RISK MANAGEMENT

I consider risk management a key element of the investment process. A strategy's advantage is only meaningful when it's wrapped in coherent exposure control principles. Therefore, I've implemented risk management in my portfolio at two levels: the portfolio (control of the total exposure of all strategies in the portfolio) and the strategy (risk discipline within each system). In practice, this means that both the portfolio and the strategies have built-in rules that manage concentration, correlation, long/short tilt, and volatility risk.

Risk Management at the portfolio level

This layer ensures that even with multiple simultaneous signals from different strategies, the portfolio as a whole remains balanced, diversified, and resilient to less favorable market conditions. All trades that pass risk screening at the individual strategy level are then reviewed by a set of portfolio limits. This serves as a second, overarching layer of transaction filtering and exposure control, mitigating risks arising from concentration, correlation, and cumulative exposure across the portfolio.

Risk adaptation after capital drawdown

When drawdown exceeds 10%, the portfolio automatically reduces the risk level of new positions by 20%. I use this to limit the rate of deepening losses and stabilize the equity curve during periods of weaker market conditions or underperforming strategies.

Volatility limits on a single instrument (ATR)

The maximum concentration on a single instrument is limited by risk calculated based on the daily Average True Range (ATR). For each position, I ensure that its "ATR exposure" (relative to equity) does not exceed 1.5% of equity. Even a good strategy can "miss" in a particular market – the limit protects against a situation in which a single error or exceptional movement in a single instrument causes a disproportionate impact on the portfolio.

Correlated Volatility Limits (ATR)

Diversification "on paper" can fail when multiple positions, under market stress, start to behave like a single aggregate position. Therefore, I also control correlation through volatility limits (ATR).

I group markets according to similarity of behavior (correlation) and limit the total ATR exposure in the following baskets:

closely correlated basket: max. 2.0% of capital,
loosely correlated basket: max. 2.5% of capital.

This limits the risk of building a portfolio that appears diversified on paper, but in market stress behaves like a single, cumulative bet on a common factor. Correlation limits reduce the risk of "everything moving at once" in an unfavorable direction in market stress.

Global Volatility Limits (ATR)

I also control risk through volatility exposure limits calculated for the entire portfolio. The goal is to maintain the portfolio's total volatility within pre-defined limits. If any of the limits are already met, or if adding a new unit would exceed any of them, the system automatically blocks the transaction (similarly to other portfolio limits).

I use three independent limits on global volatility exposure:

net volatility (long-short difference): 4% of capital,
directional volatility (on one side of the market: long or short): 5% of capital,
gross volatility (sum of long + short): 6% of capital.

These limits are primarily intended to prevent further risk accumulation once the portfolio has reached its upper acceptable volatility threshold. In my personal opinion, these limits have the greatest added value at the portfolio level and, in many cases, can significantly improve the portfolio's risk-reward.

Risk Management at the level of a single strategy

This layer ensures that each strategy operates within a structured framework and within an acceptable risk level. This is an important element of risk management, because even with robust portfolio limits, a single strategy can overextend its exposure if left unconstrained. This ensures risk is controlled at the source - at the signal generation and position building levels.

Volatility limits on a single instrument (ATR)

In individual strategies, I limit the volatility of a single position based on the daily ATR. This allows the strategy to automatically reduce exposure to instruments whose current volatility is too high, so that a single market doesn't contribute a disproportionately large risk to performance due to a volatility spike.

I use:

single instrument volatility limit (ATR): 0.5% of capital,
permissible deviation (tolerance) from the limit: 0.5% of capital.

Applying a tolerance stabilizes the filter operation near the threshold and limits frequent trading at boundary values (when volatility fluctuates around the limit).

Correlated volatility and directional limits (ATR)

I group markets according to similarity of behavior (correlation) and/or direction and limit the total ATR exposure in the following baskets:

closely correlated basket: max. 2.0% of capital,
loosely correlated basket: max. 2.5% of capital,
directional volatility basket (on one side of the market: long or short): 5% of capital.

This limits the risk of building a portfolio that appears diversified on paper, but under market stress behaves like a single, cumulative bet on a common factor/direction. Correlation and direction limits reduce the risk of "everything moving at once" in an unfavorable direction under market stress. When the strategy reaches either level, new trades are not executed.

Global Volatility Limits (ATR)

A single strategy has an additional limit on total volatility (the sum of long and short exposures), calculated based on the daily ATR. This allows the strategy to automatically reduce exposure across all instruments when current volatility becomes too high. This is intended to prevent the strategy's total risk from rising above the assumed level.

I use:

gross volatility (sum of long + short): 10% of capital,
permissible deviation (tolerance): 1% of capital.

These limits are primarily intended to prevent further accumulation of risk when the portfolio has already reached the upper acceptable level of volatility.

The end result: the portfolio controls exposure on several levels simultaneously – limiting concentration and correlations, monitoring long/short tilt, filtering excessive volatility, and automatically reducing risk after a drawdown. At the same time, each strategy has its own "risk framework," allowing the entire portfolio to maintain a more predictable risk profile across various market regimes.