WHY CAGR IS NOT ENOUGH: TESTING A NEW RISK METRIC FOR GRID AND MARTINGALE STRATEGIES
What makes an automated trading strategy robust?
It is not only the entry signal or the final annual return. In leveraged trading, the result can also be strongly influenced by position sizing, margin utilisation, basket depth, the sequence of market movements and the rules used to close positions when account conditions change.
This is the subject of my current academic research and development work.
We are now testing two new MetaTrader 5 implementations of the Prisma S&P 500 strategy:
• Classic v2.30 – designed to preserve the established trading logic while transferring it to MT5.
• FullAuto v2.40 – an extended version introducing more automated management of exposure, equity and margin conditions.
WHY CAGR CAN BE MISLEADING
Trading systems are often compared primarily by Compound Annual Growth Rate (CAGR), maximum drawdown or the Calmar ratio.
These measures are useful, but they may not fully describe the risk of grid and martingale-style strategies.
A grid strategy can generate a long sequence of relatively small profits while gradually accumulating exposure during an adverse market move. Two configurations may therefore achieve a similar CAGR and maximum drawdown, even though one of them spends much more time operating with a severely depleted free-margin cushion.
For a leveraged grid strategy, this distinction can be critical. The binding constraint is often not ordinary volatility, but available margin.
A NEW METRIC: RETURN ON MARGIN STRESS
In my current research, I propose a new evaluation metric called Return on Margin Stress, or RoMS.
RoMS relates strategy growth to the time-weighted depletion of the account’s margin cushion. In simple terms, it asks:
How much return did the strategy generate relative to the intensity and duration of the margin stress it had to withstand?
I also introduce the Margin Cushion Coefficient, which expresses the remaining margin capacity as a normalised state variable.
These measures are particularly relevant to grid and martingale strategies because their risk may accumulate gradually and remain partly hidden behind a smooth balance curve.
In seven tick-level backtests covering 2012–2026, RoMS produced a substantially different ranking of strategy configurations from the traditional Calmar ratio. The Spearman rank correlation was −0.68, meaning that configurations that looked attractive under a conventional return-to-drawdown measure could appear much less attractive once the duration and severity of margin stress were considered.
This does not mean that RoMS replaces CAGR, maximum drawdown or the Calmar ratio. We use it alongside them to obtain a more complete picture of leveraged-strategy risk.
WHAT WE ARE TESTING NOW
Our current evaluation therefore does not focus only on the highest possible backtest profit. We compare the new MT5 configurations using:
• CAGR
• maximum drawdown
• Calmar ratio
• Return on Margin Stress
• margin-cushion depletion
• peak effective leverage
• recovery after adverse periods
• balance and equity behaviour
• stability outside the optimisation period
• sensitivity to execution assumptions
The underlying scientific question is whether state-dependent exit rules can improve the behaviour of a leveraged strategy during adverse market conditions without unnecessarily limiting its performance during normal periods.
The new MT5 versions are still being tested. They are not being presented as finished products, and no preliminary backtest result should be interpreted as a guarantee of future performance.
At the same time, the existing Prisma S&P 500 strategy continues to trade on a live account with publicly available results.
I describe this methodology in detail in my new academic paper prepared at the University of Finance and Administration (VŠFS) in Prague, which is scheduled for publication in January.
If you are interested in RoMS, margin-state-dependent exits or the research behind the new Prisma S&P 500 versions, feel free to send me a private message. I will be happy to discuss the methodology and its practical application.
⭐⭐⭐⭐⭐ Follow the Prisma SP500 live signal on MQL5:
https://www.mql5.com/en/signals/2398332
The goal is straightforward: combine academic discipline with practical trading experience and improve the strategy only when the evidence supports the change.
Past performance is not indicative of future results. Trading involves risk.
What makes an automated trading strategy robust?
It is not only the entry signal or the final annual return. In leveraged trading, the result can also be strongly influenced by position sizing, margin utilisation, basket depth, the sequence of market movements and the rules used to close positions when account conditions change.
This is the subject of my current academic research and development work.
We are now testing two new MetaTrader 5 implementations of the Prisma S&P 500 strategy:
• Classic v2.30 – designed to preserve the established trading logic while transferring it to MT5.
• FullAuto v2.40 – an extended version introducing more automated management of exposure, equity and margin conditions.
WHY CAGR CAN BE MISLEADING
Trading systems are often compared primarily by Compound Annual Growth Rate (CAGR), maximum drawdown or the Calmar ratio.
These measures are useful, but they may not fully describe the risk of grid and martingale-style strategies.
A grid strategy can generate a long sequence of relatively small profits while gradually accumulating exposure during an adverse market move. Two configurations may therefore achieve a similar CAGR and maximum drawdown, even though one of them spends much more time operating with a severely depleted free-margin cushion.
For a leveraged grid strategy, this distinction can be critical. The binding constraint is often not ordinary volatility, but available margin.
A NEW METRIC: RETURN ON MARGIN STRESS
In my current research, I propose a new evaluation metric called Return on Margin Stress, or RoMS.
RoMS relates strategy growth to the time-weighted depletion of the account’s margin cushion. In simple terms, it asks:
How much return did the strategy generate relative to the intensity and duration of the margin stress it had to withstand?
I also introduce the Margin Cushion Coefficient, which expresses the remaining margin capacity as a normalised state variable.
These measures are particularly relevant to grid and martingale strategies because their risk may accumulate gradually and remain partly hidden behind a smooth balance curve.
In seven tick-level backtests covering 2012–2026, RoMS produced a substantially different ranking of strategy configurations from the traditional Calmar ratio. The Spearman rank correlation was −0.68, meaning that configurations that looked attractive under a conventional return-to-drawdown measure could appear much less attractive once the duration and severity of margin stress were considered.
This does not mean that RoMS replaces CAGR, maximum drawdown or the Calmar ratio. We use it alongside them to obtain a more complete picture of leveraged-strategy risk.
WHAT WE ARE TESTING NOW
Our current evaluation therefore does not focus only on the highest possible backtest profit. We compare the new MT5 configurations using:
• CAGR
• maximum drawdown
• Calmar ratio
• Return on Margin Stress
• margin-cushion depletion
• peak effective leverage
• recovery after adverse periods
• balance and equity behaviour
• stability outside the optimisation period
• sensitivity to execution assumptions
The underlying scientific question is whether state-dependent exit rules can improve the behaviour of a leveraged strategy during adverse market conditions without unnecessarily limiting its performance during normal periods.
The new MT5 versions are still being tested. They are not being presented as finished products, and no preliminary backtest result should be interpreted as a guarantee of future performance.
At the same time, the existing Prisma S&P 500 strategy continues to trade on a live account with publicly available results.
I describe this methodology in detail in my new academic paper prepared at the University of Finance and Administration (VŠFS) in Prague, which is scheduled for publication in January.
If you are interested in RoMS, margin-state-dependent exits or the research behind the new Prisma S&P 500 versions, feel free to send me a private message. I will be happy to discuss the methodology and its practical application.
⭐⭐⭐⭐⭐ Follow the Prisma SP500 live signal on MQL5:
https://www.mql5.com/en/signals/2398332
The goal is straightforward: combine academic discipline with practical trading experience and improve the strategy only when the evidence supports the change.
Past performance is not indicative of future results. Trading involves risk.
