8

The Long/Short Toolbox

Long-only portfolios are by definition correlated with the market. Introducing a short component enables managers to engineer the type of performance they want to deliver. We will make long/short portfolio management accessible to all dedicated market participants in this chapter.

Creating a long/short portfolio requires the consolidation of two relative portfolios: a long and a short book. Introducing a short book adds another layer of complexity. An effective way to manage this transition is to begin with the objectives in mind. We want to manage risk defined across liquidity, correlation, volatility, and performance (drawdown). Then, we want to concentrate on a handful of variables that have the highest impact on these objectives: gross and net exposure, net beta, and concentration, which make up your investment toolbox when constructing your strategy.

In this chapter, we will concentrate on risk management. We will introduce a formula that has the potential to transform how you manage your portfolio: no more fear of missing out (FOMO), guesswork.

Along the way, we will cover the following topics:

• The Long/Short toolbox
• Gross exposure
• Net exposure
• Net beta
• Concentration
• Exchange exposure
• Sector exposure

It is tempting to focus on the minutia of stop loss effectiveness, regime dependence, trading attribution, microstructure, etc. What we really need to run a long/short portfolio is a solid understanding of the big blocks that will shape the risk profile and ultimately returns. If the portfolio was a car, we would build the dashboard. We only need to look at the fuel gauge, engine temperature, and speed to drive safely. Everything else is important yet not essential. Let's start with a brief overview of the long/short toolbox.

The Long/Short toolbox

If we want to have ten definitions of risk, all we need to do is gather five market participants in one room and let them explore the subject in depth. There is no shortage of creativity on the subject of risk. However, as a practitioner, here are the four dimensions of risk that will keep you awake at night and potentially shorten both your gain and professional life expectancies:

• Liquidity: if we can't exit, we are by definition trapped
• Correlation: in a down market, the only thing that goes up is correlation
• Volatility: whipsaws that shake us out prematurely
• Performance: drawdowns: magnitude, frequency and duration

Those are the four main facets of risk. We have four major levers to manage those four risks.

• Gross exposure is leverage. This is "how much risk?"
• Net exposure is the apparent directionality. This is "are we bullish, bearish or neutral?"
• Net beta answers is the residual directionality. This is "how market‑sensitive?"
• Concentration is the number of positions. This is "how fragile?"

All this can still seem a bit esoteric. Let's summarize this in a table to visualize how each tool affects each component of risk:

Table 8.1: Risk and Exposures

The preceding table shows how each exposure affects risk. For example, a concentrated leveraged portfolio will have high volatility and vice versa.

Our objective is to understand how gross, net, and concentration ultimately affect performance.

Let's import the libraries and start coding.

Technical requirements

For this chapter and the rest of the book, we will be working with the pandas, numpy, yfinance, and matplotlib libraries. So, please remember to import them first.

import pandas as pdimport numpy as npimport yfinance as yf%matplotlib inlineimport matplotlib.pyplot as plt          

Next, let's explore the exposures one after the other starting with gross exposure.

Gross exposure

"石橋を叩いて渡る (ishibashi wo tataite wataru), knocking on a strong stone bridge before crossing it", metaphor for prudence

– Japanese proverb

Gross exposure is the absolute sum of long and short books. Selling stocks short generates cash. This can be used to enter additional long positions. In theory, leverage could be increased ad infinitum. In practice, prime brokers limit leverage to 3 or 4 times to limit their counterparty risk.

Let's define the functions for the exposures:

def market_value(position, price):    return position * pricedef net_asset_value(market_values, cash):    return cash + market_values.sum(axis=1)def gross_exposure(market_values, nav):    return market_values.abs().sum(axis=1).div(nav)def net_exposure(market_values,nav):    return market_values.sum(axis=1).div(nav)def net_beta_exposure(market_values, beta, nav):    beta_weighted_mv = market_values.mul(beta)    return beta_weighted_mv.sum(axis=1).div(nav)                            

These functions calculate key portfolio metrics used in dynamic risk management and position sizing. Let's understand them in depth next.

1. Market Value: computes the market value of each position by multiplying shares held by their current price. For a multi-asset portfolio, returns a series where each element is the dollar value of that position.

   

2. Net Asset Value (NAV): Total portfolio value equals cash on hand plus the sum of all position market values. This is the denominator for all exposure ratios.

   

3. Gross Exposure: measures total leverage by summing absolute values of all positions (long and short treated equally) divided by NAV. Values > 1 indicate leverage; = 1 indicates fully invested; < 1 indicates underutilized capital.

   

4. Net Exposure: measures directional bias by summing signed position values (long positive, short negative) divided by NAV. Range: [-1, 1].

   

5. Net Beta-Weighted Exposure: adjusts net exposure for each position's systematic risk (beta). Positions with high beta contribute more to portfolio market sensitivity, even if dollar values are equal.

   

Our job as algorithmic traders is not to take entry signals and stop losses as long as there is cash available. Since the process is semi-automated, our job really comes down to managing risk. We therefore need to set targets and closely monitor those aggregates.

Next, let's see how gross exposure affects risk.

Gross exposure has a direct impact on:

• Liquidity: The higher the leverage, the bigger the positions. This has a direct effect on market impact.
• Volatility: Leverage magnifies the volatility of returns.
• Performance: Leverage magnifies returns: A paltry 0.1% at low leverage can be juiced up to 0.5% and so on.
• Concentration: If the objective is to keep volatility low, the number of names should move in tandem with leverage: higher gross means more names and vice versa.

Let's expand on every item, starting with concentration, and work our way back. Concentration has an obvious impact on gross exposure. The more strategies, the more positions. The more positions, the bigger the exposure.

Volatility in this context stands for drawdowns. Leverage is a double-edged sword. It will make and then break a business when not wielded carefully. Deleveraging is not always beautiful, as Ray Dalio likes to say.

Gross exposure and leverage are often used interchangeably. There is however a subtle nuance. Some market participants used to hold the belief that as long as both books were balanced, then leverage could be increased to juice up the returns. In theory, it works. Pairs trading is the immediate example that comes to mind. Since the relationship between pairs is stationary, then risk should be contained.

The natural temptation for left skewed/positive mode (mean reversion, arbitrage) strategies is to increase leverage. After all, if the win rate is high, the relationship stable, yet returns modest, then leverage seems to be the solution to juice up returns. This is an evolutionary stage of development for those strategies. Managers become overconfident and lever up. It all works well until it ends in one swift blow. Left skewed strategies go out of business because of a few unpredictable devastating losses.

Gross exposure has a direct impact on liquidity. It tends to evaporate when needed the most. The quants debacle in 2007 illustrates how overleveraged funds with seemingly low market exposure found themselves in a messy elephant stampede in a China shop. In summary, all the multi-strategy funds were milking the same pairs trading strategies. They were forced to deleverage to post collateral in other strategies. This forced them to cover. The cross-sectional volatility killed one year's performance in one week. Investors were not forgiving of this sudden volatility.

Gross exposure is ultimately a reflection of market participants' confidence in the markets. The more confident they are, the more they tend to increase leverage to capture opportunities. Nothing looks more like a kid in a candy store than a growth manager in a tech bubble. Conversely, market participants tend to reduce their gross and increase their cash exposures to protect capital in periods of turbulence or uncertainty. Strategically managing gross exposure can be a powerful tool to boost performance in good times and protect capital in rough markets.

Market participants tend to manage gross exposure in one of two ways. Either they stick to a fixed level of exposure, or they take an unstructured approach. Fixed exposure participants stick to the same level of gross exposure throughout the cycle. The rationale for a constant level of gross exposure is simplicity. Participants like to keep the number of names to a level they can safely manage. They may leave money on the table in good times, yet they keep drawdowns to a manageable level during losing streaks. On the other hand, unstructured participants get bulled up in good times, load up the truck, and hoard cash in bad times. This translates into massive outperformance followed by lackluster returns.

Let's illustrate the concept of de/releveraging with an example from . We explored various position sizing algorithms at different risk levels across the global autos stocks. We did not impose any limitation on the gross exposure level. The investment universe was comprised of only 9 stocks. As such, gross exposure remained at an average of 0.8 as Figure 8.1 shows.

Chapter 6, Gross Exposure Comparison">

Figure 8.1: , Gross Exposure Comparison

Gross exposure of 0.8 with 9 stocks meant that position sizes were on average a little below 10%. High concentration led to high volatility. Drawdowns were already significant.

Chapter 6, Drawdown Comparison Scenario 3">

Figure 8.2: , Drawdown Comparison Scenario 3

Drawdowns coincide with disorderly deleveraging. It is therefore prudent to err on the side of cautious optimism.

This leads us to a powerful indicator that could potentially change your game.

Dynamic risk appetite management

We all understand the relationship between gross exposure and drawdowns: the bigger the bets, the bigger the potential losses. Our dilemma is: we want to deploy as much capital as possible, but we also want to quickly trim exposures when things get wobbly. If our objective is to deliver attractive returns with reasonably low volatility, there may be a more scientific approach. This approach is a variation on a ruthless technique employed by a famous hedge fund. At -5% drawdown year to date, they reduce assets under management (AUM) by 50%. At -7.5% they stop loss to 0. It has worked spectacularly for them, yet there may be a smoother approach. What we want is a responsive tool that will collapse the gross exposure when there is a shock and promptly re-accelerate.

Here are the steps:

1. Decide on the upper and lower bands for gross exposure. Remember that deleveraging has serious market impact. A reasonable range between the lower and upper band should be somewhere between 50–100% for multi-strategies and 40–60% for single strategy.
2. Decide on the magnitude of drawdown to trigger the minimum target gross exposure. In the example above, a -5% drawdown would collapse the gross exposure by 50%.
3. Plug in the equity curve. The risk appetite function will automatically calculate the target gross exposure.

Let's convert this into code.

def risk_appetite(equity_curve, max_drawdown_tolerance,                           min_risk, max_risk,                           smoothing_span=20,                           curve_shape='linear',                           drawdown_window=0):    """    curve_shape : {'linear', 'aggressive', 'conservative'}, default='linear'        'aggressive': Scale up quickly when recovering (convex < 1)        'conservative': Scale up slowly when recovering (concave > 1)        'linear': Proportional response    """    equity = pd.Series(equity_curve)    if drawdown_window > 0:        running_max = equity.rolling(window=drawdown_window, min_periods=1).max()        else:        running_max = equity.expanding().max()  # All-time high    drawdown = equity / running_max - 1    normalized = 1 - np.minimum(drawdown / max_drawdown_tolerance, 1)    smoothed = normalized.ewm(span=smoothing_span).mean()    # mapping: convex (<1 aggressive), concave (>1 conservative)    power_map = {'aggressive': min_risk / max_risk, 'conservative': max_risk / min_risk, 'linear': 1}    power = power_map.get(curve_shape, 1)    transformed = smoothed ** power    risk_appetite = min_risk + (max_risk - min_risk) * transformed    return risk_appetite                                                      

Dynamically adjusting portfolio risk allocation based on current drawdown, the function scales risk between minimum and maximum thresholds, reducing exposure during drawdowns and increasing it near equity peaks.

Here is a brief explanation of the code:

1. Calculating percentage decline from peak equity:
   1. Uses expanding window (all-time high) by default
   2. Optional rolling window for shorter-term drawdown measurement
2. Normalizing drawdown: we scale drawdown relative to maximum tolerance. We convert the absolute drawdown to a 0-1 range:
   1. 0 = at maximum drawdown tolerance (minimum risk)
   2. 1 = at peak equity (maximum risk)
3. Smoothing the Signal by applying exponential moving average. This reduces noise and prevents chaotic risk changes.
4. Applying response curve: We transform the normalized drawdown based on risk preference:
   1. Linear: classic proportional response (risk scales linearly with drawdown)
   2. Aggressive: Convex recovery (risk increases quickly as drawdown worsens)
   3. Conservative: Concave recovery (risk increases slowly, stays defensive longer)
5. Scaling to Risk Range: Map transformed value to actual risk allocation. Interpolating between min_risk and max_risk.

This function returns a pandas Series of risk appetite values between min_risk and max_risk, inversely correlated with portfolio drawdown. Higher values near equity peaks, lower values during drawdowns.

The best real life analogy for this function is driving. We simply want to drive at cruising speed. Unfortunately, impredictable incidents happen. We need responsive brakes. When out of danger, we want to re-accelerate and keep on cruising. This risk appetite runs gross exposure close to optimal in normal conditions, but collapses it as soon as markets turn wobbly. It re-accelerates as soon as equity curve starts to pick up.

Let's take that function for a spin. In the following block of code, we simulate a few equity curves using geometric Brownian motion. We apply our risk appetite with the linear, aggressive, and conservative shapes. We plot the scenarios along with the risk appetites by shape.

1. First, let's define the geometric Brownian motion function:

   def simulate_equity_curve(initial_capital, annual_return, annual_vol, days, dt=1/252, seed=42):      """      dS = μ*S*dt + σ*S*dW      """      np.random.seed(seed)         mu = annual_return * dt      sigma = annual_vol * np.sqrt(dt)      equity = np.zeros(days)      equity[0] = initial_capital        # Euler scheme: S(t+dt) = S(t) * (1 + mu*dt + sigma*dW)      for t in range(1, days):          dW = np.random.normal(0, 1)          equity[t] = equity[t-1] * (1 + mu + sigma * dW)      return equity

   Simulating an equity path for strategy/risk testing:

2. Inputs: initial_capital, annual_return, annual_vol, days, dt=1/252, seed
3. Method: Daily update with Euler GBM
4. μ = annual_return * dt, σ = annual_vol * sqrt(dt)
5. equity[t] = equity[t‑1] * (1 + μ + σ * dW) with dW ~ N(0,1)
6. Output: NumPy array of length days (equity curve)
7. Notes: Set seed for reproducibility; keepdt=1/252 for daily

   This generates a geometric brownian motion (GBM) that will be used to simulate equity curve paths. Next, we will write a few scenarios:

   np.random.seed(42)  initial_cap = 1_000_000  annual_ret = 0.12  annual_volatility = 0.18  num_days = 252 * 3  # 3 years    scenarios = {}  for scenario_id in range(1, 4):      equity_sim = simulate_equity_curve(          initial_capital=initial_cap,          annual_return=annual_ret,          annual_vol=annual_volatility,          days=num_days,          seed=scenario_id * 100)      scenarios[f'Scenario {scenario_id}'] = equity_sim

   Showing equity paths and risk appetite under linear, aggressive, or conservative curves per scenario.

8. Simulate 3 geometric Brownian motion paths (different seeds, 3y).
9. Set min/max risk and drawdown tolerance.
10. For each scenario, plot equity (left) and risk appetite (right) with markers every 50 points.

    This generates one figure per scenario with equity plus comparative risk-allocation curves bounded by min/max lines.

    We can run our little function. We will generate side-by-side plots for each scenario showing equity evolution and dynamic risk-appetite response across three curves. This is a long block of code, primarily because of the charting function. We have therefore commented each step for clarity purposes and split it into small, digestible steps:

    # Calculate risk appetite for each scenario with different curve shapes  max_dd_tol = -annual_volatility * 0.4  min_risk_val = 0.02  # 2% minimum risk  max_risk_val = 0.10  # 10% maximum risk

11. Initializing max drawdown tolerance, min/max risk bounds. We set the drawdown tolerance at 40% of the annual volatility. The objective is to respond before the damage is done.

    # Create separate figure for each scenario  for scenario_name, equity_path in scenarios.items():

12. Opening a loop to iterate over each scenario's equity path. Everything that follows is contained within this loop.

    fig, axes = plt.subplots(1, 2, figsize=(16, 5))  # Left: Equity curve      ax_equity = axes[0]      ax_equity.plot(equity_path, linewidth=2, color='steelblue')      ax_equity.fill_between(range(len(equity_path)), initial_cap, equity_path, alpha=0.2, color='steelblue')      ax_equity.axhline(y=initial_cap, color='black', linestyle='--', alpha=0.5, label='Initial Capital')      ax_equity.set_title(f'{scenario_name} - Equity Curve', fontsize=14, fontweight='bold')      ax_equity.set_ylabel('Equity ($)', fontsize=11)      ax_equity.set_xlabel('Trading Days', fontsize=11)      ax_equity.grid(True, alpha=0.3)      ax_equity.legend(fontsize=10)        # Right: Risk appetite by curve shape with markers      ax_risk = axes[1]        # Define markers and colors for each shape      shape_styles = {          'linear': {'marker': 'o', 'color': '#1f77b4', 'markersize': 8},          'aggressive': {'marker': '^', 'color': '#ff7f0e', 'markersize': 8},          'conservative': {'marker': 's', 'color': '#2ca02c', 'markersize': 8}

13. For each scenario, we create a 1×2 subplot: equity plot (left) with fill and initial-capital reference line; risk appetite plot (right) with three colored/marked lines.

    #Calculate risk appetite for each curve shape      for shape in ['linear', 'aggressive', 'conservative']:          risk_app = risk_appetite(equity_path, max_dd_tol,                             min_risk_val, max_risk_val,                             smoothing_span=20,                             curve_shape=shape,                             drawdown_window= 126

14. Computing risk_appetite() for each shape; plot as % with markers every 50 days.

           style = shape_styles[shape]          # Plot with markers at every 50th point for clarity          ax_risk.plot(risk_app * 100,                      linewidth=2,                      label=f'{shape.capitalize()}',                      alpha=0.8,                      color=style['color'],                      marker=style['marker'],                      markevery=50,                      markersize=style['markersize'],                      markerfacecolor=style['color'],                      markeredgecolor='white',                      markeredgewidth=0.5)

15. Adding grid, labels, legends, and min_risk/max risk horizontal reference lines to risk plot.

      # Secondary axis for drawdown (keep subtle to avoid clutter)      dd = pd.Series(equity_path) / pd.Series(equity_path).cummax() - 1      ax_dd = ax_risk.twinx()      x_dd = np.arange(len(dd))      ax_dd.fill_between(x_dd, dd * 100, 0, color='red', alpha=0.1, step='mid')      ax_dd.plot(x_dd, dd * 100, color='red', alpha=0.4, linewidth= 2, linestyle='--', label='Drawdown')      ax_dd.set_ylabel('Drawdown (%)', fontsize=10, color='red')      ax_dd.tick_params(axis='y', labelsize=9, colors='red')      ax_dd.set_ylim(dd.min() * 120, 0)      ax_dd.grid(False)        ax_risk.set_title(f'{scenario_name} - Risk Appetite by Shape', fontsize=14, fontweight='bold')      ax_risk.set_ylabel('Risk Appetite (%)', fontsize=11)      ax_risk.set_xlabel('Trading Days', fontsize=11)      ax_risk.set_ylim([min_risk_val*100, max_risk_val*100])      ax_risk.grid(True, alpha=0.3)      ax_risk.legend(fontsize=10, loc='best')      ax_risk.axhline(y=min_risk_val*110, color='red', linestyle=':', alpha=0.5, linewidth=1.5)      ax_risk.axhline(y=max_risk_val*110, color='green', linestyle=':', alpha=0.5, linewidth=1.5)        plt.tight_layout()      plt.show()

16. Using plt.tight_layout() and plt.show() to render each figure.

    Generating side-by-side plots for each scenario showing equity evolution and dynamic risk-appetite response across three curve shapes (linear/aggressive/conservative).

    As is often the case with charting functions, the explanation is inversely proportional to the length of the code:

    This produces one figure per scenario (total 3 figures) with paired equity and risk-appetite subplots. This allows visual comparison of how each curve shape (linear ≈ proportional, convex ≈ aggressive recovery, concave ≈ conservative recovery) adapts during market cycles.

Let's publish each scenario. Some shapes probably look all too familiar to some market participants out there. Let's start with scenario 1:

Figure 8.3: Scenario 1 Equity Curve Simulation, Risk Appetite by Shape

We have the equity curve on the left and the risk appetite by shape on the right. In all scenarios, aggressive is the top, linear the middle, and conservative the bottom lines. The drawdown is on the right-hand scale.

The linear/proportional method is the diagonal line. Risk will be reduced to its minimum when the tolerance is reached. There is no acceleration or dampening effect. It is still far more responsive than absorbing a drawdown. At one third of the annual expected volatility, risk appetite collapses to its minimum. It stays down as long as the equity curve does not give signs of recovery.

This scenario epitomizes trend followers' equity curves. They make lots of money in up or down markets, only to give back some during sideways markets. This oscillator was designed for this shape of equity curves. Drawdowns are inevitable. The difficulty is to arbitrage the trade-off between collapsing risk early and re-accelerating fast.

Let's go through scenarios 2 and 3:

Figure 8.4: Scenario 2 Equity Curve Simulation, Risk Appetite by Shape

This scenario is known as round trip. Every single market participant has been through this. We hop on to something that works. We ride the way up. We become cocky and overconfident: "this time it's different." Then, the market starts to get wobbly. We watch profits melt away and of course we rationalize this with some: "but the market is wrong!"

Now, here is a less emotional way to manage risk. We may be right. We may be wrong. We just don't know. Deep down, all we want is to participate if things pan out. This is symptomatic of the FOMO. This oscillator takes care of the deceleration/acceleration. We will not miss the boat and we will get out of a sinking ship.

Finally, let's go through scenario 3:

Figure 8.5: Scenario 2 Equity Curve Simulation, Risk Appetite by Shape

This scenario illustrates sideways markets. After a prolonged rally or bear, markets tend to take a breather. How to allocate risk in those times is the essence of risk management. The oscillator does it for us.

The aggressive is more responsive than the linear shape. Sideways periods require vigilance. They precede breakouts or breakdowns. Sometimes, markets seem to tank only to rally rapidly. Those are known as bear traps. As a short seller, you need to stay vigilant and agile.

Overall, this risk appetite oscillator seems to do the job. This function alone is worth the price of the book. It dynamically manages the amount of capital to be deployed. Everything else flows from that.

The risk_appetite() oscillator deploys the optimal amount of capital based on the shape of the equity curve. In , we will take a deep dive into this oscillator.

Once we are confident about deploying money on the markets, we need to bet on which way they will go, which leads us to net exposure.

Net exposure

Market neutral refers to net exposure zero. Shorts balance longs. In theory, this should cushion market risk. In practice, some of the most spectacular blow-ups, such as Long-Term Capital Management (LTCM), were "market neutral" funds. Net exposure has a direct impact on:

• Liquidity: is one of the most overlooked and critical components. Long and short positions have opposite dynamics. To keep net exposure low, the short book needs to be constantly replenished. Meanwhile, the supply of borrow is finite. This leads to an increase in borrowing costs. Always keep an eye on borrow utilization. Do not let it go past 40%. Similarly, do not let your borrowing cost creep up on You with adjusting your position size.
• Correlation: The lower the net exposure, the lower the correlation. Mutual funds have a correlation of 1, for instance, meaning they mirror the market gyrations. Markets go up, so do mutual funds, and vice versa on the way down.
• Volatility: Net exposure has the largest impact on volatility. The lower the net, the lower the volatility. Targeting zero net exposure, however, has its own systemic risk. It requires constant adjustment of the long and short books. Successful longs grow, while successful shorts shrink. It works well in stable markets but can be extremely painful during major regime changes.
• Performance: The lower the net exposure, the lower market directionality influences returns.
• Yield: Short-sellers are liable for dividends. Low net exposure implies that dividends received on the long side are canceled out by dividends paid on the short side. This has implications on stock selection on the short side: high-dividend yield stocks are often under-represented on the short side.

Net exposure is not a risk measure. It is a view on directionality. Most market participants implicitly believe that reducing the net exposure will reduce risk. This is not true, and there are numerous examples to substantiate that claim. The most telling example is the collapse of the quants in 2007. Quant shops in the US were running pairs trading strategies as far as Japan and Europe. Pairs trading is market neutral by definition. Those strategies were deemed low risk. When the debacle happened in the summer of 2007, those shops found themselves in the same trades that they had to unwind at the same time. There was not enough liquidity. Market impact was devastating.

Another example was the 2008 Lehman shock (GFC). Hedge fund managers promptly collapsed their net exposure from 50% to sub 10%. They were talking a bearish game. Every newsletter, every interview sounded like financial Armageddon was upon us. And yet, their net beta was in positive territory. They were residually correlated to the index. As such, they kept on bleeding performance.

Net exposure can be a deceiving metric. Low net exposure does not necessarily mean low correlation. Most market participants resort to several expedients to bring their net exposure down. The most classic technique is to oversize their few short bets. Concentration increases volatility. Another technique to reduce net exposure is to sell futures, which have lower beta than single stocks. Both techniques have a material impact on correlation and volatility. But, net exposure is to portfolio management what price-to-earnings ratio is to valuations. It is a good enough shortcut, but it should not be the sole basis for a decision. It is better to contextualize net exposure along with other variables. It is therefore important to go past the net exposure headline figure and look at the composition of the portfolio through net beta.

At the beginning of the chapter, we showed the chart on gross exposure for . Let's have a look at the net exposure.

Chapter 6, Net Comparison Scenario 3">

Figure 8.6: , Net Comparison Scenario 3

The net exposure oscillates from +/-1. In theory, this might be the best way to navigate Mr Market's frequent mood swings. In practice, this might be a little hard to execute. Market participants generally have no problem running at a net exposure of 30% and above. Bullish is the factory setting for most market participants.

In comparison, sub-zero net exposure seems to be a physiological malfunction. Very few market participants are brave enough to venture into the frigid minus territory of their net exposure. And there are at least 3 good reasons for that:

1. Liquidity: trading volume dries up on the short side.
2. Volatility: the short side is notoriously volatile. Bear market rallies are treacherous. It is difficult to maneuver an entire fund from net short to net long.
3. Discomfort: for 8 years, I woke up every morning with a net exposure at -100%. I envied the blissful simplicity of the lives of the coffee shop employees and they probably envied mine in return.

A more sensible approach is to cap net exposure between bands. People who are serious about their long/short mandates will gravitate toward lower exposures, while opportunists will more freely oscillate. Average net exposure is therefore a quick and dirty way to reclassify long/short market participants. Here are some broad buckets that participants fall into:

• Above +50%: Those are special situation long-only funds charging hedge fund fees. They offer something else than downside protection. They are not trying to hedge the downside with the short book. They have views on their short positions as individual stocks.
• Above +30%: Those market participants are usually referred to as directional hedge funds. This is the most relatively unsophisticated bunch of the long/ short club. They are usually asset aggregators who pretend to know what they are doing on the short side yet they struggle in the idea generation department. Their characteristics are high correlation, limited upside participation, and zero downside protection. They are the sushi conveyor belt of hedge funds. When one fails, the next colorful one rolls around.
• Net exposure contained between -20% to +20%: The most serious long/ short players operate in that space. They have understood that risk management is the business and exposure control the tools. This low net exposure offers low volatility returns but still benefits from limited market directionality.
• Sub -10% to +10% net exposure is the province of market-neutral funds. Alpha strictly comes from stock selection and position sizing. Only a few strategies can accommodate that level of exposure. Those are usually arbitrage, pairs trading. Bear in mind that low net exposure does not equate to low risk. The risk with those funds is often in the tail. It is difficult to constrain exposures at all times. This often comes at the cost of some factor. Think of it as rewinding a spring. At some point, pent-up energy may spring up and cause damage.

We just saw that low net exposure does not equate to low risk. It is now time to talk about Beta.

Net beta

"Market exposure is a choice; it should be a deliberate one."

– Cliff Asness

In early 2005, the Japanese equities market had an epic year. It felt like this time, it was different. The party came to a screeching halt when the Japanese authorities decided to arrest Takafumi Horie, CEO of Livedoor (JT:4753) and symbol of the new Japan in January 2006. High-flying small caps rediscovered Newtonian physics.

We were long high-flying small caps with esoteric business models, and short a few asthmatic "structural shorts" along with index futures. The fund manager quickly responded to the crisis by selling futures. Despite a reasonable +30% net exposure, the ship was taking on water fast. As a self-appointed risk manager, I promptly brought to his attention that we were synthetically exposed on the beta, market cap, exchange, and liquidity sides. We were long small caps with a beta of 1.7. We had a few asthmatic shorts with a beta of 0.8. Our only significant short position was futures with a beta of 1. Our weighted average net beta was hovering around 0.7. As one investor later pointed out, we had a "beta of 1.5 on the way up, and 3 on the way down."

Beta is a central concept of the relative long/short business. Beta is the sensitivity to the markets. If the market returns $1, a beta of 1.5 would return $1.5. In mathematical terms, this is the slope of the return of a stock versus the benchmark. Net beta is the difference between the beta-adjusted long and short exposures expressed as a decimal value. It goes past the net exposure headline to reflect a portfolio's underlying market sensitivity. When risk is on and Mr. Market is in a good mood, high beta stocks tend to do well. The market rewards risk takers who go for riskier issues. Liquidity is the primary risk of any market. As a result, small caps have higher beta than large caps. High beta stocks are also industries where there are significant failure risks such as tech or biotech, or balance sheet risks such as financials. Critics of the hedge fund industry often describe hedge funds as beta merchants, or beta disguised as alpha. There is some truth to it. A relative long/short portfolio is by definition an arbitrage on beta. Similarly, a painter is someone who throws pigments at a blank canvas.

For reference, let's have a look at the high/low beta custom index we built in the previous chapter, .

Chapter 7, High/Low Beta Sectors' Relative Returns">

Figure 8.7: , High/Low Beta Sectors' Relative Returns

The high beta sectors clearly outperformed the low beta one. This is a clear example that in a bull market, a long book would have a beta higher than 1. The short book would be constituted of stocks that return less than the market, or beta below 1. This can result in net exposure hovering around +20%, but net beta firmly at +0.5. Headline net exposure may appear low, but residual sensitivity to the markets can be elevated. In a bear market, the long book would be constituted of defensive issues such as foods and utilities with a beta traditionally below 1. The short book would list all the leaders of the previous bull phase, disgorging their euphoria. This would result in a positive net exposure of somewhere between +5 to +20% and a negative net beta of -0.1 to -0.5. Bearish portfolios have defensive stocks on the long side and market-sensitive stocks on the short side.

Despite the apparent bullishness of positive net exposure, negative net beta demonstrates a bearish stance. Conflicting net exposures is the sign of a balanced portfolio. Negative net exposures in bear markets will generate alpha, but they will have violent volatility swings. Net beta is an essential component of the toolbox. Market participants intuitively understand how beta works. Adding a bit of discipline goes a long way to deliver smooth performance.

Beta is not this monolithic measure: high = high returns, low = boring. In the previous chapter, we ran a small experiment with a Beta momentum strategy. Here is the chart for reference:

Chapter 7, Momentum Falling versus Rising Betas cumulative returns">

Figure 8.8: , Momentum Falling versus Rising Betas cumulative returns

One would intuitively think that rising beta translates into rising returns. Well, it turned out that falling beta delivered excess returns over the benchmark, while rising beta consistently underperformed.

This was a rudimentary exercise carried out on a small sample size of 10 stocks, which leads us to the next topic: concentration.

Concentration

"Diversification is protection against ignorance"

Warren Buffet Big bets have made a few fortunes and destroyed a lot more. Concentration refers to the number of stocks in a portfolio. Net concentration is the difference between the number of stocks on the long minus the short side, expressed either as a percentage of the total number of positions or in absolute terms.

Concentration has a direct impact on:

• Liquidity: Building and liquidating positions has a market impact. Concentrated portfolios are difficult to manoeuver.
• Volatility: Volatility is inversely correlated with the number of names.
• Performance: Big concentrated bets hit home runs. Small bets compound singles.

Concentration is often perceived as an attribute of an investment style and as such, does not receive the attention it deserves. It is a bit like garlic—a tasty seasoning not recommended on a first date, but also a cancer-fighting super-food and vampire repellent.

Some market participants believe the key to superior performance is hitting home runs with a few high-conviction big bets. This is often true, but it is beside the point. The best-performing products are not the best-selling ones. Sports cars attract a limited clientele.

The number of positions in your portfolio is an existential question. It is a trade-off between performance, volatility, and control.

The paradox of low-volatility returns

The way to sustainably attract and retain investors is to have structurally more names on the short than the long side, or net negative concentration. Portfolios often have more names on the long side than the short. When confronted with the scarcity of ideas on the short side, market participants resort to oversizing their short bets to bring net exposure down. They end up rationalizing these decisions by calling their outsized positions "high-conviction shorts," "structural shorts," or "tactical shorts."

The problem is that big concentrated bets increase volatility. Since the short side is notoriously volatile to begin with, this results in the short book driving the volatility of the overall portfolio. Investors want low volatility. The only way to lower volatility is to reduce the concentration on the short side. This means smaller bets and more diversified names. As we saw in , this is feasible in a relative long/short portfolio but much harder to accomplish in an absolute long/short. The short book needs to have structurally at least as many if not more names than the long side to bring the volatility down.

Ratio of big to small bets

An internal study by Fidelity Boston measured the ratio of biggest to smallest bets in managers' portfolios, outperformance, and ultimately, the retention of assets under management. The results were unequivocal:

• The higher the ratio of large to small bets, the higher the volatility. Big concentrated bets drive performance but the downside is elevated volatility of returns. This is a big deterrent for investors.
• Managers with ratios lower than 2.5 had substantially lower tracking errors than managers with big bet disparities. Their biggest bets were not so big as to hijack the volatility of their entire portfolios. Low tracking errors are correlated with the stability of AUM. In execution trader English, this means that you will be heralded as a "stock picker" when things work but branded as a high tracking error risk when your style falls out of favor. This is the institutional investor equivalent of "no one ever got fired for buying AAPL."

Exchange exposure

Who wants to buy S&P 500 big boring blue chips when it is "risk-on" and Nasdaq mints billionaires? Whoever came up with the concept of "rational investors" has obviously never set foot on a trading floor during a raging bull market. The "winner effect" is in full swing. Testosterone is wildly pumping through the systems of cold, calculating, rational, self-serving economic agents otherwise referred to as "stock punters." Yesterday's highly speculative issues are knighted as "new investment paradigms." The backward rationalization sings the same old routine: "This time it's different!"

Such nouveaux riches stocks are rarely found on the venerable main exchange. They list on NASDAQ in the United States, JASDAQ in Japan, KOSDAQ in Korea, and so on. Tech and AI have driven the post-Covid bull market.

During bull markets, a classic strategy is to go long on speculative stocks and short on blue chips. This is, however, a one-way street. The reverse does not work that well. Going long blue chips and short speculation is a difficult strategy to execute in practice. Past the biggest capitalization, borrow is hard to locate, liquidity evaporates, and the bid-ask spread is wide enough to dock a supertanker without scratching the hull.

Liquidity is the currency of bear markets. When turbulence hits, stick to the liquid issues on the main exchanges.

Sector exposure

In the second edition of this book, we have talked at length about sectors in Chapters 3, 6 and 7. Sectors do the heavy lifting for you. Defensive sectors lag the benchmark in bull markets. Now, you do not want your entire long/short sector exposures to look like technology long food short in 1999, and vice versa in 2000. Keeping a diversified sector exposure is good practice.

High intra-sector disparity offers more opportunities to capture alpha. These are the famous Netflix/Blockbuster, Meta/MySpace examples. Go back to the framework we used in about fundamental analysis. If one stock zigs when all its industry peers zag, then it is worth investigating with further fundamental analysis.

Unless you engage in intra-sector pairs trading, you do not need to fully hedge the sector exposure to mitigate industry risk. A case in point: In 2008, one would have been hard-pressed to find any stock in the financial sector on the long side.

Now that we have looked at exposures, let's put everything together in a document called mandate.

Design your unique mandate

"People live up to what they write down,"

– Robert Cialdini

Finance is the only segment of the fashion industry where last year's collection sells well. Money flows effortlessly to last year's best performers. Everybody does more or less the same things: fundamental analysis, company visits, earnings models, quantitative or technical analysis. It is difficult to come up with a differentiated pitch. If you want to leave a durable impression on your potential clients, show them something they rarely get to see: a documented process. They may choose not to invest today, but they will remember you. It happened to us 20 years ago. Our mandate was so transparent one potential investor decided to use it as a template in his subsequent interviews in Tokyo.

Investors love to put managers in boxes. They already have a set number of them. If you don't fit in any box, they probably will not make a new one for you, unless performance compels them to. So, make their job easy: tell them which box best suits you. Few market participants take the time to explicitly formalize their process.

If investing is a process, then automation is the logical conclusion. This has stood the test of time since the industrial revolution.

Here is a simple concise guideline:

1. Investment objectives: impeccable clarity reflects confidence, discipline, and precision.
2. Asset class and universe: what are you trading?
3. Strategies: how are you trading?
   1. Style description: it helps people put you in a box
   2. Strategies description: how many, synopsis (go beyond saying "fundamental" or "quantitative")
   3. Statistical description: think in terms of distributions of returns. Are you right skewed/negative mode (trend following), left skewed/positive mode (mean reversion), positive skew/positive mode (hybrid)? This gives investors an idea of what to expect.
4. Exposures: Gross, net, net beta, concentration, turnover (trading frequency)
5. Risk management: expected and annualized volatility, drawdown analysis (magnitude, duration, frequency).

This list is far from exhaustive. This process is iterative.

Managing a long/short portfolio feels like upgrading from a family SUV to a sports car. Without proper training, people tend to take their shiny new toy for an off-road spin. In this chapter, we looked at the major levers to engineer the returns.

Summary

This chapter showed that disciplined exposure management, rather than prediction, is the foundation of durable performance. The central idea is simple: "risk management is the business; positions are merely the instruments".

We learned that risk is multi‑dimensional and must be managed explicitly across four axes: liquidity, correlation, volatility, and performance. To manage these risks, the chapter formalized four essential levers: gross exposure, net exposure, net beta, and concentration.

We saw why gross exposure is the primary accelerator and brake of a long/short portfolio. It directly affects liquidity, volatility, and drawdowns. Left-skewed strategies are especially vulnerable to excessive leverage, while deleveraging under stress is often disorderly and destructive. Net exposure is the headline directionality, while net Beta adjusts for market sensitivity.

The chapter's centerpiece is the Dynamic Risk Appetite function. This drawdown‑aware oscillator collapses risk during stress, re-accelerates rapidly as the equity curve recovers, and allocates optimal risk at all times. Through simulations, we observed how linear, aggressive (convex), and conservative (concave) responses behave across trending, round‑trip, and sideways markets. The key insight is that path‑dependent risk control dominates static exposure rules.

Finally, we learned why low net exposure does not imply low risk, why net beta provides deeper insight than net exposure alone, and how concentration, sector, and exchange exposures shape portfolio behavior. The chapter concludes with a unifying principle: the clarity of a mandate will impose itself, even on those who do not appreciate its sophistication.

In this chapter, we briefly touched upon asset allocation. In the next chapter, we will go deeper into the topic of asset allocation.

Get this book's PDF copy, code bundle, and more

Scan the QR code (or go to ). Search for this book by name, confirm the edition, and then follow the steps on the page.

Note: Have your invoice handy. Purchases made directly from the Packt website don't require an invoice.