentrated positions can produce losses large enough to jeopardize an
entire trading account.
The Weak Point Of Stop-Losses: The gap
If you trade a concentrated portfolio of a few stocks, the risk of
one gapping down would be unbearable, rendering any stop-loss moot. This
is a hidden risk for which there is no “standard” protection (unless
you use puts or similar sophisticated hedging strategies). A stop-loss
will not protect you against a stock that drops 60% overnight. The
takeaway is clear: even if you are a technical trader for whom the
stop-loss is a must, you must strive for a diversified portfolio.
I have a hard time understanding traders who brag about their “high
conviction,” “focused” portfolio. In this game, the objective is
survival. Survive, and performance will follow.
Choosing Your Path
The mistake to avoid is not picking the wrong philosophy; it is
borrowing the confidence of one philosophy while practicing another. A
concentrated breakout trader who refuses to use a stop because
Renaissance did not use one is not being sophisticated; he is simply
unprotected because he lacks the breadth of positions and the
statistical edge that made the stopless approach survivable at scale.
Equally, a diversified fundamental investor who panics into a tight
mechanical stop on every position is importing a discipline built for a
different game entirely.
The honest question to ask before every position is not, “What is my
stop?” It is, which game am I playing, and does my protection actually
match it?
Sincerely,
Manuel Blay
Editor of thedowtheory.com
